November 2026
From 17 November 2026 the European Commission accepts 28 listed goods descriptions in the ICS2 Common Repository, among them Battery case, Jewellery box, Model toy and Pharmaceutical waste (DG TAXUD, "Whitelisting of Goods Descriptions in the Common Repository", read again on 20 November 2026). Twelve of them contain a word on the stop-words list, and the check flagged them. It now answers those whole descriptions as accepted and names the whitelist entry, the stop words it contains, the document and its date, in a new whitelist_match field; singular and plural count as the same, as the Commission's document says. A longer description that contains one of them, such as "new battery case for retail sale", is checked as before, because the document lists word combinations and does not say a longer line is accepted. Every other answer is unchanged, apart from one added sentence in the disclaimer saying the whitelist is applied first. The check takes no date, so this applies from now.
The tool list the hosted MCP server sends an agent is lighter again. Measured with the cl100k tokenizer, the whole list went from 48,828 to 36,062 tokens, and the output schemas in it from 28,216 to 15,450; what an agent reads to choose a tool and build its call is unchanged at 20,342. Each tool's output schema is the response envelope, and it now names only the fields an agent acts on: ok, status, result, confidence, warnings, blocking_errors, missing_inputs, _source and citation, with envelope_version, edition and as_of. It is open, so the fields it no longer names (normalized_input, validity and _meta, which holds the identifier scheme, the upstream state and a check's receipt) still validate in a client that checks answers against the schema. The schema for one tool is 567 tokens where it was 1,058. The scoped servers follow: /api/mcp/dg went from 15,484 to 12,538 tokens, /api/mcp/ops from 30,569 to 22,222 and /api/mcp/customs from 7,376 to 4,921. The booking check keeps its own output schema. No answer changed: every field is still served in every answer that carries it, and the tools take the same inputs.
The UK duty calculator's methodology (GET /api/duty?methodology=true) cited five HMRC guidance pages at addresses GOV.UK does not hold: it has no page at any of them and no redirect. Four now link the GOV.UK pages that state the rule each was cited for, all read on 19 November 2026: Tariffs on goods imported into the UK (the UK Global Tariff applies unless an agreement or an exception does), Valuing imported goods using Method 1 (the costs of transport and insurance to the UK border are part of the customs value), Working out the VAT value using the customs value of the imported goods (customs duty is added to the customs value before VAT), and Check when you need to pay anti-dumping, countervailing and safeguard duties. The fifth, on Binding Tariff Information, stays as it was for now: GOV.UK's page for it covers Northern Ireland and the EU, and goods into Great Britain use an Advance Tariff Ruling instead. No rate, formula or answer changed.
The /containers page said the UK requires containers over 3.5 tonnes gross to be weighed before they are loaded on to a ship. SOLAS regulation VI/2, as amended by IMO resolution MSC.380(94) and read on 17 November 2026, sets no weight threshold: the shipper must verify the gross mass of every packed container, either by weighing it packed or by weighing its contents and adding the container's tare, state it in a signed shipping document, and a packed container without one is not loaded. The page now says that, with the regulation named. The page also said the EU road limit is typically 44 tonnes; it now gives the limits with the instruments that set them: 40 tonnes for a five- or six-axle articulated vehicle in international EU traffic, 44 tonnes there only for containers or swap bodies in intermodal transport (Directive 96/53/EC, Annex I, point 2.2.2), and 44,000 kg for a six-axle articulated vehicle in Great Britain (Road Vehicles (Authorised Weight) Regulations 1998). Figures with no source were removed: the tractor and chassis weights, a practical road payload of 25,000 to 28,000 kg and a reefer temperature range. The /vehicles page gives the same EU and Great Britain limits. Separately, every stored value of the 17 vehicle reference records was cross-checked against the 32 pages they cite, all read again on 17 November, and against the legal texts and three further manufacturer pages: of 178 values, 23 legal limits match their provisions, 40 agree with two or more publishers, 9 lie inside a published range, 20 rest on one source, 72 have no source, 2 are derived and 12, on four records, lie outside every source. No value was changed; the 12 are queued for correction from a source. A vehicle answer's source now names the instruments behind its limits (the Authorised Weight Regulations 1998, 23 CFR 658 and 23 U.S.C. 127) in place of "UK DVSA, 49 CFR", and six articulated records quote the Directive's 44-tonne intermodal provision in its own words. The dataset stays pending verification.
The road-weights paragraph on /containers gave the weight of an empty container as 2,300 to 5,000 kg, a range typed into the page that no longer matched the records after the two container corrections of 15 November. It now gives the range the ten container records span, 2,200 to 6,200 kg, read from the records themselves, so it follows any later correction. The other figures on the page that restate a record (the maximum gross weight of the standard containers, and the 40ft standard against high cube comparison) are read from the records too; their visible values are unchanged. The page's structured data for search engines now gives the 40ft standard's internal height as 2.38 m, as its record does (it said 2.39 m).
The values of the container specifications that lay outside every publisher read this morning were corrected the same afternoon, from the pages of Ocean Network Express, Evergreen Line, Seaco Global and Textainer, read again before each change, and by the rule the first three corrections used: the lowest rating two publishers state and, within it, the row with the lowest published payload; an internal measurement takes the smallest published figure, rounded down; a capacity takes the figure two or more publishers state. The 20ft and 40ft open tops now give an internal height of 234 cm with the roof fitted (was 238, about 3 cm above every published unit) and a width of 235 cm (was 234); the 20ft open top also gives a capacity of 32.5 m³ (was 32.9), external length 606 cm (was 610) and 30,480 / 2,350 / 28,130 kg (was 30,480 / 2,360 / 28,120). The 20ft standard gives external length 606 cm and tare 2,200 kg with payload 28,280 kg (was 610 cm and 2,300 / 28,180). The 20ft reefer gives internal 545 × 228 × 226 cm (was 544 × 228 × 218), capacity 28.2 m³ (was 27.0), external length 606 cm and a door of 229 × 222 cm (was 228 × 218). The 45ft high cube gives an internal width of 235 cm and capacity 86.0 m³ (was 234 and 85.9). No publisher states a pallet count, so each count was checked against its own record's floor in straight rows: the 20ft flat rack now gives 10 Euro and 8 GMA pallets (was 11 and 10), and the 20ft standard and 20ft open top 9 GMA (was 10). One value is still outside every publisher, the 40ft open top's capacity of 67.3 m³: the four publishers state four different figures, from 65.9 to 67.0 m³, and the rule does not pick one. A load-fit answer for these containers changes with them; the dataset stays pending verification. Separately, the booking check page's title is now short enough to show whole in search results.
Every value of the ten container specifications that /api/containers and container_lookup serve was compared with what four publishers state on their own pages, all read on 15 November 2026: Ocean Network Express, Evergreen Line, Seaco Global and Textainer. Of 180 values, 69 agree with at least two publishers, 23 lie inside the range the publishers span, 2 are stated by one publisher only, 66 are stated by none (pallet counts, deck areas and most ISO size-type codes) and 20 lie outside every publisher. Three records were corrected. The 40ft reefer now gives a maximum gross of 34,000 kg, tare 4,760 kg and payload 29,240 kg (was 30,480 / 4,800 / 25,680), internal height 254 cm (was 242), capacity 67.5 m³ (was 63.9) and door width 229 cm (was 228). The 20ft flat rack now gives a deck of 555 × 219 cm (was 583 × 234), external length 606 cm (was 610), deck area 12.15 m² (was 13.64) and 34,000 / 2,800 / 31,200 kg (was 30,480 / 2,740 / 27,740). The 40ft flat rack now gives a deck length of 1,165 cm (was 1,196), deck area 27.26 m² (was 27.99) and 45,000 / 6,200 / 38,800 kg (was 30,480 / 5,000 / 25,480). A load-fit answer for these three containers changes with them. The 20 values still outside every publisher, on six other records, are queued for correction; the dataset stays pending verification.
The tool list the hosted MCP server sends an agent is lighter. Measured with the cl100k tokenizer, the whole list went from 63,265 to 48,828 tokens, and what an agent reads to choose a tool and build its call (names, descriptions and input schemas) from 34,779 to 20,342. The descriptions are now the text of the freightutils-mcp 2.34.4 package, and each input field's text is the package's one-line wording. The rules every tool shares (the response envelope, the rate limit, unknown fields, ADR editions, identifiers, lists, batch arguments and the consignment object) are sent once, as the server's instructions, instead of in each description. Nothing in any answer changed: the same tools take the same inputs and return the same results. There are still 26 tools, and 27 on the hosted server.
The HS codes file no longer carries the aggregate row of the source file it was built from: the row TOTAL, the sum line of the UN Comtrade classification file, which is not a code of the HS 2022 nomenclature. /api/hs?code=TOTAL now answers not found, like any code that does not exist, and no HS search lists it. The coverage matrix (/coverage and /api/coverage) counts 6,936 records and 6,936 identifiers for the HS codes file: one per code. The row's identifier, fu:hs:TOTAL, is retired and will never be issued again: GET /api/resolve and resolve_reference return no candidate for it and say it was issued and retired on 14 November 2026, and why. It is the first retired identifier. A chapter's answer changed in one field: parent is now null, since the nomenclature places a chapter in a Section (still given in section) and gives it no parent code; it read TOTAL, the source file's marker. The change feed carries the removal as release 7. The /hs page and /api-docs now say 96 chapters.
A new tool for the agents and systems that take bookings in: POST /api/booking-check, and booking_check on the hosted MCP server. Send a road booking's fields — origin and destination as codes, the pieces with their dimensions and weights in stated units, the booked vehicle, any dangerous-goods lines — and the answer says, in three lists, what the booking lacks (each field with the calculation that needs it), what the job needs as booked (the places resolved, volume, gross and chargeable weight, loading metres, whether the load fits the vehicle, and the ADR consignment check line by line) and what stops it (a code that does not resolve, a load the vehicle cannot take, a dangerous-goods line that is refused or fails a check), with a verdict: no_blocker_found, missing_information or blocker_found. It adds no rule of its own: every line is the answer of a calculator or lookup already on this site, with that engine's source, and what no engine answers yet is listed as not checked. The answer is short by default; detail: true adds each engine's full answer. Nothing is stored, for any caller: the answer carries its receipt hash only. A party name, address or free text sent with a booking is ignored and never repeated. Road bookings only for now. It is the 26th tool and the 28th REST endpoint, and the 27th tool on the hosted MCP server; the freightutils-mcp package gains it at its next release. It is not a booking system and gives no rate, route, drivers' hours or legal clearance: a clean answer means no engine found a blocker in the fields sent.
The HS lookup's answers now count codes, not the rows of the file. total_codes (GET /api/hs with no parameters) and meta.codes on every HS code lookup, search and section answer now say 6,936: 96 chapters, 1,228 headings and 5,612 subheadings. They said 6,937, which also counted the dataset's aggregate row TOTAL, and that row is not a code. The /hs page now says 6,936 throughout: in the FAQ, in the search box and in its heading. So does the HS operation in the OpenAPI spec. No code answers differently, and /api/hs?code=TOTAL answers as before.
The /eval page now opens with the latest result of each condition: on the 30 questions about what ADR 2027 changes, the models with the FreightUtils tools against web search and against the models alone, and the whole current question set with the tools; the earlier runs follow, newest first, and the first runs now say in the past tense that the tools did not yet serve ADR 2027 then. The /mcp page shows the answer envelope as the hosted server and the npm package return it (version 1.2, with its _source fields and a citation object), says plainly where the two differ (result field names on nine tools, and how far the output schemas type the result), and lists the ADR consignment check with the dangerous-goods tools. The site counts 6,936 HS codes, the chapters, headings and subheadings of HS 2022; the dataset's one aggregate row is not a code. freightutils-mcp 2.34.2 is now the version the hosted server runs, with the same HS count in its tool description.
A new page, /changes, gathers what changed for a list you hold: paste your UN numbers to see what ADR 2027 changes in Table A for each, follow every data change on this site by JSON or RSS, and find HMRC's own watch lists for UK commodity codes. The ADR 2027 list check is now also an API, GET /api/adr/changes-2027, the 27th REST endpoint: the same answer the check page shows, per UN number and cited to the UNECE document and page, for up to 200 UN numbers a call. A check result now has a Copy link button; the list travels in the link and is never stored.
The Verified, Pending, Computed and Live labels on the homepage, the coverage page, the methodology page and the tool pages are now one label style, whatever group the tool sits in. Before, a label took its group's colour, so Verified showed in red on the dangerous goods tools, where it could read as a warning. Now Verified, Computed and Live share one neutral style, each with its own small icon beside the word, and Pending has a dashed outline. The group colour stays on the tool card and the group heading. No dataset's status changed. On the methodology page, Live and Computed are now shown as the two states the data carries, instead of one "Computed live" label.
GET /api/changes is new, the 26th REST endpoint. It gives one release per data change that merged here, naming for each dataset the records added, amended or deleted by fu_id, the served fields that changed and the version ids before and after, so an integration that caches our answers knows which records to re-fetch and when an edition moved; join an event to its record through GET /api/resolve?q=<fu_id>. Ask with ?since=2026-11-05 (a UTC date, or a timestamp ending in Z) and continue with the next_cursor each answer gives; ?format=rss gives RSS 2.0, one item per dataset per release. The history starts on 6 November 2026 at 01:30 UTC, when identifiers were first served, and its first release is that morning's airport refresh: 33 added, 38 amended, none deleted. Five more releases follow, each for a whole dataset (no record-by-record diff is declared for them yet): the airline corrections of 7, 8 and 10 November and the UN/LOCODE status meanings, now UNECE's own names, of 12 November. Before and after values are carried only for datasets marked clean or attribution required on Data Licensing, with the attribution line, and never for a deleted record — today, airports. A dataset whose change has no record-by-record diff yet appears as one entry for the whole dataset, and a dataset whose served file the feed cannot account for is listed as incomplete, never as unchanged. A release is dated by the commit that recorded the data change, not by a deploy or by the publisher. Not in the feed: changes a publisher has made that have not merged here (those are _meta.upstream and /data-status), and anything before 6 November. Each release is generated from the committed data files and checked against them before any change merges. The feed is not an MCP tool yet.
Every one of the 116,086 UN/LOCODE codes FreightUtils serves was compared, field by field, with UNECE's own 2025-1 release file, downloaded again from UNECE's publication repository (the same file the data was built from). Country, code, name, name without diacritics, subdivision and status match on every code, and every coordinate served equals UNECE's own decimal reading of it. What differs is by design and labelled on the record: codes UNECE lists on more than one row are one record with the other names; the 19 entries UNECE marks for removal are served as deleted; coordinates whose published string is impossible or a placeholder are withheld with the reason; IATA codes UNECE leaves blank are filled only where OurAirports confirms them, and each one says so. No served value was wrong. UN/LOCODE now reads verified on the coverage page: an exact-code answer carries high confidence without the pending warning (a location whose status is not approved still carries the provisional warning), and resolve_reference reports every code verified. The status meanings in those warnings now use UNECE's own words; one had named an organisation UNECE does not.
Where ADR 2025 Table A has more than one row for a UN number, its page opened on one generated sentence per row. On UN 1202, diesel, the three sentences read the same except the special provisions, while the tank code, the vehicle for tank carriage and the carriage operation provision also differ between the rows and were shown only in the tables further down. These pages now open on one sentence that states what every row shares, then one line per row with only the columns whose values differ, each value as Table A prints it after its column number, and the name and description where it differs. Each row's full sentence now sits above that row's Table A table, and a link under the answer goes to the rows in full. Pages for a UN number with one row are unchanged, and so is the API answer.
Since 10 November a free API key brings one email, the key itself. The privacy policy now says the same in each place that named the two follow-up emails: what an email address is used for, what Resend sends for us, and the lawful basis. What is left from before is deleted on its original terms, and the policy gives the dates: the sending schedules by 20 November 2026 and the record of each follow-up email sent by 10 December 2026.
IATA's own list of member airlines gives each airline's three-digit code beside its designator: 416 is National Airlines (N8), 828 is Hong Kong Air Cargo (RH) and 851 is Hong Kong Airlines (HX). Prefix 416 had answered both National Airlines and Fika Salaama Airlines of Uganda, and 851 had answered National Airlines, Hong Kong Air Cargo and Fika Salaama. National Airlines now serves 416, Hong Kong Air Cargo 828, and Fika Salaama Airlines no prefix. /airlines/awb/416 goes to National Airlines' page, /airlines/iata/N8. No record serves 851 for now, so /airlines/awb/851 answers not found: Hong Kong Airlines' record has no prefix yet. IATA lists DK for Sunclass Airlines ApS of Denmark. The reference set held that carrier under its former name, MyTravel Airways (ICAO VKG, callsign VIKING), and the carrier's own history page names MyTravel Airways, Thomas Cook Airlines and Premiair as its earlier names. That record is now Sunclass Airlines with IATA DK, and /airlines/iata/DK shows it. The carrier says it carries air freight on most of its routes and names no AWB prefix, so the record has none. Airport answers still name OurAirports, but its page ourairports.com/data/ has served an expired certificate since 6 November. Each airport answer's source link is now the Internet Archive copy of that page from 14 January 2026. No airport value changed.
Between 04:27 UTC on 8 November and about 14:30 UTC on 9 November, every API call made with a key returned an error, because the database that holds API keys and usage counts was suspended by its provider. Calls without a key, the website and the hosted MCP server kept answering. A call with a key is now answered when that database cannot be reached: a key the server checked in the last minute keeps its own limits, and any other call is answered as an anonymous one, not counted, with the header X-FreightUtils-Degraded: store_unreachable (and _meta.degraded on a ?envelope=1.2 answer). Errors from that database are now logged as a short code, without the request that failed. A free API key now brings one email, the key itself: the two follow-up emails, tips after two days and a note after five, are no longer sent.
On every HS code page, such as /hs/code/847130, the heading is now the code, and directly under it is the cited answer: the code, its description and the edition (WCO HS 2022), with its source. The same card links "UK duty for" the code and the codes before and after it under the same heading. The agent and API hand-off follows, then the classification: section, chapter, heading and level, each with its description. /duty now answers a code given in its address. /duty?code=847130 lists the declarable UK 10-digit codes under that subheading, from the UK Trade Tariff, each a link. /duty?code=8471300000 opens on HMRC's description of the code and its third-country duty expression, with the code already in the form, ready for the origin and the customs value. If the UK Trade Tariff cannot be reached, the page says so and does not show an empty list.
Fika Salaama Airlines (Uganda) answered IATA code N8 and Eastland Air (Australia) answered DK. IATA's own code search lists N8 for National Air Cargo Group Inc dba National Airlines and DK for Sunclass Airlines ApS of Denmark, finds no code for either carrier, and lists no airline under EL, the code a public database gives Eastland Air. No other first-class source states a code for either, so both records now serve no IATA code, in the same form as Magma Aviation and Niger Air Cargo. /airlines/iata/N8 now lists National Airlines alone, and /airlines/iata/DK answers not found. Eastland Air's page is its AWB prefix page, /airlines/awb/630, and Fika Salaama's is its prefix page /airlines/awb/851. ICAO codes, callsigns and AWB prefixes are unchanged, and each record's rationale lists every page read. The dead citations on the Fika Salaama record (the carrier's former website and an old Uganda CAA operator list, neither ever archived) are removed.
On /adr/consignment-check, an answer now opens above the form. It starts with the load as you entered it (each line's UN number, packing group or Table A row, quantity and packages; the journey, the vehicle and the driver), then the load status and the 1.1.3.6 points, then every check that did not pass, each with its state, its ADR section and the input that would settle it, exactly as the check returns them. The checks that pass follow, one line each, in a list that is open when the page loads; a check that fails, is conditional, is not assessed or does not apply is never inside it. The form follows under "Change the load" and keeps every value. It now takes the ADR 8.1.5 equipment on board, the specialisations on the driver's certificate, photo identification, and the vehicle type and axles, and a changed load no longer drops an input that came in the page's address. A fourth sample load, 2,000 L of petrol with the vehicle and driver declared, shows a load the check passes. The "Which ADR tool answers which question" box on the UN number lookup, the 1.1.3.6 calculator and the LQ / EQ checker now names the consignment check.
Cargo Plus Aviation (8L) and Eastland Air (DK) were shown as having ceased operating because their ICAO designator is no longer listed for them by the FAA and "their former website" no longer answered. Neither website was ever theirs: one was a Dubai web-design firm's, the other belonged to a Curaçao air-taxi firm trading as "ELA Airlines". With those set aside, no first-class source states whether either carrier operates today, so both record pages and /api/airlines now say exactly that (operational status unknown), as they do for every other record in the same position. The record's rationale names the old basis and why it fell. Codes, names and AWB prefixes are unchanged. A re-run of the airline data build now keeps the archived copies that replaced dead carrier websites on 7 November, and a check fails the build if a dead or never-the-carrier website is cited again.
Opening /api-docs#adr-editions in a new tab, or any long page at one of its sections, now lands on that section. The browser fixed where to scroll before the page's fonts had loaded; as they arrived the text above re-wrapped, about 350 px shorter on the API reference, and the section ended up above the screen. Once the fonts have settled, the page now puts the section back where the link asked, unless you have already scrolled, tapped or pressed a key. Links followed within the site already landed, and the carrier links in the /chargeable-weight divisor table land on the preset line and the calculator at both phone and desktop widths.
For 22 airlines, a source cited on the record page no longer answered: a carrier's site gone or moved, a press article taken down, a liquidator's page removed. Each now links the same page where it moved, another primary page that states the same fact (for Fiji Airways' code, IATA's own code search), or an archived copy of the cited page from the Internet Archive, dated, with the live result kept in the citation's words — for CanJet, its own 2008 homepage beside "the live domain no longer resolves". 4 citations are removed because the page never supported the fact: 3 "former carrier domains" belonged to other businesses, and one was an encyclopedia article deleted as unsourced. 8 sources with no replacement stay as they were and are listed for a later pass. No airline code, name, prefix or status changed.
A walk of the whole site, every page in the sitemap and every link on it, found two links that did not answer. On /changelog, the 13 May entry on ADR provenance linked to /api/adr, which answers an error when it is opened without a UN number; it now opens the ADR section of the API reference. On /eval, two links to the run audits pointed to a repository readers cannot open; the page now says those audits are not yet published. The same walk now runs every day over every hand-written page and a sample of the HS, ADR and airline pages, and checks each internal link and #anchor once: a dead link, a redirect that does not end on a page, or an anchor that is not on its page fails the daily check. Once a week it also reads the links to other sites and lists the ones that no longer answer.
/adr/consignment-check checks a road load of dangerous goods in one go, the way POST /api/adr/consignment-check does: enter each line (UN number, quantity, packages and, if you want, the inner packaging and the packaging mark), the date of carriage and the tunnel categories on the route, and, if you know them, the vehicle and the driver. The answer gives the load status and whether the load is carried under 1.1.3.6 or full ADR, the 1.1.3.6 points, and every check with its own state and ADR section: limited and excepted quantities, the whole-load tunnel code, the transport-document lines, mixed loading, placards and plates, the equipment on board and the driver training, then what would settle the checks that stayed open. A UN number with several Table A rows, such as diesel (UN 1202), asks you to choose the row; none is chosen for you. Three sample loads open on their answers. The load is carried only in the page address, so a link re-opens the same answer; nothing you enter is stored. Under the answer, the same check for an agent (the MCP call) and through the API (the curl line with the body). The ADR Consignment Check card on the homepage and in the Tools menu now opens this page.
On the homepage, the Shipment Summary, ADR Consignment Check, Nearest Airport and Resolve Reference cards opened the top of /api-docs, a long page that documents every endpoint, and so did their entries in the Tools menu. Each now opens its own section of the reference (for example /api-docs#consignment-check), and Nearest Airport opens /airports with the "Nearest to me" control in view and focused; your location is still asked for only when you press it. The four cards now say API · MCP, because these tools have no page of their own and answer through the API and the MCP server. A section of the reference opened from a link now lands below the header on a wide screen; three of them were half hidden under it.
Under the answer and the agent / API hand-off on each ADR UN page, a row now links the 1.1.3.6 points calculator and the limited / excepted quantity checker for that UN number. Both now read the UN number from the address: /adr-calculator?un=1203 opens with petrol selected (transport category 2, ×3) and asks for the quantity, and with &qty=200 shows the 600 points; /adr/lq-eq-checker?un=1203 opens with UN 1203 filled in and its Table A limits (limited quantity 1 L, excepted quantity E2). For a UN number with several Table A rows, such as diesel (UN 1202), both list the rows and you choose; nothing is chosen for you, and variant_index picks one as in the API. The calculator link further down each UN page used to open an empty calculator; it now opens on the UN number. Entries not subject to ADR, or prohibited, get no row. On the calculator, the "1.1.3.6.3 wording, verbatim" toggle under the quantity is now in the page's text colour (it was orange at 2.1:1 contrast).
On every ADR UN page whose Table A entry ADR 2027 changes, deletes or adds to, a line directly under the answer now names the columns that change (for diesel, UN 1202: (2) Name and description and (6) Special provisions), gives the dates from the acceptance notice it cites, and links the ADR 2027 check for that UN number and the full change table further down the page. Pages whose entry does not change show nothing new near the answer; the ADR 2025 answer itself is unchanged.
The monospace font used for numbers, codes and inputs is now cut to Latin-script characters, punctuation, arrows and maths signs: 78 KB for its two weights, down from 186 KB, on every page. While it loads, the stand-in font now has the same width and line height, so fields and figures keep their size when it arrives (on /ldm, a link to a result moved the form down 4 px). The links in the header, on the homepage and on each ADR UN page no longer fetch their pages in the background while the first page is still drawing; each is still an ordinary link.
Under the result on /cbm, /ldm and /chargeable-weight, a row now opens the same pieces in the other two calculators. From CBM: the chargeable weight by air and by sea, with the dimensions and pieces filled in, and the loading metres for the same footprint and count. From loading metres: CBM and the chargeable weight for the same footprint and count, with a weight per pallet carried over as the total gross weight. From the chargeable weight: the CBM result, and the loading metres with the total gross weight carried over as a weight per pallet. Where the other calculator needs something this one does not have (the gross weight, the height or the vehicle), the link says so and opens on that field, ready to type; nothing is guessed. A loading-metre link now opens on its own result in the page itself, rather than on the one-pallet default. Nothing you enter is stored: the pieces are only in the links.
On /ldm the "Pallets can be stacked" switch is now announced by a screen reader as "Stackable? Pallets can be stacked", with its hint; before, it had no name at all. The switch looks the same. On a phone, this changelog no longer scrolls sideways: a long path or word in an entry now wraps. No entry's wording changed.
Under the result on /cbm, /ldm and /chargeable-weight there is now a link that reproduces the calculation, and the same answer as a hosted MCP tool call and as a curl line for the REST API, built from the numbers you entered, each with a copy button. They carry the citation the API returns and the status it gives the answer: Computed. The link's query string is the API's own (for example /cbm?l=120&w=80&h=100&pcs=5 and /api/cbm?l=120&w=80&h=100&pcs=5), so a link pasted into an email or a booking opens on the same result. A sea (W/M) chargeable weight shows the API line alone, because the MCP tool computes the air answer only. Loading-metre links now keep a custom pallet size and a custom vehicle length, and chargeable-weight links keep the sea mode; before, both were lost. A link to a CBM or chargeable-weight result is answered in the page itself, and opens at the answer (the link ends in #result). On a phone the calculators' headings take less room. On /ldm the trailer drawing now sits under the result. Links to a section of any page (such as /api-docs#cbm) now open at that section instead of jumping back to the top. Nothing you enter is stored: the numbers are only in the link.
The site header now takes one row on a phone (about 53 px, down from three rows and 139 px): the brand, the search button, "Get an API key" and a Menu that holds API docs, MCP, Pricing, sign-in, the theme switch and every tool. Every page's first answer moves up by about 85 px, and on the ADR 2027 check the paste box and its submit button now fit the first screen of a phone. On a phone the header scrolls away with the page instead of staying pinned. The search button on any page opens the homepage's paste box in place: paste a UN number, an HS code, an air waybill, an airport or port code or a list of UN numbers, and the cited answer appears under it with its verification status and the same call for an agent and the API, without leaving the page; on the homepage the button goes to the page's own box. When the text holds no code, the box lists the tool pages whose name matches it. At 200 % zoom on a phone, pages now reflow to one column instead of scrolling sideways, and the header no longer shifts while it loads. Nothing typed into the box is stored.
The record pages search lands on — a UN number (/adr/un/1203), an HS subheading, heading or chapter, an airline by IATA code or AWB prefix — now carry, directly under the cited answer, the same record as a hosted MCP tool call and as a curl line for the REST API, each with a copy button, the MCP server address, and the verification status the call returns, as the dataset states it (Verified or Pending). The call names the record's own identifier, never the text that was searched, and an ADR page's call pins ADR 2025, the edition the page shows, so a copied call keeps answering what the page said after ADR 2027 applies. The curl line carries the page's citation as a comment, so it can be forwarded with its source. On a phone the answer sits higher in the first screen. On the UN number pages this replaces the "Use this data programmatically" box at the foot of the page.
The homepage now opens on one box. Paste a UN number, an HS code, an air waybill, an airport or port code, a list of UN numbers or a line of text, and the answer appears under the box with its citation and its verification status, as the full answer pages give it: a code is resolved to its typed, ranked readings (an ambiguous one such as FOB shows every reading in order, never one picked), a list of UN numbers is checked against the ADR 2027 Table A changes with the amendment document and page, and an air waybill in a line of text is checked for its check digit and its airline prefix. Under each answer is the same call for an agent on the hosted MCP server and as a curl line for the API, ready to copy. A row of examples answers on a tap. Nothing a visitor pastes is stored. The site-wide MCP announcement bar is gone, and on a phone the ADR 2027 check now shows its paste box first.
In free text (GET /api/validate?text= and the validate MCP tool), an air waybill written "125 1234567 0", "176 12345675", "176-1234567-5" or "176-1234-5675" was dropped without a word, and a spaced one with a wrong check digit was dropped too, so "157 4321 8765, 618-55555555 and 125 1234567 0" answered only 618-55555555. Text mode now reads eight spellings. Next to an AWB, MAWB, HAWB or air waybill label every spelling is checked exactly as typed mode checks it, so "MAWB 157 4321 8765" answers INVALID_CHECK_DIGIT; a check digit written apart is read whenever the prefix is an airline's other than 020 or 077, so "125 1234567 0" now answers INVALID_CHECK_DIGIT too. Because UK telephone numbers are also 11 digits, the shapes they share are read without a label only when the check digit is right and the prefix is an airline's — not after a word such as "call" or "tel", and, written with spaces, not under 020 or 077, how London and UK mobile numbers begin. "Please call the desk on 020 7946 0006" no longer answers a valid air waybill 020-79460006, and an unspaced telephone number is no longer reported as an air waybill with a wrong check digit. What text mode no longer reports: an unlabelled spaced number under 020 or 077, and an unlabelled unspaced 11-digit run whose check digit fails, whose prefix no airline in the dataset holds, or that follows a telephone word; add the label, or check the number in typed mode (value= with type=awb), which answers exactly as before. Checked against 14,534 generated lines, including every number in the ranges Ofcom reserves for drama.
GET /api/resolve?q=<fu_id> and the resolve_reference MCP tool resolved an airport identifier only when the airport had an ICAO or IATA code, so 65,433 of the 86,281 airport identifiers the answers carry — heliports, airstrips and closed fields without a code, such as fu:airport:07TT — answered "names no record"; 24 airline identifiers and 1,325 HS chapter and heading identifiers (fu:hs:01) did the same. Every issued identifier now resolves to exactly its record, and a check before each release proves it for all of them. An airport with no code answers as airport_ident, with api_url /api/airports?ident=07TT, a new exact lookup by the OurAirports ident; an airline with no usable code answers as airline_slug, with api_url /api/airlines?slug=…, a new exact lookup by record slug. A record is never labelled with a code it does not carry, and every identifier that resolved before answers exactly as before. The /api/airports answer now says when its data was last verified: meta.last_verified gives the date of the last audited refresh, 2026-11-06, instead of "June 2026".
A free-text search on GET /api/hs?q= and the hs_code_lookup MCP tool matched any description that contained the letters typed, so "charger" answered heading 8416, furnace burners, because the tariff text mentions "ash dischargers". Words now match only as whole words. When the tariff text matches a word only weakly, the UK Trade Tariff search references (HMRC's index of everyday goods names) are asked as well, and their answer is served: "charger" now answers heading 8504 (static converters, from "wireless charger"), with 8544 for a charger cable, and "trainers" answers chapter 64, footwear, instead of the aircraft "ground flying trainers" of 8805, which the message names as a different reading of the word. When neither source has a confident match, the answer is "no confident match", with the weak candidates named in the message as candidates, not as the result. A code search ("8471.30") is unchanged. Confirm any code against the goods.
The upstream state that record lookups have carried since earlier today now rides on the calculators that read a served dataset: the loading-metre calculator (its vehicle and pallet presets), the UK duty calculator (the live UK Trade Tariff and the country table), the ADR 1.1.3.6 exemption calculator (the ADR edition it computed on), the shipment summary and the consignment calculator (what their parts read for that call), and validate when an air waybill is checked (the airline prefix list). It is under _meta.upstream on a v1.2 envelope and every hosted MCP result, and in the X-FreightUtils-Upstream header on REST; the flat body is unchanged, and ?if_stale=refuse works on these routes as on the others. A calculation that reads no dataset — CBM, chargeable weight, pallet fitting, unit conversion, a container or IMO check digit — carries none.
The airport list behind /airports, GET /api/airports, GET /api/nearest-airport and the airport_lookup and nearest_airport MCP tools is refreshed from the OurAirports files of 6 November 2026 (public domain): 33 airports added and 38 amended, none removed, 86,281 in all. Nine amendments record an ICAO code that OurAirports had held only as a GPS code — seven airfields in Uganda, Princeton in British Columbia, and Malmstrom Air Force Base (KGFA), which ?icao= now matches on its ICAO code; one gives CFB Gagetown Heliport the IATA code YCX. Luanda Quatro de Fevereiro (LAD) and Doha International (DIA) no longer have scheduled service, so the nearest_airport default ranks Luanda's new airport (NBJ) and Hamad International (DOH) instead. No airport with an IATA code moved by more than 1 km. Every airport the list already held keeps its identifier: fu_id is unchanged and fu_version_id names the new edition, for example fu:airport:EGLL@2026-11-06. The 1,175 large airports are cross-checked against OpenFlights again.
The freightutils-mcp package on npm, version 2.34.0, now declares _meta.upstream in its output schemas: the upstream state of the datasets an answer was drawn from, with each dataset's state, how it is watched, its last successful check and window, and stale, which is true only when a dataset missed its check window. The field is optional and open, so an answer without it still validates, and 2.33.0 already accepted every answer that carries it. No answer changed and the tool count is unchanged at 26.
Every API answer and MCP tool result drawn from one of the datasets FreightUtils serves now says what the last check of that dataset's upstream found. A daily check run by the site fetches each watched source — the UN Treaty Collection's ADR notifications, the UK Trade Tariff, the UN/LOCODE releases, the OurAirports files, the emissions-factor publications and the EU ICS2 list — rejects an error page or a truncated file before reading it, validates its shape and compares it with what is held. Under ?envelope=1.2 and on the hosted MCP server the answer carries _meta.upstream; every REST answer of a record lookup carries the same object in the X-FreightUtils-Upstream header, and the flat body is unchanged. Per dataset it gives the state (in step, changed upstream, schema drift, unreachable, window missed, or unknown when the state cannot be read), how the dataset is watched, the last successful check and the window. stale is true only when a dataset had no successful check inside its window; the answer is still served from the last verified edition held. One unreachable check, a known difference upstream or a state that cannot be read never makes an answer stale. A caller who would rather be refused adds ?if_stale=refuse to a REST request and gets HTTP 424 with the reason while a dataset of that answer is in a missed window. A new page, /data-status, shows every dataset's state, the last successful check, the edition held and the window, and /coverage and /api/coverage show the same state. It says whether the upstream was checked inside its window, not that an answer is right. Every existing field keeps its name, type and value; the tool count is unchanged.
The freightutils-mcp package on npm, version 2.33.0, now declares fu_id and fu_version_id in the output schemas of the fifteen tools that return records, and _meta.ids on the envelope, so a client that reads the tool list can see that each record carries a stable identifier before it calls a tool. Both fields are optional and every record stays open, so an answer without them still validates. The same fifteen tools, here and on the hosted MCP server, now end their descriptions with one sentence saying that each record carries fu_id, and fu_version_id where the dataset has editions, and that resolve_reference resolves one to exactly its record. resolve_reference in the package now takes an fu: identifier of up to 200 characters, as the hosted tool already did; every other identifier keeps the 32-character limit. No answer changed and the tool count is unchanged at 26; earlier package versions still accept every answer.
Every record an API answer or MCP tool result returns now carries fu_id, a stable identifier that stays the same across editions, such as fu:adr:un:1203:II:F1, and, where the dataset has editions, fu_version_id, which names that edition's record, such as fu:adr:un:1203:II:F1@2025. Store fu_id to join on a record across editions; store fu_version_id to pin the edition an answer came from. Records inside other answers carry them too: the dangerous goods of a consignment line, the row an LQ/EQ check used, the candidates of /api/resolve, the emission factor a calculation used and the ICS2 stop-words a description matched. Airlines, ULDs, containers, vehicles and pallets have no editions, so they carry fu_id only. Every identifier issued is kept in an append-only ledger, and a check before each release proves none is reused for another record; where two ADR entries share a UN number, packing group and classification code, a short alias such as viscous-vp-le110kpa tells them apart and stays fixed even if the name is later corrected. /api/resolve and the resolve_reference tool now also take an fu identifier and return exactly its record, or none. Under ?envelope=1.2 and on the hosted MCP server, _meta.ids states the scheme. An identifier is not a regulatory reference, a citation of the source or a promise that a record never changes. Nothing else in any answer changed: every existing field keeps its name, type and value.
A new page, /coverage, lists every dataset an API answer or MCP tool result comes from: one row for each dataset and edition held, with the jurisdiction it is law or reference for, the edition and the dates the authority gives it, whether that edition was the authority’s current one at the last check, the verification state every answer already carries (with the count of records in each state where a dataset records one), the licensing status from /data-licensing, and when the authority was last checked and is next due. Beside them it names the neighbouring rule sets that are not served and why: RID and the EU and US tariffs are not served; the IMDG Code, the ICAO Technical Instructions, the IATA Dangerous Goods Regulations, IATA’s coding directory and the Cargo-IMP tables are never served as data, because their publishers sell them under their own terms. The same matrix is served as JSON at /api/coverage, with each row’s authority, source link, retrieval method, record count and file checksum, and the reason for anything not recorded; /data-licensing is served word for word as JSON at /api/data-licensing. Both are generated from the dataset registry, and a build check proves that the page, the JSON and the registry agree. No tool’s answer changes.
The section on /eval that judges the ADR consignment check whole left nine answers open, where the first readers had no majority — on three of the four loads concerned one reader had declined to answer. Each of those four loads went, whole and unchanged, to two further independent readers, again without a FreightUtils answer, and an open answer now settles only on a reading held by more than half of every reader who answered it. Six settled that way, three readers in four each time, and the check already gave all six outcomes. Three stay open, two readings against two, and are listed in the audit for a person to decide. The outcome table now shows the six on their own row; no other figure moved.
The packaging check that reached the ADR consignment check today is now described and typed on MCP: on the hosted MCP server and in the npm package freightutils-mcp 2.32.0, adr_consignment_check takes the optional packages.un_mark of each line — the UN packaging mark as printed, e.g. UN 4G/Y145/S/02/NL/VL823 — which the hosted tool refused until now, and its answer types the packaging_code_validity block: the mark read field by field in the order of ADR 6.1.3.1 (the symbol, the packaging code of the 6.1.2.7 table, the packing-group letter and the density or gross mass, S or the test pressure, the year, the State and the maker) and its letter held against the line's packing group, with the sub-paragraph named on a fail. The description says the other answers of today's build too: a limited or excepted quantity line whose stated regime is met is left out of the 1.1.3.6 total (1.1.3.6.5, named in exemption_1_1_3_6.engine.excluded_lines); an N.O.S. entry with special provision 274 is conditional on the transport document until its technical name is added; a CARRIAGE PROHIBITED line gets no document entry; an ID-prefixed line without the ICAO limited-quantity statement leaves the document insufficient_data. Its Limitations now say what the mark check does not do: whether the packaging is one the packing instruction allows, IBC and large-packaging marks, and a row without a packing group. The same field is accepted in the consignment argument of adr_exemption_calculator, consignment_calculator and shipment_summary, which list it as not read. Nothing removed: every 2.31.0 argument and answer field still validates.
The guide What a freight reference tool changes for an agent said scoped manifests were in the queue and that the MCP tools would follow the REST API onto envelope 1.2. Both have shipped, and the guide now says so: the hosted server serves the dangerous-goods, customs and operations scopes at /api/mcp/dg, /api/mcp/customs and /api/mcp/ops (and the npm package takes --scope), each a filter on the full tool list, and the MCP tools answer with envelope 1.2, refusals included. No figure in the guide changed; every figure is still filled from the pinned eval results and checked by lint.
The For AI agents sample on the homepage now opens on UN 1230 (methanol), a row with a row-level audit, so its answer reads "verified": true; it is generated from the live lookup at build time, as before, and the MCP, curl and GPT tabs show the same row. Each counter in the stats row and each example card now carries its verification state as a small badge — Verified with the audit date, Pending, Live or Computed — read from the same per-dataset record the API's _source carries, in place of the seven repeated "provenance pending verification" lines. One line under the counters says what Pending means and links data licensing and methodology. On a screen less than 500 px tall, such as a phone in landscape, the site header now scrolls away with the page instead of staying pinned, so the whole screen is left for reading; there is no animation and nothing else moves. On phones and touch screens the header links and buttons, the Tools menu, the copy curl and JSON chips, the code-tab buttons and the footer links are now at least 44 px tall, and no control outside running text is smaller than 24 × 24 px.
The /eval page has a new section on the ADR consignment check. We wrote a set of whole road loads the way a practitioner describes them, and two independent readers and a tie-breaker set the outcome of every one of the check's answers from the ADR text, without seeing a FreightUtils answer. Those outcomes were fixed before the check ran. The section shows who set them, what the check answered against them, and what was fixed before the table (above all, limited and excepted quantities left out of the small-load total). It lists the two false clears: both are loads whose description contradicts the booking on purpose. Every difference is explained in the audit, and the section says plainly what the readers are not.
The ADR consignment check (POST /api/adr/consignment-check) now reads the UN packaging mark of each line. Send the mark as printed in the new optional field packages.un_mark (for example "UN 4G/Y145/S/02/NL/VL823") and packaging_code_validity reads it by the grammar of ADR 6.1.3.1, field by field — the symbol, the packaging code of the 6.1.2.7 table, the packing-group letter and the relative density or gross mass, S or the test pressure, the year, the State, the maker — and holds the letter against the line's packing group (X covers I, II and III; Y covers II and III; Z covers III only). A malformed mark, or a letter that does not cover the packing group, fails with the sub-paragraph named; no mark is conditional, like the other optional inputs, so a load that passed with every vehicle-side input now reads conditional until its marks are sent. IBC and large-packaging marks are named and not read in this version, and whether the packaging kind is one the packing instruction allows is not checked. Second: a limited- or excepted-quantity line whose stated regime is met is no longer counted in the 1.1.3.6 total — ADR 1.1.3.6.5 leaves goods carried under those exemptions out — and exemption_1_1_3_6.engine.excluded_lines names it; a line that states no regime is still counted. The hosted MCP tool gives the same answer; it takes packages.un_mark from the next freightutils-mcp release.
The vehicle-side checks and the regime field that reached the ADR consignment check today are now described and typed on MCP: on the hosted MCP server and in the npm package freightutils-mcp 2.31.0, the adr_consignment_check description says it answers the vehicle side — orange-coloured plates and placards (ADR 5.3), the documents, fire extinguishers and equipment on the transport unit (8.1) and driver training (8.2) — and what fail, not_applicable and result.regime mean; its answer types the three blocks (placards_and_plates_5_3, equipment_8_1, driver_training_8_2), result.regime and the per-line verdict of the limited and excepted quantity checks; and its input takes the optional vehicle-side fields (vehicle.fire_extinguishers_kg, orange_plates_displayed, placards_displayed, equipment_8_1_5, instructions_in_writing) and the crew object, which the hosted tool refused until now. The same fields are accepted in the consignment argument of adr_exemption_calculator, consignment_calculator and shipment_summary, which list them as not read, as the web API already did. The description also names what the vehicle side does not assess: the vehicle approval of Part 9, the elevated temperature mark of 5.3.3, and handling and stowage. Packaging codes come next. Nothing removed: every 2.30.0 argument and answer field still validates; the one changed reading — not_applicable, not fail, for a relief that does not apply — is the web API's from earlier today, now in the package too.
Since the ADR consignment check (POST /api/adr/consignment-check) arrived on 4 November, a load outside the 1.1.3.6 small-load exemption answered status fail, because the exemption check read "not within 1.1.3.6" as a failure. That was wrong: a load above the 1.1.3.6 thresholds is carried under the full requirements of ADR, as most dangerous goods are, and a lawful fully regulated load must not read fail. The exemption check now answers not_applicable for such a load (the full requirements apply, and the other checks answer for a fully regulated load), and not_applicable does not change the status: 2 000 L of petrol in drums with the plates, extinguishers, equipment and a certificated driver declared now reads pass, and conditional without those inputs. The same applies to the limited and excepted quantity checks: a line that states no regime and fits neither is not_applicable, not fail; a line that states regime "lq" or "eq" and misses it still fails, and so does a load with an entry whose carriage is prohibited. fail now means a check the load does not meet. A new field, result.regime, says which rules the checks answered under: "1.1.3.6", "full_adr" or "undetermined". The same answer comes from the adr_consignment_check MCP tool on the hosted server.
The ADR consignment check (POST /api/adr/consignment-check) now answers the vehicle side of a road load in three checks: placards_and_plates_5_3 (the two orange-coloured plates of ADR 5.3.2.1.1, placards on a vehicle carrying packages of Class 1 or Class 7 under 5.3.1.5, the environmentally hazardous substance mark with placards, and the limited-quantity mark of 3.4.13 on a unit over 12 tonnes as a condition), equipment_8_1 (the documents of 8.1.2, the portable fire extinguishers of the 8.1.4.1 table by the transport unit's maximum permissible mass, and the 8.1.5 equipment selected by danger label) and driver_training_8_2 (the 8.2.1 certificate and the Class 1 or Class 7 specialisation). The consignment object describes the load, not what is on the vehicle, so each check first lists the requirements the load puts on the vehicle side, read from the Table A rows and the 1.1.3.6 verdict, and then checks them against new optional inputs — vehicle.fire_extinguishers_kg, orange_plates_displayed, placards_displayed, equipment_8_1_5 and instructions_in_writing, and a crew object (training certificate, specialisations, photo identification, Chapter 1.3 training). Without those inputs a dangerous-goods load is now conditional: acceptable provided the listed requirements are met. A load within 1.1.3.6 is read by 1.1.3.6.2: no plates or placards, one 2 kg extinguisher, the transport document, and Chapter 1.3 training for a driver without the certificate. Every rule is cited by volume and PDF page of the free UNECE ADR 2025 text; the extinguisher table and the equipment lists were read cell by cell from the page. The hosted MCP tool answers the three checks now; the hosted schema and the npm package freightutils-mcp carry the new inputs from 2.31.0. Packaging codes come next.
The mixed-loading check that reached the ADR consignment check this morning is now described and typed on MCP: on the hosted MCP server and in the npm package freightutils-mcp 2.30.0, the adr_consignment_check description says it checks mixed loading (ADR 7.5.2, with the foodstuffs precautions of 7.5.4), its answer types the mixed_loading_7_5_2 block (every pair of lines with the deciding cell of the 7.5.2.1 or 7.5.2.2 table, the quoted note and its page; the foodstuffs part; the conditions), and its input takes the optional vehicle.foodstuffs_carried, which the hosted tool refused until now. The same field is accepted in the consignment argument of adr_exemption_calculator, consignment_calculator and shipment_summary, which list it as not read, as the web API already did. What is still to come is named in the description: plates and placards (5.3), vehicle equipment (8.1) and driver training (8.2.1), then packaging codes. Nothing removed: every 2.29.0 argument and answer field still validates, and no answer changed.
The ADR consignment check (POST /api/adr/consignment-check) now checks mixed loading. The lines of one consignment object are one load in one transport unit, so every pair of lines is read as packages loaded together: against the 7.5.2.1 table by the danger labels of their Table A rows, subsidiary labels included, and, for two Class 1 lines, against the 7.5.2.2 table by compatibility group. Each pair gives the deciding cell of the table and its page. A lettered note the rows settle is settled (1.4S goods; the Class 9 life-saving appliances; safety devices UN 0503 with UN 3268); a note that turns on a fact the consignment object does not carry, such as approved segregation of compatibility groups B and D, answers conditional with the note quoted. A line whose limited or excepted quantity check passes is still read as a package of its row, because the object does not say it is carried that way, and the limited-quantity relief is stated as a condition. A load within 1.1.3.6 is checked all the same, because 1.1.3.6.2 does not lift 7.5.2. Goods with special provision CV28 engage the foodstuffs precautions of 7.5.4, and a new optional field of the consignment object, vehicle.foodstuffs_carried, says whether foodstuffs, other articles of consumption or animal feeds are in the same vehicle or container; for ADR 2027 the amended 7.5.4 is applied. Both tables were read cell by cell from the free UNECE ADR 2025 publication and every rule is cited by volume and page. The hosted MCP tool gives the same answer today; its description, and the npm package with the new field, catch up with freightutils-mcp 2.30.0. Nothing removed: every earlier field keeps its meaning.
The ADR consignment check that reached the web API yesterday is now an MCP tool, adr_consignment_check, on the hosted MCP server (the default tool list and the dangerous-goods scope) and in the npm package freightutils-mcp 2.29.0. One call on the consignment object gives the 1.1.3.6 verdict, the limited and excepted quantity check of each line, the tunnel restriction code of the whole load against the tunnels on the route, and each line's transport-document entry, with one status and every check cited to the ADR section and page it rests on. It gives the same answer as the web API, and a call made with an API key stores a check receipt, as the other checks do; without a key the answer carries the receipt's hash and nothing is stored. Mixed loading, plates and placards, vehicle equipment, driver training and packaging codes are not checked yet and are listed as such. The hosted tools that take the consignment object now also accept its four newer line fields: an ID-prefixed number, a quantity of 0, the ICAO limited-quantity statement and the inner packaging. The package also types two fields the 1.1.3.6 calculator began answering yesterday for a line whose entry is deleted in ADR 2027, such as UN 2941: strict_adr_2027.entry_deleted and items[].edition_basis. The MCP server now has 26 tools. Nothing removed: every 2.28.0 argument and answer field still validates.
The liability cap in the terms of service is now the greater of £50 and the fees paid in the twelve months before the claim, so the Free tier's cap reads £50 instead of £0; Pro stays up to £228. Changes to the terms no longer bind anyone by continued use. Every change is posted with a new revision date; a change that affects a Pro plan (its price, its request allowance, its listed features, or the terms as they apply to subscribers) is emailed to Pro subscribers at least 14 days ahead (30 for a change to rate limits, response formats or endpoints, as before), and takes effect for each subscriber at the start of the first billing period after that notice; a subscriber who does not accept it may cancel before then and keeps the current terms to the end of the period paid. The cap no longer opens by limiting what the terms say cannot be limited. Other changes take effect when posted. The refund policy's revision date moves to November 2026, with one word added to its cooling-off clause.
A new API endpoint, POST /api/adr/consignment-check, takes the consignment object (the lines of a road load, the journey date and the tunnel categories on the route) and answers in one call: whether the load is within the 1.1.3.6 small-load exemption (the exemption calculator's own answer, whole), whether each line's inner packaging fits the limited or excepted quantity of its Table A row (the LQ/EQ check's answer), the tunnel restriction code of the whole load and which tunnels on the route it may use, with and without the 1.1.3.6 relief, and each line's entry for the transport document in the order ADR prescribes, such as "UN 1203 PETROL, 3, II, (D/E)". The answer has one status (pass, fail, conditional or insufficient data), names any missing input by its path, and lists every check with the edition, the date and the ADR section and page it rests on. The tunnel and document rules were read from the free UNECE ADR 2025 publication with their page numbers. Where the text does not settle a case, such as a code that restricts tank and other carriage differently on a line with no packages stated, the check asks instead of guessing. This first version does not check mixed loading, plates and placards, vehicle equipment or driver training; those follow. A call made with an API key stores a check receipt, as the other checks do. The tool reaches the MCP server with the next freightutils-mcp release.
The /eval section on the dangerous-goods checks judged independently now shows every written load settled. The one load the three model readers still split on, an ADR 2027 load whose Table A entry ADR 2027 deletes, was delegated by the owner and settled by the owner's delegate after a fourth, independent reading agreed with one of the three; the table of who set each outcome gains that row, and the run of the checks over every settled load is the one made after today's calculator change. The audit linked from the section records the reading and the decision.
ADR 2027 deletes some Table A entries without naming where the goods go next; UN 2941 (fluoroanilines) is one. Until today the 1.1.3.6 calculator (the web API and the hosted MCP tool) refused such a load under ADR 2027. Transitional measure 1.6.1.1, as amended for ADR 2027, lets the whole of ADR 2025 be applied until 2027-06-30, so on a stated as_of from 2027-01-01 to 2027-06-30 an entry deleted in ADR 2027 is answered on ADR 2025's row inside the 1.6.1.1 window, with the strict result beside it: basis 1.6.1.1, basis_until 2027-06-30, the line counted on its ADR 2025 row (edition_basis "2025"), and strict_adr_2027 with entry_deleted true and no ADR 2027 total or verdict. The answer warns that the ADR 2027 entry for those goods must be established before 2027-07-01. After that window, or without a stated as_of, the request is still refused and names the deletion. Every change is additive (no field changes meaning) and is held by a practitioner test.
A free API key now stores up to 25 check receipts a day for its account (UTC day). From the 26th check that day the answer is the same in full and carries the receipt's hash exactly as a call without a key does, with _meta.receipt reason "quota", and nothing is stored. Pro keys have no daily receipt limit, and every receipt is still kept 90 days. New: GET /api/receipts/{check_id}/verify lets anyone holding a receipt's id confirm, without a key, that it exists and read its hash, schema version, creation time, tool and data edition, and nothing else. The pricing page now lists what each plan gets today, adds two questions (what a check receipt is, and whether the price will change) and states the Pro price hold: the plan's price will not rise for 12 months from the first payment. The hold lasts for as long as the subscription stays active, cancelling ends it, and it does not cover taxes. Pro is sold to businesses and professionals, and the pricing page and the terms now say so. The terms gain a paragraph on what a check result and its receipt are, and are not; they now say the cooling-off waiver is given by the confirmation at checkout, that liability for direct loss is capped rather than excluded, that nothing in them limits liability for death or personal injury caused by negligence, for fraud, or for anything the law does not allow to be limited, and that a price hold is not changed by the right to change limits or the terms. The Stripe checkout page now states the price, that it recurs monthly and where to cancel. The privacy policy says what a receipt id lets anyone read.
The npm package freightutils-mcp 2.28.0 now carries what the 1.1.3.6 calculator began answering today. adr_exemption_calculator takes icao_limited_quantity: an ID-prefixed line such as ID 8000 is counted out of 1.1.3.6 only when the request states that its packages are limited quantities to the ICAO Technical Instructions, and is otherwise held as needing its own UN entry. It accepts an item quantity of 0, read as goods not carried (EXCLUDED_NIL). It types the new answer fields: figure_basis when a gross mass on an entry counted by mass clears the load as an upper bound, and on ADR 2027 answers basis, basis_until, strict_adr_2027 and transitional_1_6_1_1. Once ADR 2027 applies, an answer for a stated date headlines the transitional measure that clears the load and shows the strict ADR 2027 figure beside it. adr_lookup types the code registry's review statuses, owner_reviewed among them. The hosted MCP server's tool descriptions come from the package and say the same. Nothing removed: every 2.27.0 argument and answer field still validates.
The ADR code pages under /adr/codes are now open to search engines and listed in the site map. They had been kept out of search while the disagreements found by the machine review waited for the owner; the owner has now answered the last of them and cleared both review batches. Every page still says plainly how far it has been reviewed: machine-reviewed means two independent readings of the free UNECE text agreed with the record, not a person's review. Two records were corrected on the owner's answers: packing instruction P203 now gives the test pressure of a vacuum-insulated closed cryogenic receptacle as at least 1.3 times the sum of the maximum internal pressure and 1 bar (1.3 times the maximum internal pressure for the others), as Volume II prints it, and P602's second option now says that each inner closure is held in place so it cannot back off or work loose in carriage. Those two are now marked reviewed by the owner. The /eval page also gains a section on the dangerous-goods checks judged independently: written test loads whose outcomes were set from the ADR text before the checks answered, with the results and their limits.
Four answers of the ADR 1.1.3.6 exemption calculator (the web API and the hosted MCP tool) change, each where independent readers of the regulation text, judging test loads before seeing the calculator, disagreed with it and then chose the change by majority. Under ADR 2027, where ADR 2027 itself does not clear a load but a transitional measure open on the stated carriage date does, the answer is now given on that measure and names it: basis 1.6.1.61 (the ADR 2025 note-a rule for UN 1005 and the eight other listed numbers, until 31 December 2030) or 1.6.1.1 (ADR 2025 as a whole, for as long as 1.6.1.1 allows), with its end date and the strict ADR 2027 figure beside it in strict_adr_2027, and without the words "full ADR compliance required". Without a stated as_of the strict reading stays the headline. A gross mass on an entry counted by mass is now used as an upper bound of the net mass: the load is answered exempt on it, with figure_basis upper_bound_gross and a warning, only when it clears even on the gross figure; otherwise, and always for an entry counted in litres, the answer still asks for the net quantity. An item sent with quantity 0 is read as not carried: it is left out (state EXCLUDED_NIL) and named in a warning, and a load of nothing but zeros is refused. An ID-prefixed line such as ID 8000 is now counted out only when the request states icao_limited_quantity: true — that its packages are limited quantities conforming to the ICAO Technical Instructions; without that statement the line is blocked and the load has no verdict unless the other lines settle it, because the marking alone does not establish 3.4.9 or 3.4.10. Every change is additive (no field changes meaning) and is held by a practitioner test.
The 304 ADR code records the machine review left for the owner have been read a second time against the provision's full text in the free UNECE publication of ADR 2025 (ECE/TRANS/352), and against the section or Table A row each disagreement pointed to. Almost every disagreement was settled on the page: the value is printed as the record has it, belongs to a special packing provision with its own record, is printed as deleted, or is a table row the record deliberately does not reproduce. Where the page showed the record short or off, it was corrected: CV15 now says what 7.5.5.3 sets (organic peroxides, self-reactive and polymerizing substances limited to 20 000 kg per transport unit) instead of calling it a mixed-loading rule; LP03, LP904, LP905, P004, P112(c), P505, P910, S1 (Class 1 supervision), TE7, TT11 and label 9A restate what their pages print; P910 gains the note that its packagings may exceed 400 kg net, and 14 records gain cross-references printed in their own text. Nothing was removed. 962 records are now machine_reviewed and 2 contested (both go to the owner); none says needs_fix. Code pages still stay out of search indexes until the owner clears each review batch.
Every one of the 964 ADR code meanings on /adr/codes has been read again against the free UNECE text of ADR 2025 (ECE/TRANS/352) by two independent machine readers: one re-extracted each code's values from the page without seeing the record, one checked the plain-words paraphrase against the page, and a second reader re-read every record where they disagreed. 660 records came through with nothing to look at and now say machine_reviewed; 302 where a reading disagreed with part of the record say contested and go to the owner to adjudicate; 2 say needs_fix (a limit both readers found that the record leaves out: P115's 30 kg propellant limit and P407's 700 matches per inner packaging). No meaning, number or code was changed in this review. A machine review is not a person's review: pages say machine-reviewed in those words, the hub and family pages count each status, and code pages stay out of search indexes until a person has reviewed them or the owner has adjudicated the review's disagreements. verification_status on /api/adr/codes, /api/adr resolved_codes and the plain-words statements carries the new values machine_reviewed and contested.
October 2026
The npm package freightutils-mcp 2.27.0 now carries what the hosted MCP server and the API began serving on 30 and 31 October. airline_lookup types the airline record's new fields: operational_status (active, ceased or unknown) with its basis — a separate thing from the verification status — prefix_shared and prefix_holders on every record with an AWB prefix, restrictions[] when the EU or UK Air Safety List names the carrier (a flag with the list and its date, not a legal determination), and the carrier's own cargo, tracking and dangerous-goods pages where it publishes them. nearest_airport takes include_all: by default it ranks airports with scheduled airline service and an IATA code (no heliports, no closed fields), include_all: true lists every type, and every answer states distance_basis and the filter applied; 2.26.0 refused include_all before the call reached the API. Airport records type the refreshed OurAirports fields (86,248 airports). validate types failure (reason, message, expected format, example) and resolve_reference types validation and failure. Nothing was removed. With the hosted server pinned to 2.27.0, its airport_lookup and nearest_airport descriptions say 86,248 airports and name the default and include_all. Published from GitHub Actions with npm provenance.
The identifier validator (GET /api/validate and the validate tool) now says why an identifier failed, in a reason an agent can branch on: INVALID_CHECK_DIGIT, WRONG_LENGTH or BAD_CHARSET, with a message, the expected format and an example; an AWB whose airline cannot be named says UNKNOWN_PREFIX (no record holds the prefix — the dataset is not the IATA registry) or AMBIGUOUS (several do), and a type the validator does not check answers UNSUPPORTED_TYPE. resolve_reference (GET /api/resolve) now resolves a full 11-digit AWB number and an IMO number, runs the check digit as it resolves an AWB, container or IMO number, and still names the airline of an AWB whose check digit fails, with the failure stated; it says NO_MATCH or AMBIGUOUS when it cannot give one answer. A valid answer is unchanged.
nearest_airport (GET /api/nearest-airport and the hosted MCP tool) now ranks, by default, the airports a shipment can fly from: airports with scheduled airline service and an IATA code, never a heliport or a closed field — from Heathrow's coordinates the answer is LHR, London City, Gatwick, Luton, Stansted, not RAF Northolt or a business-park heliport. include_all=true lists every type as before, and a type filter works as before. Every answer states its distance basis (great_circle) and the filter applied. The airport data is refreshed from OurAirports' 30 October 2026 file: 86,248 airports, with scheduled service, ISO region code, GPS code and keywords now in every airport answer. Rows OurAirports marks "[Duplicate]" answer their exact code only and no longer appear in name searches or nearest results.
The 395 cargo airline records carry four new things, each with its own source, URL, access date and sha256. Whether the carrier operates: active, ceased or unknown, with the basis — the carrier's own page, or its ICAO designator listed for it in the current FAA Order JO 7340.2 (Change 3); ceased also when the FAA no longer lists the designator for the carrier and its former website no longer answers. 237 records are active, 38 ceased and 120 unknown: Wikidata is shown as second-class evidence and never makes a carrier ceased on its own. Whether a public air-safety list names the carrier: 11 records carry the EU Air Safety List entry (Commission Implementing Regulation (EU) 2026/1317, 8 June 2026) or the UK Air Safety List entry, matched on the ICAO designator or on name and state, printed as "listed in …" — a flag with its source, not a judgement. Whether the AWB prefix is shared: every record with a prefix says so and lists every holder (61 of the 354 prefixes sit on more than one record). And where the carrier publishes them: its cargo website (86), its AWB tracking page as a template (2) and its dangerous-goods page (22), each answering when read. The verification status of the identifiers is unchanged, and no code, prefix or name changed. On /api/airlines, the hosted airline_lookup tool and the airline pages.
Page titles and descriptions now lead with the identifier the way lookups write it. HS code pages: "HS Code 850990 — …", with the dotted form, the heading and the chapter in the description. HS heading pages: "HS Heading 8431 — …", with the chapter in the description. Airline IATA pages: the ICAO code beside the airline name ("Airline IATA Code M6 — Amerijet International, ICAO AJT"). Shaped from the search queries recorded in the public eval; page content, addresses and data are unchanged.
The public freight-reference eval has a v0.5 run: 40 new questions whose answer is fixed by what an ADR Table A code means — a packing instruction's limit, a special provision's threshold, the 1.1.3.6 maximum of a transport category, a tank code's calculation pressure — each expected value taken from the ADR code registry and checked against the live site (340 of 340 values). With the tools the model answered 38 of 40 in every include_codes mode, and every answer it gave was right: the two misses were questions the provider's API blocked before the model replied. Per correct answer, compact cost 1.02 times and full 2.04 times what none cost, because without the meanings the model looked the codes up itself; full changed no answer. The default stays compact. The questions and results are in release v0.5.0 of the eval repository (DOI 10.5281/zenodo.23820777); /eval and the agent-builders guide show the v0.5 figures.
The npm package freightutils-mcp 2.26.0 now carries what the hosted MCP server and the API already serve. The five tools that read ADR data — adr_lookup, adr_exemption_calculator, adr_lq_eq_check, consignment_calculator and shipment_summary — take edition (2025 or 2027) and as_of (a date): edition wins, otherwise the date picks the edition — ADR 2025 before 2027-01-01, ADR 2027 from then, with ADR 2025 still allowed until 2027-06-30. ADR 2027 is accepted and applies from 2027-01-01; an ADR 2027 answer for an earlier date says it is not yet applicable. The ADR 2027 answer fields (adr_edition, the 1.6.1.61 note-a basis beside the 1.1.3.6 verdict, the UN 2372 tank-code note, the amended code meanings) and the three edition codes are typed, so a validating client accepts them; 2.25.0 refused edition before the call reached the API. A check receipt is stored only for a call made with an API key — a free key is enough; without one, the answer carries the receipt hash in _meta.receipt, now typed in the package, and nothing is stored. Nothing was removed. With the hosted server pinned to 2.26.0, the five tools' descriptions there name edition and as_of too. Published from GitHub Actions with npm provenance.
Check receipts are stored for calls made with an API key — a free key is enough; a call without a key gets the receipt hash in the answer and nothing is stored. A check tool called without a key answers exactly as before, with the X-FreightUtils-Receipt-Hash header and, under envelope 1.2 and on the hosted MCP server, _meta.receipt: the hash and the fields it covers, so it can be recomputed from the answer alone. There is no check_id and no record to retrieve. A keyed call, free or Pro, gets a check_id and a receipt kept for 90 days, as before. Receipts already stored are kept as they are, and reading one by its id stays free. The privacy policy says the same.
The Data licensing page now has a row for the ADR 2027 Table A values the ADR tools serve with edition=2027: UNECE names no open licence for the ADR publication or its amendment documents, the values are served per query for the UN number asked about, and their status is the same as ADR 2025's — served per query, under review. The dataset registry names the five tools that answer from it.
The public freight-reference eval has a v0.4 run. The 30 questions on what ADR 2027 changes in Table A are now answered through adr_lookup with edition=2027: with the tools, three of the four models answered all 30 (on 21 October, with the tools serving ADR 2025, no model answered more than one); the fourth was not run. On all 300 questions, the model re-run with the tools answered 300 of 300. The run also measured what the code meanings cost: on the ADR Table A and 1.1.3.6 questions every include_codes mode answered every question right, and compact cost 1.45 times and full 2.84 times what none cost per correct answer. The results, the question set and a finding aid to the ADR 2027 amendment documents are in release v0.4.0 of the eval repository (DOI 10.5281/zenodo.23801448). /eval and the agent-builders guide show the v0.4 figures.
The five tools that read ADR data — the ADR lookup, the 1.1.3.6 calculator, the LQ/EQ check, the consignment calculator and the shipment summary — now take edition (2025 or 2027) and as_of (a date), on the API and the hosted MCP server. edition wins; otherwise the date picks the edition by the legal timeline: ADR 2025 before 1 January 2027; ADR 2027 from then, with a notice that ADR 2025 may still be applied until 30 June 2027 (1.6.1.1), and edition=2025 still answers until that day. ADR 2027 is accepted (C.N.395.2026); an ADR 2027 answer for an earlier date says it does not yet apply. Under ADR 2027, UN 2956 has no limited quantity, UN 2941 is deleted, UN 2348 and 2862 have a packing group II row (their packing group III row moves to variant_index 1 — send packing_group), and UN 3561 to 3564 are new. ADR 2027 deletes 1.1.3.6.3 note a: 25 kg of UN 1005 scores 1,250 points and is not exempt, and the answer shows beside it the ADR 2025 note-a basis (500 points, exempt), which 1.6.1.61 allows until 31 December 2030; the shipment summary and the /adr-calculator page show the same. UN 2372 under ADR 2027 names tank code L4BH and the measure that keeps LGBF for tanks built before the switch. Code meanings that ADR 2027 amends say so. A caller who sends neither edition nor as_of gets the same answers as before until 1 January 2027; the check receipts name the edition and the date.
FreightUtils now holds an ADR 2027 Table A: the ADR 2025 table, already checked cell by cell against the free UNECE publication, with every amendment of ECE/TRANS/WP.15/274, its Addendum 1 and its Corrigendum 1 applied — 2,944 rows and 2,350 UN numbers (six new entries, UN 2941 removed, 212 values changed on 183 rows). A build check matches every difference between the two tables to its amendment and every amendment to its difference, 223 of 223 both ways. ADR 2027 is accepted (C.N.395.2026) and applies from 1 January 2027; when UNECE publishes the consolidated ADR 2027 volumes, the table will be checked against them cell by cell. No answer changes today: every tool still answers on ADR 2025.
The npm package freightutils-mcp 2.25.0 now carries what the hosted MCP server already served: --scope dg, customs or ops starts it with only that scope's tools (5, 5 or 17 of 25); adr_lookup, unlocode_lookup, airport_lookup, airline_lookup and hs_code_lookup take a list of up to 50 codes and uk_duty_calculator up to 10 lines, answered in input order with a status for each; the six checks' answers carry their check_id and receipt_url and name the edition they rested on; adr_lookup answers what each Table A code means (resolved_codes, include_codes); and resolve_reference decodes ADR Table A codes such as TU9. Nothing was removed: every earlier call answers as before. With the hosted server pinned to 2.25.0, its adr_lookup description now names all 20 Table A columns and resolved_codes, and its resolve_reference description names the ADR code families. Published from GitHub Actions with npm provenance.
The build checks that keep the ADR code meanings complete, cited and in FreightUtils' own words, and the check that every dataset lists what its source holds that is not served, now have tests of their own. On copies of the files, the tests delete a code's record, change a derived count, move a citation outside its part of ADR, mark a meaning as reviewed with no review on record, copy ten words of the publication into a meaning, add an unlisted column to a source's field list and let an undecided field go past 90 days; each check must fail on each one, and pass on the untouched copies. Both tests run on every build. No served value changed.
An ADR lookup's resolved_codes now sits beside results[] instead of on every row: each code's meaning is given once per answer, with columns (the Table A columns it is printed in) and rows (the results[] rows that carry it). A UN number with several rows no longer repeats the same meanings on each one: UN 1263, six rows, is 24,485 bytes by default instead of 45,418. The compact form leaves out url (a code's page is /adr/codes/{family}/{code}); include_codes=full keeps it. The rows still carry the codes as Table A prints them, and include_codes, its defaults and every meaning are unchanged. Same on /api/adr and the adr_lookup MCP tool.
The six check tools — the ADR 1.1.3.6 calculator, the LQ/EQ check, the UK duty calculator, the consignment calculator, the shipment summary and the ICS2 check — now name the edition of the data each answer rested on, in the answer (?envelope=1.2 and the hosted MCP) and in its check receipt: ADR 2025 with the dates it applies, for the ADR checks and for a consignment or shipment line with a UN number; the UK Trade Tariff, read live, from the date of the answer; the version of the ICS2 stop-word list. A check that read no regulatory dataset says so in _meta.edition_basis instead of leaving the edition empty. Receipts issued from now on are receipt.v2, with the as-of date and that basis inside receipt_hash; earlier receipts are kept as issued and still verify, and the X-FreightUtils-Receipt-Schema header says which version a receipt is. No answer value changed.
Every ADR 2025 Table A value FreightUtils serves (2,939 rows, 24 fields each: 70,536 cells) and the excepted-quantity limits of 3.5.1.2 (E0 to E5) were checked on 29 October 2026 against the free UNECE publication ECE/TRANS/352, Volume I, with Corrigendum 2. Every cell and every limit is the same. No value changed. The Data licensing and Methodology pages now say so, and a build check fails if a served Table A value changes without the comparison being run again.
For each dataset behind an API route or MCP tool, FreightUtils now keeps a copy of the source's own field list — the columns of the source file, the fields of its records, or, for the ADR code meanings, the codes Table A prints — and a list of the fields it does not serve, each with the reason and the date that was decided. A check fails the build when a source field is neither served nor listed, when a listed field has gone from the source, or when a field is listed as unserved although it is served. ADR Table A: all 20 columns served. The monthly audit prints, per dataset, how many source fields are served.
Each UN page now has an "In plain words" block under its Table A table: what the tunnel code allows, the transport category's small-load limit, the labels, and the packing, tank and carriage provisions, each with the ADR section it comes from and whether it has been reviewed. The text is assembled from the meanings of the row's codes and from the tunnel-code and transport-category tables, never written separately, and a statement is only as reviewed as the least-reviewed code it uses. The same statements come with /api/adr?include=statements and with include_statements on the adr_lookup MCP tool; nothing changes unless you ask for them.
An ADR lookup by UN number now says what each of the row's codes means: /api/adr and the adr_lookup MCP tool add resolved_codes, one entry per code in Table A columns (5) to (20), with a one-sentence summary, the ADR section and whether the wording has been reviewed (include_codes=none, compact or full; compact by default for a UN number, none for name searches, class lists and batches). GET /api/adr/codes lists the 17 families and their codes, and GET /api/adr/codes/{family}/{code} returns one, with its limits and conditions as data. resolve_reference now recognises a code such as TU9, lists every family that holds an ambiguous one, and says when a well-formed code is not in Table A. /adr/codes has a page for every family and every code, and every code in the Table A table on a UN page links to its page. All 964 meanings are FreightUtils' paraphrases, not yet reviewed, and say so; code pages stay out of search indexes until reviewed.
FreightUtils now holds a record for each of the 964 distinct codes printed in the ADR 2025 Table A columns that carry codes: special provisions, packing instructions and the special provisions inside them, mixed packing, portable tank and bulk container codes, tank codes decoded letter by letter, tank special provisions, vehicle designations, carriage provisions, label models and hazard identification numbers. Each record says what the code means in FreightUtils' own words, written from the free UNECE publication (ECE/TRANS/352) and cited to its volume, section and page; only a provision's own heading is quoted. None has been reviewed yet, and each record says so; once a review lands it says who reviewed it, by role. The record format is published at /schema/adr-code.v1.json. The next update serves the records.
The hosted MCP server now also answers at /api/mcp/dg (the five dangerous-goods tools), /api/mcp/customs (the five customs and trade tools) and /api/mcp/ops (the seventeen freight-operations tools). An agent that connects to a scope loads fewer tool definitions: the dangerous-goods scope is about a quarter of the full list. The tools, their answers, the key and the limits are the same as on /api/mcp, which still serves all 25 tools. A refusal for a missing input now names only the inputs the call lacked, never one it sent or an optional one, and the consignment-object argument and the 1.1.3.6 calculator's arguments state their rules once instead of on every field.
The ADR, UN/LOCODE, airport, airline and HS lookups now take up to 50 items in one call (un_numbers or codes, comma-separated, or as an array on MCP), and the UK duty calculator up to 10 lines. Each item is answered exactly as its single call would be and comes back in the order sent, with its own status, so one code that does not exist is named instead of failing the rest. A batch counts as one call against the daily limit and as one call per item against the Pro monthly allowance. Single calls answer as before.
When some lines of a load cannot be resolved, for example a UN number with several Table A rows and no packing group, the ADR 1.1.3.6 calculator could already answer "not exempt" if the lines it could resolve were over the limit on their own. It now says so in four fields: the points of the resolved lines, whether those alone exceed 1,000, whether the verdict rests on all items or on the resolved items alone, and which lines were left unresolved. 350 L of petrol beside paint and aerosols with no packing group reads 1,050 points from the petrol alone, over the threshold, not exempt on the resolved items alone. The total and the exempt verdict are unchanged, on the website, the API and both MCP surfaces.
Each answer of the ADR 1.1.3.6 calculator, the LQ/EQ check, the UK duty calculator, the consignment calculator, the shipment summary and the ICS2 check now carries a check_id and a receipt URL (in the X-FreightUtils-Check-Id and X-FreightUtils-Receipt headers, in _meta under envelope 1.2, and in the MCP answer). GET /api/receipts/{check_id} returns the record of that check for 90 days: the inputs it used, with units and defaults spelled out, the edition and data files it read, the rules applied with their section, the result and anything left unresolved, a hash to check a copy against, and a sentence on what the receipt does not prove. Receipts never hold names, addresses, keys, IP addresses or long free text. Lookups carry no receipt. Answers that carry one are no longer cached, and their values are unchanged.
A load can now be described once and sent to three checks: the consignment object holds its lines (UN number, packing group, the dangerous-goods quantity with its unit and basis, packages with their count and gross mass, HS code, value, dimensions), the journey, the vehicle and options, and only the lines are required. POST /api/consignment, /api/shipment/summary and /api/adr-calculator, and the matching MCP tools, accept it as { "consignment": ... } and give the same answer as the equivalent request in their own shape. When the object lacks an input a check needs, the answer names it by path, such as lines[0].dims, instead of guessing. POST /api/validate checks an object and says which of the three checks it can feed. The JSON Schema is at /schema/consignment-object.v1.json. Every existing request shape answers as before.
Every tool on the hosted MCP server and in the npm package freightutils-mcp 2.24.0 now answers with response envelope 1.2: beside the answer, its status (ok, partial, not found, unsupported or ambiguous), the inputs a partial answer still needs, the edition it is served from with the dates that edition applies, and the date its dataset was last audited. A refusal carries the same fields, so an agent can read which input is missing. No value in any answer changes, and the text of a successful answer is unchanged. The unlocode_lookup description now says 116,086 UN/LOCODE codes. Published from GitHub Actions with npm provenance.
The first guide under /guides retells the eval for people building agents: what a model gets right alone, what the FreightUtils tools change, what web search changes instead, and what each costs, with the questions where the tools still say “unknown” (ADR 2027). Every figure in it is filled from the same pinned eval results the /eval page renders, and a check fails the build if the two ever differ. The /guides page now lists it, links the dangerous-goods guides that live beside their tools, and is open to search engines.
A client generated from /openapi.json could reject real answers: the spec marked fields as never null that are null on some records (ADR packing group, special provisions, tunnel code and others on the scope-flagged rows; airline IATA codes, callsigns and AWB prefixes), named the container tare weight with a typo, and described the resolver’s source block with the wrong shape. Its nullable markers were in an older OpenAPI dialect that the declared version ignores. The spec now types those fields as they are served, documents the API key (X-API-Key header or bearer), the 429 rate-limit answer with its headers, and the envelope and client parameters on every operation. A check now runs every example call the spec documents against the schema it documents.
Until 29 October, the emissions calculator used ADEME’s 2,000–5,000 km factor for a large air freighter at every distance, so a 9,000 km flight came out about a fifth too high and a short hop far too low. It now chooses ADEME’s own band for the distance — under 500 km, 500–1,000, 1,000–2,000, 2,000–5,000 or over 5,000 km — and the answer names the ADEME element and the band. The six standard articulated trailers, tagged as European, showed the UK’s 44-tonne gross weight with no mention of the regime; each now says its 44 tonnes is the UK figure and gives the 40 tonnes Directive 96/53/EC allows in general international traffic, with both sources.
Free-text searches on the ADR, airline and airport lookups, through the API and the MCP tools, now take at most 200 characters; a longer one is refused straight away with a 400 that names the limit. A search far longer than any name could hold a server for minutes while it looked for near matches, and the near-match ranking is now bounded as well. The optional ?client=<your-app> label, which changes no answer, was refused with a 400 on the ADR 1.1.3.6 calculator, the LQ/EQ checker and the shipment summary; every route now accepts it.
Until 29 October, looking an airport up by ICAO code on the airport lookup, its API and the MCP tool matched the OurAirports identifier, which is usually but not always the ICAO code. About 690 airports could not be found by their real ICAO code — Lima (SPJC) and Lusaka (FLKK) among them — or the code found a different record. The lookup now matches the airport’s ICAO code first and the OurAirports identifier only as a fallback, says which it matched, and returns both as separate fields. The airport data itself is unchanged: the ICAO codes come from the same OurAirports file already served.
From 10 August to 29 October, about 12,600 UN/LOCODE place names with accented letters were shown garbled on the lookup, its API and the MCP tool — Bodø as “Bodø”, and a search for “Malmö” found nothing. The names are now read correctly from the same UNECE 2025-1 release files and checked against UNECE’s XML release and its current pre-release. Codes that UNECE lists twice, under a bilingual name in both orders, are now one location with the other name as an alternate name, so the count is 116,086 locations rather than 116,232. The 19 entries UNECE marks for removal from the next issue now say so. Where UNECE leaves an airport’s IATA column blank because the IATA code equals the location code, the code is now given when OurAirports confirms it, and the answer says how it was derived. One airline record’s garbled name was corrected too (Fundació Rego).
Until 29 October, the UK duty calculator, its API, the MCP tool and the shipment summary answered 0 % duty for any commodity whose UK third-country duty is not a plain percentage: white sugar (1701991000), charged £35.00 per 100 kg, came back as 0.0 % duty with no warning, and so did compound duties such as garlic’s 8 % plus £100 per 100 kg. The calculator only computes percentage duties, so it now says so for these codes: the answer is marked unsupported, quotes HMRC’s duty expression word for word, and gives no duty, VAT or landed-cost figure. Every answer also lists what the figures never include, and names any preference for the origin’s trade group, tariff quota, suspension or import restriction HMRC lists. When the UK Trade Tariff does not answer, the API now returns a retryable 503 instead of saying the commodity code was not found.
Until 29 October, the machine-readable description each ADR page gives search engines (its schema.org Dataset block) named FreightUtils as the creator of the ADR 2025 Table A data and stated a licence for it: CC BY 4.0 on every UN page, and our terms on the ADR lookup page. Neither was right. The values come from UNECE’s ADR 2025 publication, for which UNECE names no open licence and we hold none, as the data licensing page says. The blocks now name UNECE as the source and creator, FreightUtils as the publisher of the page, point to the data licensing page, and state no licence. A check now fails the build if any page states a licence the data licensing page does not record.
Since 28 October, the npm package freightutils-mcp (2.23.0 and earlier) failed every tool except get_subscribe_link when run without an API key in a client that checks answers against the tool’s output schema, such as Claude Desktop or Cursor: the anonymous answer carried a field the package’s schema did not allow. From 29 October the sentence telling anonymous callers how to get a free key travels in an X-FreightUtils-Identify response header instead, and in the answer body only when you ask for ?envelope=1.2. Installed packages work again with no upgrade, and a daily test now runs the published package against the live API.
Add ?envelope=1.2 to any API call and the answer states its status (ok, partial, not found, unsupported or ambiguous), the inputs a partial answer still needs, the edition it is served from with the dates that edition applies, and the date its dataset was last audited; no value in the answer changes, and the existing envelopes are unchanged. A call made without an API key now says so in its answer, with the anonymous daily limit and how to get a free key; add ?client=<your-app> to label your calls.
The ADR lookup, airline code and HS code pages and the ADR 1.1.3.6 exemption calculator no longer download a whole dataset with the page. Their search boxes now ask the server as you type and show the first 50 matches, from the same search the API runs, so each page drops from 1.2–1.8 MB of HTML to under 200 KB. The LQ/EQ checker looks up a UN number’s packing-group rows when you type it, the ULD and vehicle pages stop carrying source notes they never displayed, and the HS page now links every chapter directly. The airline table’s CSV download now saves the page of results on screen.
The adr_lookup tool description now names the packing, tank, vehicle and carriage provisions the ADR lookup has returned since today’s Table A update, on the npm package and the hosted MCP server, and the package’s result schema lists them; the package moves to zod 4.6. Published from GitHub Actions with npm provenance.
The ADR lookup, its API and the MCP tool now return every column of an ADR 2025 Table A entry, not only the identification, classification, quantity, transport-category and hazard-identification columns: packing instructions and packing provisions (columns 8, 9a, 9b), portable tank and ADR tank codes and provisions (10 to 13), the vehicle for tank carriage (14) and the special provisions for carriage (16 to 19). The values are the codes Table A prints, read from the UNECE Table A workbook and checked field by field against it. Each UN page now opens with one sentence giving the entry’s Table A values and the source, then the full Table A entry as a table (one per entry when a UN number has several), then what ADR 2027 changes for that entry — or that it changes nothing in Table A. ADR 2027 amendments to the packing, tank and carriage columns now show on the UN pages too.
The /eval page now answers the first question an agent builder asks: why not just let the model search? A third condition gives Claude Fable 5.1 its provider’s own web search and no FreightUtils tools, on the same questions with the same scoring, and the page shows its accuracy, how many searches it made, its cost per answer against the tools, and which sites the answers came from, freightutils.com included. The results table now leads with the questions outside ADR 2027, the ones the FreightUtils tools cover; the ADR 2027 questions are their own section beneath it, as the gap in our own tools, and the figures over all questions stay in a second table. Every figure is still rendered from the committed results of the public freightutils-eval repository.
A new page, /eval, shows how often AI models answered freight reference questions correctly on their own and with the FreightUtils tools, from the committed results of the public freightutils-eval repository: the results by model and condition, accuracy by question family, every confident-wrong answer with the primary source it was checked against, ADR 2027 as the gap in our own tools, and the method with its limits. FreightUtils ran the eval of its own tools; every figure on the page is rendered from the committed results, and anyone can re-run it.
ADR 2027 check: paste a list of UN numbers and see what changes for each, with the UNECE reference; CSV and print. The new page /adr/changes-2027/check takes up to 200 UN numbers, written 1203 or UN 1203, separated by commas, spaces or new lines. For each it shows what that UN page’s ADR 2027 block shows: the ADR 2025 and ADR 2027 values of every served column that changes, the columns that change but FreightUtils does not serve, new and deleted entries, or that its Table A columns do not change, each with the UNECE document and page. A number that is not in ADR 2025 Table A is listed as such. A summary counts each group, the result downloads as a CSV file, and the print view keeps every citation. The ADR 2027 changes page and every UN page link to it. It is free and needs no login.
A new page, /data-licensing, lists every dataset FreightUtils serves with its publisher, what we know of the publisher’s terms, the attribution to show beside the values, and a status. ADR 2025 Table A values, the HS 2022 descriptions and the airline code compilation are marked “served per query, under review”: they come from publications whose publishers assert rights, and we serve one answer at a time with attribution. The Terms of Service have a new section, Use of the data: you may store our answers, show them in your own products and cite them with the source line; you may not extract the datasets in bulk or redistribute them as data. A rights-holder who believes our use goes beyond their terms can write to us through the Contact page, and we reply within ten working days.
HS and ADR UN pages limit how many pages one address can open in a few minutes, to stop scrapers. That limit also slowed down Google’s and Bing’s crawlers. A crawler that proves who it is now crawls those pages without the limit: its address must look up, in reverse, to a Google or Bing crawler host name, and that name must look up back to the same address, the check Google and Bing publish. The answer is remembered for a day per address. Calling yourself Googlebot or Bingbot is not enough: an address that does not pass the check keeps the same limit as everyone else.
Some HS headings have no subheadings: their only 6-digit code is the heading followed by 00, so natural honey is heading 04.09 and code 0409.00 and nothing else. Such a heading had a page of its own that said the same as its code page. The heading’s address now redirects permanently to the code page (/hs/heading/0409 opens /hs/code/040900), the chapter page links the code page directly and says the heading is not subdivided, and the code page names the heading without linking back to the redirect. Every other heading page is unchanged.
The 30 airline chargeable-weight pages each repeated the calculator with one carrier’s volumetric divisor. They are now one table on the chargeable weight calculator: carrier, IATA code, volumetric divisor and source, with a link to each carrier’s airline code page. Picking a carrier sets the calculator to its divisor and says so above it. Each old address, for example /chargeable-weight/emirates-skycargo, redirects permanently to the calculator with that carrier preset (/chargeable-weight?airline=emirates-skycargo). The divisor table records no source for each carrier, so the table says to check the carrier’s published conditions.
Most AWB prefixes belong to one airline that also has an IATA code, so the prefix page and the airline’s code page said the same thing twice. Such a prefix’s page now redirects permanently to the airline’s code page, which states the prefix: /airlines/awb/176 opens /airlines/iata/EK, Emirates, AWB prefix 176. A prefix page stays where it says something the code page cannot: a prefix held by more than one record, or one whose holder has no IATA code. Those prefix pages and their airlines’ code pages now link each other. The resolver’s page link for an AWB prefix now names the page that answers, and a record with no page of its own no longer carries a link that went nowhere. Airline results in the site search now open the airline’s code page; they used to open a chargeable-weight page that exists for 30 carriers only. Code pages for airlines with no cargo operations still answer, but are no longer offered to search engines.
ADR 2027 was accepted on 9 October 2026 (C.N.395.2026) and applies from 1 January 2027; ADR 2025 may still be applied until 30 June 2027. Every ADR UN page now shows what the ADR 2027 amendments change for that entry: the ADR 2025 value, the ADR 2027 value and the UNECE document and page it comes from. Entries with no Table A change say so. A new page, /adr/changes-2027, lists all 38 changes on served columns across 23 UN numbers, the 6 new entries and the deleted UN 2941. FreightUtils keeps serving ADR 2025 data until the switch on 1 January 2027.
Thirteen new pages, one per ADR class or division (for example /adr/class/3 for flammable liquids), list every UN number in that class with its proper shipping name and packing groups, each linked to its entry. The ADR lookup page links all thirteen, and every UN page links back to its class. The “Other substances in this class” block on a UN page now shows the UN numbers nearest to it in the same class, instead of the same first five on every page. HS code pages no longer say “Tariff & Duty” in their titles: they show the HS 2022 description and link to the UK Trade Tariff, and the duty calculator is where duty rates are.
An HS code written with dots or dashes in the address (for example /hs/code/0901.11) opened a second copy of the code’s page; it now redirects permanently to the one page for that code (/hs/code/090111). On the Airline Codes page, the popular AWB prefix cards opened the unfiltered list instead of the prefix, and each carrier in the list linked through a redirect; the cards now open the prefix’s own page and each carrier links straight to its code page.
/sitemap.xml is now a sitemap index with one child per kind of page: site pages, HS sections and chapters, HS headings, HS codes, ADR UN numbers, airline pages, and the CBM, pallet, container and Incoterms pages. Each data page is dated by the last change to its own data; pages with no dataset behind them carry no date, instead of the build date. The empty Guides page is no longer offered to search engines until it has a guide, robots.txt no longer asks crawlers to wait between requests, and each deploy now tells IndexNow (Bing and others) only about the pages it changed, not the whole site.
Georgian Airways was in the airline dataset twice, both with IATA A9 and ICAO TGZ, so a lookup by A9 answered two holders and said the code alone did not identify one carrier. The two records are now one: the record verified against Georgian Airways’ own conditions of carriage and the FAA’s list of ICAO designators. The old address of the duplicate redirects to it. The dataset now holds 6,356 airlines, and every page and answer that states the count says so.
The airline_lookup tool description now says 6,356 airlines, matching the airline dataset after the Georgian Airways duplicate was merged. Description and README text only; no tool, argument or result changed. Published from GitHub Actions with npm provenance.
The airline lookup answered IATA code PH as Polynesian Airlines’ code without saying anything else. The airline’s own website no longer serves the airline, and IATA’s code search now lists PH on other carriers, so the record is now marked historical: PH is kept as a former code, and the answer says the code may since have been reassigned. ICAO code PAO, the call sign POLYNESIAN and the country (Samoa) stay, because the FAA’s list of ICAO designators still carries them. Every historical airline record now says so in its cited line on the API, the hosted MCP server and the airline pages, not only in the record’s status field.
An airline lookup by ICAO code NAB answered Mina Airline Company (Egypt). The FAA’s list of ICAO three-letter designators (Order JO 7340.2P, Change 3) gives NAB to Niger Air Cargo, and it has no entry for Mina Airline Company. NAB now answers Niger Air Cargo, on the API, the hosted MCP server and the airline pages, and the Mina Airline Company record no longer shows an ICAO code, because no source gives it one. No other airline record changed.
Some labels said something untrue about where our data comes from or what licence it carries. Airline answers named "IATA / ICAO airline code registries" as their source. IATA and ICAO assign airline codes, but they did not publish this dataset: FreightUtils compiled it from public sources, and verified records cite carrier pages and FAA Order JO 7340.2 field by field. The labels now say that. HS answers said "PDDL". That was the label of a third-party copy, not a licence the UN Statistics Division or the World Customs Organization gave, so they now say "no named open licence" and name the descriptions as UN Comtrade’s H6 wording (UN Statistics Division), not "official WCO" text. UN/LOCODE also said "PDDL"; UNECE’s own pages say CC BY 4.0, so every UN/LOCODE answer now carries "UN/LOCODE © UNECE, CC BY 4.0" with a link to the licence, and the UN/LOCODE page shows it beside the results. ADEME emission factors were labelled "Licence Ouverte"; ADEME’s own terms allow use in a tool like ours but not selling or licensing the data, so they now name those terms. The Terms page lists the sources we actually use. No values or answers changed, and no response field was added.
The npm package and the hosted MCP server now say the HS code search runs first against "the HS 2022 descriptions (UN Comtrade H6)" instead of "the official WCO HS descriptions", and the emissions calculator description calls its DEFRA, EPA and ADEME factors "published" instead of "open". The package README’s data sources match the corrected labels. Only descriptions changed; every argument and answer is the same.
Three tools use UK Government data published under the Open Government Licence v3.0: the UK duty calculator (live UK Trade Tariff rates from HMRC), the HS code search when it finds an everyday name such as "laptop" through HMRC’s Trade Tariff search references, and the emissions calculator when it uses a UK DESNZ/DEFRA conversion factor. The licence asks anyone who re-uses that data to say so, in its words: "Contains public sector information licensed under the Open Government Licence v3.0.", with a link to the licence. We named the licence but did not give that statement or the link. Now the duty, HS search and emissions pages show it next to the result, and the API and the hosted MCP server add it, with the link, to the citation qualifier of those answers. Answers built only on US EPA or French ADEME factors do not claim it, because those are not UK Government data. No numbers changed and no response field was added.
The npm package’s shipment summary now refuses a freight cost, insurance cost or item customs value below zero before it calls the API, with the same words the API and the hosted endpoint already use ("freight_cost must be 0 or more (GBP), or omitted for 0"), and its tool list advertises a minimum of 0 for all three. Zero and omitted still answer as before. The validate and airline lookup descriptions (npm package and hosted endpoint) now say what happens when several airline records share an AWB prefix: no single airline is named, and every holder is listed in airline_candidates with a note saying why. For 020 that is Lufthansa and Lufthansa Cargo. Answers did not change.
When a Pro key reaches its monthly allowance on the hosted MCP server, the limit message named a support address to write to about raising it. That address does not exist, so mail to it came back undelivered. The message now says contact@freightutils.com, the address on the contact page. The emails we send (your API key, sign-in links, password resets, the welcome emails and the ADR quick reference guide) came from a no-reply address, and the welcome email invites you to reply. A reply now goes to contact@freightutils.com too.
A Pro API key’s allowance renews at 00:00 UTC on the 1st of each month, and that is when the count starts again. The reset time sent with each answer said something later: X-RateLimit-Reset, Retry-After and the reset_at in a rate-limit answer counted 32 days from the 1st, so for October they named 2 November instead of 1 November. That is a day late in a 31-day month, two days late in a 30-day month and four days late in February. They now name the 1st of next month at 00:00 UTC, on the REST API and in the hosted MCP server’s limit message. The allowance itself, and the daily reset for free keys and calls without a key, did not change.
The HS code pages answered any code at any level. /hs/code/0901 showed heading 0901 on the subheading page, /hs/heading/090111 showed subheading 0901.11, and /hs/chapter/0901 showed the heading again, each claiming to be the original. So one code had up to three pages, and search engines saw duplicates. A code asked for at the wrong level now goes to its own page with a permanent redirect, so /hs/code/0901 opens /hs/heading/0901. A code that exists at no level gives a 404. Every code reached at its own level, and every link and sitemap entry on the site, is unchanged.
Some UN/LOCODE entries are not approved locations. BEAAB, Aalst in Belgium, carries the UNECE status RN: a request still under consideration. AUFIS, Fisherman Islands in Australia, carries UR: added on a user's request. The web API has said so since July, with a PROVISIONAL_STATUS warning, medium confidence and a citation that names the status. The unlocode_lookup tool on the hosted MCP server gave the same entry without any of that, so an agent could cite a provisional code as if it were approved. It now gives the same warning, confidence and citation as the web API, from the same code. Approved locations such as NLRTM, Rotterdam, answer exactly as before. Four new practitioner tests hold it.
Most API answers are cached at the edge, keyed on the URL. The response envelope can be asked for with ?envelope=1 or with the header Accept: application/vnd.freightutils.v1+json, and the header is not part of the URL. So whoever asked a URL first decided the shape every later caller got from the cache: a plain request for an Incoterms rule could come back as the envelope, and a request for the envelope could come back flat. Cacheable answers now carry Vary: Accept, so the cache keeps the two shapes apart. ?envelope=1 works as before, and the answers themselves have not changed.
Asked whether 0.1 L of hydrofluoric acid, UN 1790, is a limited quantity with no packing group given, the LQ/EQ check rightly gives no verdict: the two packing group I rows do not permit limited quantities at all, and the packing group II row allows 1 L. It says so with the code AMBIGUOUS_UN_VARIANT, but only in the structured envelope. The plain response, which the published API specification documents and a ChatGPT action reads, carried the candidate rows without the code, and so did the text the MCP server hands an agent. The code and its recovery hint are now in all of them, as they have been on the 1.1.3.6 calculator since 3 October. The npm package reads the same plain response, so its text carries the code too. Two new practitioner tests hold it.
Asked about a UN number with more than one ADR Table A row and nothing to say which, the 1.1.3.6 calculator and the LQ/EQ check give no verdict: they return AMBIGUOUS_UN_VARIANT with the candidate rows and a recovery hint. The hint's parameters named the first candidate's variant_index, so an agent that sent them as given got an answer on whichever row was listed first, not on the goods. UN 1263 paint was scored on its packing group I row, at fifty times the points of packing group III. UN 1950 aerosols were scored on the asphyxiant row, 100 points and exempt, although a corrosive or toxic aerosol is transport category 1 and the same 100 scores 5,000. The hint now repeats the call without a row, keeping a packing group only when the caller sent one, so sent unchanged it is asked again. It names the line it is about and the field that chooses the row: packing_group or variant_index, picked from the candidates by the goods' own packing group, concentration or description. Where the rows have no packing group, as for aerosols, it asks for variant_index alone. The web API, the hosted MCP server and the npm package, which reads the same answer, all behave this way. Eight new practitioner tests hold it.
The Pro plan's quota is 50,000 requests a month, but its REST responses said X-RateLimit-Window: 86400, a day. The hosted MCP endpoint already said 2592000, 30 days, for the same key. The quota itself was always counted by the month and X-RateLimit-Reset always pointed at the month's end; only the label was wrong. Pro REST responses now say 2592000 too. Anonymous and free callers still see 86400, because their quota is daily. The limits have not changed.
The shipment summary's UK duty estimate adds freight_cost and insurance_cost to the goods' customs value to get the CIF value the duty is charged on. A negative cost was subtracted instead: a 5,000 GBP coffee line with freight_cost -1000 came back with a CIF of 4,000 and a duty of 240 instead of 300. A negative customs_value on one line came off the total in the same way. Text such as "abc" or "NaN", and an infinite number, made every duty figure empty while the request still succeeded. Each of these is now refused with HTTP 400 on the API, and with a tool error on the hosted MCP endpoint, naming the field and saying that a cost must be a number of 0 or more. Zero is still a valid cost. A cost left out, sent as null or left blank, as Zapier, n8n and Make send an empty field, still means no cost. Valid costs give the same answers as before. The duty calculator has refused the same values since 7 October.
The status page used to ask UptimeRobot for fresh figures on every visit, so it took as long as UptimeRobot took to answer, sometimes more than five seconds. It now keeps the last answer for up to 60 seconds and refreshes it in the background, so a visit never waits on UptimeRobot. Under the status line it shows when the figures were read. If UptimeRobot cannot be reached and nothing has been read yet, the page says the status data is unavailable and why, and shows the site's own internal checks instead of figures it does not have.
Since 8 October, some endpoints refused an optional field that was sent empty. /api/cbm?l=120&w=80&h=100&pcs= answered "pcs must be a whole number" instead of the volume of one piece, and /api/emissions with sub_mode= and basis= answered "basis must be one of: wtw, ttw". Other endpoints already treated the same blank as not given. Zapier sends an optional field you leave empty that way, and so do HTML forms and other automation tools. An optional query parameter sent empty, or holding only spaces, is now read as not given, and its documented default applies: the answer is the same as leaving it out. This now holds on every GET endpoint. The same applies to the optional mode field of the LQ/EQ check, the consignment calculator and the shipment summary, and to an item's unit in the LQ/EQ check. Required fields sent empty are still refused by name. Values that are present but wrong, such as 2.5 pieces, 1e+21, "abc" or an unknown basis, are still refused. A parameter sent twice and a misspelt parameter name are still refused too.
Asking /api/ldm for quantity=10 (the API reads qty) used to answer for one pallet, as if the 10 had never been sent. A query parameter that is a near miss of a real one is now refused, naming the right one: another case or separator, a plural, or the MCP argument name. Other unknown parameters, such as cache-busters and tracking tags, still pass. The consignment and duty request bodies refuse any field they do not read. Every MCP tool, hosted and in freightutils-mcp 2.22.2, refuses an argument it does not read. Before, most dropped it silently, so ldm_calculator with qty: 10 answered for one pallet. A UN/LOCODE search for “Roterdam” now suggests Rotterdam (NLRTM). A code one character off, such as NLRTN, names the real codes beside it, on the lookup and on resolve.
The search box on the HS codes page now asks the same HMRC everyday-goods index as the API when no WCO description contains your words, so “laptop” shows 8471.30 there too, marked as an HMRC match to confirm. On the hosted MCP endpoint, the UN/LOCODE and HS lookups called with nothing to look up now refuse and name what to send, as the API does. They used to answer 20 arbitrary locations or the section list. The LDM calculator refuses a pallet preset sent with a different length or width, instead of quietly using the preset. A web address containing an encoded backslash now gets a plain “not found” instead of a server error, and the signed-out account check in the site's menu no longer logs an error in your browser console on every page.
The MCP tool descriptions (npm package and hosted endpoint) now say what today's tools do. The LDM calculator's says a load fits only when it fits the vehicle's length, its floor and, when you give a weight, its payload, and that a stored pallet count is used as floor space only when it is one. The shipment summary's says the same of its vehicle suggestion. The HS lookup's no longer says “laptop” and “computers” find nothing: they now find 8471.30 and heading 8471 through HMRC's everyday goods names, at medium confidence. The npm package's duty calculator also accepts a commodity code written with spaces, as the API already did.
Earlier today the LDM calculator and the shipment summary started reading the 7.5t rigid's stored pallet count, 8, as its floor space. That 8 is a payload figure: the record's own notes say the 6.1 × 2.4 m deck holds 15 Euro pallets. So ten 150 kg Euro pallets were answered “does not fit”, and the summary suggested an 18-tonne rigid. A vehicle's stored count is now used as its floor space only when the record says that is what it is. Otherwise the floor positions are worked out from the deck, and the payload is checked against the vehicle's maximum. Ten 150 kg pallets fit the 7.5t at 65.57 % of its length; ten 300 kg pallets do not (3,000 kg is over its 2,500 kg payload), and the answer says which limit failed. The response says where its floor-position figure came from, and when no weight is given it says the payload was not checked. Answers for the 13.6 m artic are unchanged.
A paged or summary airline list by country (/api/airlines?country=…&view=summary, which the npm MCP package uses) counted only the first 100 matches, so Germany’s 137 airlines read “total 100” and paging stopped there. Paged and summary lists now count and page the whole match, as the hosted MCP already did. The plain list is unchanged.
Searching the HS codes for “laptop” found nothing, because the WCO text calls them “portable automatic data-processing machines”. When no WCO description contains the words, the HS lookup (API and MCP) now asks the UK Trade Tariff’s search references — HMRC’s own index of everyday goods names — and returns the code they name: laptop → 8471.30. The answer says it came from HMRC’s index and asks you to confirm the code. Searches the WCO text already answers are unchanged.
The npm package's vehicle, ULD, container, Incoterms and airline lists now come back as summary pages with a total and the offset of the next page, and new full, offset and limit parameters page the full records. The whole ULD list was more than 600,000 characters, more than an assistant will read. The package accepts UN 1203, 8471.30 and GB LHR as the API already did; it refuses a negative value to convert; and its tool descriptions now say that totals are worked from unrounded figures, that gross weight is the total for all pieces, and that a shipment's customs value is per item. On the API, /api-docs documents the same paging (view=summary) on the five list endpoints, and its MCP install boxes no longer push the page sideways on a phone.
Every ICAO airport lookup through the freightutils-mcp npm package failed: the API answered the ICAO code, and an unknown code, without the response envelope the package asks for. It now does — no package update needed. A request to the hosted MCP endpoint whose body was not valid JSON waited about five minutes for a timeout; it now gets the standard JSON-RPC parse error straight away. Container lookups read the forms planners type — any letter case, the ISO size-type code (45G1) and short forms such as 40HC — and a typo is pointed at the closest container (20ft-standrd → 20ft-standard) rather than at a fixed 40ft high cube. Other not-found answers no longer suggest a fixed example that leads somewhere else (an unknown UN number on the exemption calculator was told to try petrol), and the hosted duty calculator takes a commodity code written with spaces, as the API does.
Listing vehicles or ULDs through the hosted MCP endpoint returned every record's full audit trail — 70,000 to 600,000 characters — and AI clients refused to read it. Lists from the vehicle, ULD, container, Incoterms and airline tools now return summary rows (the same field names, a subset of them) with the total, whether the page is cut, and the next offset; full: true returns the full records a page at a time, and looking up one record still returns everything, provenance included. The REST API keeps its full lists by default and gains view=summary, offset and limit.
Several hosted MCP tools refused a call with a bare text line such as “Error: Provide pallet preset or length_mm + width_mm”. Every refusal from the tools' own checks now comes back in the response envelope — ok false, a code, a message naming the field and how to fix it, and a recovery call — while still flagged as an error, so nothing that relied on the flag changes. The ICS2 check refuses an empty goods description instead of calling it clean, on the API and the MCP tool. The emissions calculator says when it answers on a different basis than the one you asked for (the US factors publish tank-to-wheel only): basis_used and a warning, instead of a silent switch. The hosted container lookup now computes the load fit from item dimensions, as the npm package and the API already did, instead of ignoring them.
A search that found nothing used to offer one fixed example: a typo of “acetone” was told to try UN 1203, which is petrol; “heathrw” was told to search “london”. The ADR, airport, vehicle and ULD lookups now suggest the closest names in their own data (ACETONE, UN 1090; London Heathrow, LHR; standard-curtainsider), and the airline search finds “Emirats” as Emirates. DAT, which Incoterms 2020 replaced, now names DPU with the ICC source. HS codes written with dots (8471.30), UN/LOCODEs written spaced (GB LHR), “UN 1203” on the limited-quantity and exemption checks, “LD3” for the AKE container, and an air waybill written “176 1234 5675” or without its hyphen are all read as what they are. Nearest-airport results leave out closed airfields unless you ask for them.
A 6- or 8-digit code such as 610910 was padded with zeros to 6109100000 — a code nobody sent — and answered “not found”. It is now refused with the declarable 10-digit codes beneath it, read from the same GOV.UK Trade Tariff (for 610910: 6109100010 T-shirts, 6109100090 Other). An origin country that is not an ISO country code is refused on every surface, and the duty citation states the rates as percentages and the amounts in GBP.
A class filter stopped at 100 rows (50 through the hosted MCP tool) and said nothing; class 8 alone has 368. Name searches and class filters now return total, truncated and next_offset, page with offset and limit, and cite themselves as a list rather than as the variants of the first UN number. Name and class lookups through the freightutils-mcp npm package failed outright since the response envelope was introduced; they now answer — no package update needed. The hosted validate tool refuses a value with no type, as the API does, instead of answering that nothing was found.
The CBM and chargeable-weight calculators rounded each piece before multiplying by the piece count, so many small pieces lost or gained a lot: 10,000 cartons of 5 × 5 × 6 cm read 1.0 m³ instead of 1.5 m³, and 300 kg chargeable instead of 250 kg. Totals are now computed from the unrounded figures and rounded once, for display — the same rule the consignment calculator, the shipment summary and the unit converter now follow. The unit converter uses the exact legal factors (1 ft = 0.3048 m, 1 lb = 0.45359237 kg), so 1 m³ is 35.314667 ft³ everywhere, and 9.6 m³ converts to 1,600 chargeable kg, not 1,600.03. It refuses a negative value, and a conversion between a mass and a volume now comes back as a refusal listing the units it accepts instead of a blank answer.
Ten euro pallets on the 7.5t rigid read “fits” beside “10 of 8 pallet spaces”. The loading-metre calculator now says a load does not fit when it needs more floor positions than the vehicle record holds, and says why. UK pallets are counted as UK floor positions (26 on the artic, as vehicle_lookup says), not as 33 euro equivalents, and 34 euro pallets on the artic are refused for the 34th position rather than for a length they do not exceed. The shipment summary no longer suggests a vehicle whose own record cannot take the load — ten euro pallets now get the 18t rigid, with utilisation measured against that vehicle — and suggests the smallest container that holds the cargo (9.6 m³ and 1,500 kg: a 20ft, not a 40ft High Cube).
The documentation says adr_quantity is the quantity for the dangerous-goods line; the summary multiplied it by the number of pieces, so 200 L of petrol on ten pallets scored 6,000 points and “not exempt” where the answer is 600 points, exempt — the figure the ADR exemption calculator gives. It is now read as the line total, and the response says so. customs_value stays per item and weight stays per piece, now written plainly in the schema and echoed in the response with the total used; the chargeable-weight calculator says that its gross weight is the total for all pieces. Pallet fitting adds volume utilisation beside the footprint figure it always reported, says when a weight limit rounds the count down to whole layers, and says why when no box fits. Consignments state the factor behind a road or sea chargeable weight, flag an air waybill on a sea consignment, and a multimodal summary says which mode its headline figure is.
Choosing a custom vehicle and giving its length changed nothing: the API and the /ldm page answered for the 13.6 m artic instead. Ten euro pallets on a van with a 3 m load floor — 4.0 loading metres — read “fits” at 29% of the artic. They are now measured against the length you give: 133% of 3 m, does not fit, with the over-length warning. The hosted MCP tool was already right; it now refuses a custom vehicle with no length, as the API does, and the API refuses a vehicle it does not know instead of answering for a vehicle with no length at all.
Some API inputs were turned into a different question. A count written 1e+21 — what a JavaScript client sends for 1e21 — was read as 1, so the CBM, chargeable-weight, loading-metre and container-fit calculators answered for a single piece, and a volumetric divisor of 1e21 became 1 (120 × 80 × 100 cm read as 960,000 kg); 2.5 pieces read as 2. The excepted-quantity check answered “qualifies” when the inner packagings per outer packaging were -20, 0, 16.5 or “abc”. The UK duty calculator answered a customs value of “abc” with every figure blank, and took a negative freight or insurance cost off the value duty is charged on. Container fits took -3 items and -500 kg; the UN/LOCODE search with limit=abc returned every match; and chargeable weight with mode=SEA was computed as air. Each is now refused with HTTP 400 naming the field and what arrived — sea mode is read in any letter case — and the MCP tools refuse the same values. A container type that is not found now suggests 40ft-high-cube, a real type, where it suggested 40hc. Every well-formed request is answered exactly as before.
The npm MCP server refuses a count above 9,007,199,254,740,991 before it is sent — such a count travelled as “1e+21” and was read as 1 — and a negative freight or insurance cost on the UK duty calculator. adr_lookup accepts “UN 1203”, the form a transport document prints and the one its description already promised, and its hazard class examples all answer: explosives are class 1, the division being part of the classification code (e.g. 1.4S). The hosted endpoint at /api/mcp makes the same changes. Earlier versions of the package now get a refusal from the API instead of a wrong answer.
An audit of the packages the site runs on found two critical and eleven high advisories, all in libraries we use rather than in our own code. We updated them: Next.js to 16.3.8 (a critical advisory in the Node.js version of next/og’s image generator — our social-card images use the Edge version, which the advisory does not affect), the MCP SDK behind the hosted endpoint to 1.32.1, and sharp, nanoid, the Sentry SDK, Resend and the libraries beneath them. A check now fails on any high or critical advisory in what the site runs on — on every dependency change and every day — and Dependabot opens security updates. Vulnerability reports go to contact@freightutils.com, or privately through GitHub on the freightutils-mcp repository.
Since 3 August, every ULD lookup made through the freightutils-mcp npm package returned an error: the ULD endpoint refused legacy_source, a response-format parameter the package sends with every call. The hosted MCP endpoint and the REST API answered normally. The endpoint now accepts it, as every other endpoint does — no package update needed.
A code such as 0901AB was padded, looked up at HMRC and answered “not found”, with a hint suggesting codes like 0901AB0010 that cannot exist. A commodity code is 6 to 10 digits; anything else is now refused with that rule. The code also forms part of the address we request from HMRC, so nothing but digits reaches it.
The npm MCP server now reports a failed call by kind — network failure, timeout (30 s), the API’s own refusal, a rate limit with its reset time, or a server error without the error page — and stops a stalled call instead of waiting for ever. Its schemas carry the limits the API already enforces (quantities up to 1,000,000,000, 1–50 shipment lines, weights from 0, consignment line limits), so an out-of-range value is refused before it is sent, with the limit named; the hosted endpoint at /api/mcp now refuses the same values. With no API key set, it says once at start-up that calls use the anonymous tier. Underneath: MCP SDK 1.32.1, a security policy, Dependabot and code scanning on its GitHub repository.
We checked every statement in the privacy policy against the code and our providers’ own terms, and corrected the ones that said less than the service does. It now says that our hosting provider’s request logs, kept for 30 days, record each request’s address — including any URL parameters, such as a calculator input — beside your IP address, so an API key should go in the X-API-Key header rather than the URL; that every sign-up joins our contact list at Resend and a free key taken on /pricing or /api-docs brings two follow-up emails; that feedback and the newsletter exist and what they collect; and that email to our @freightutils.com addresses is forwarded by Cloudflare to a Gmail mailbox. Two sections are new: the lawful basis for each kind of processing, and the international transfers with the safeguard each provider relies on. Some of our own log lines wrote the email address of a new sign-up, subscriber or Pro customer in clear; they now write a one-way hash, and a new check fails the build if a clear address comes back.
Opening the ADR lookup or the ADR 1.1.3.6 calculator downloaded the whole of ADR Table A to your browser — a 1.4 MB block of JavaScript — although the page only needed four numbers and the six-line excepted-quantity table for its threshold panel. The panel now loads just those, so the scripts the two pages load drop from about 1.5 MB to about 160 KB before compression. Every answer on the pages is unchanged. A new check fails the build if a dataset file is pulled into a browser bundle again.
The ADR 1.1.3.6 small-load calculator, the limited and excepted quantity check and the shipment summary read only the field names they document, and dropped every other one without a word — on a line, beside the list of lines and in the query string, on the API and on both MCP servers. A declaration sent under any other name changed nothing. 300 of petrol with quantity_basis “gross” — the very name the answer uses — read exempt at 900 points, where basis “gross” withholds the points because ADR 1.1.3.6.3 counts the goods, not their packaging. UN_NUMBER “1051” sent beside the list vanished, and a load carrying hydrogen cyanide, which can never use the small-load exemption, read exempt. An excepted-quantity check sent with inner_packagings: 20 checked one inner packaging per outer and read “qualifies”, where 20 × 30 ml is over the 500 ml E2 allows per outer packaging. A shipment line with un: “1051” was not treated as dangerous goods at all. Now any field a tool does not read is refused — HTTP 400 on the API, a tool error on MCP — naming the field, suggesting the one it most likely meant and listing the accepted ones; one field sent under two names on the shipment summary is refused the same way. Requests that use the documented names are answered exactly as before.
freightutils-mcp, the npm package that runs the FreightUtils tools inside an agent’s own MCP client, now refuses a field it does not read on the three tools whose answers settle a dangerous-goods question: the ADR 1.1.3.6 small-load calculator, the limited and excepted quantity check and the shipment summary. Those tools declared their inputs strict, but the server registered them in a way that quietly stripped any other field before the tool ran — while telling clients that no other field was allowed. A field sent under the wrong name, such as quantity_basis for basis, un for un_number or inner_packagings for inner_packaging_qty, now comes back as a tool error naming it and suggesting the right one, and the call never reaches the API. The hosted endpoint at freightutils.com refuses the same fields with the same words, and its tool descriptions, which come from this package, now say so. Every correctly named call is answered as before.
FreightUtils can now be added to Claude — claude.ai, the desktop and mobile apps and Cowork — straight from Claude’s connector directory, as well as by pasting the server URL as a custom connector. It is the same server with the same read-only tools either way. The MCP page, the API docs and llms.txt now link the directory listing.
A password you set from today — when you sign up, on the account security page, or through a reset link — needs at least 15 characters, and it is refused if it appears in a list of passwords exposed in data breaches, with the reason given. There are no rules about symbols, digits or capitals and every character is allowed: three random words works well. Passwords you already use keep working, and nobody has to change one. The breach check sends only the first five characters of the password’s SHA-1 hash to the Pwned Passwords service run by Have I Been Pwned, which answers with every breached hash that starts with them, padded with decoys, so the password itself never leaves our servers; if the service cannot be reached, a list of breached passwords held on our own servers is used instead. This follows NIST’s 2025 password guidance, SP 800-63B-4. The privacy policy now says so, and it also gives the real retention of request logs (30 days), error reports (30 days), sign-in and reset links (30 minutes) and sessions (up to 30 days after your last visit, 90 days at most).
Dry ice (UN 1845) is listed “NOT SUBJECT TO ADR except for 5.5.3” in ADR Table A, and 5.5.3.1.1 says no other provisions of ADR apply to it — so there is no 1.1.3.6 small-load calculation for it to be exempt from. The shipment summary already said so when only the gross weight was given, but once a line declared its ADR quantity it answered “exemptionApplicable: true” with a total of 0 points, beside its own warning that there was no exemption to apply. A load of dry ice alone now gets no points total and no exemption answer, with a note saying why; the line stays out of scope with the 5.5.3 conditions. Where dry ice travels with goods that are counted — say 200 litres of petrol, 600 points against 1,000 — the exemption still applies to the load, and the dry-ice line now reads “excluded from the 1.1.3.6 calculation”, the same note the exemption calculator gives, instead of “there is no 1.1.3.6 exemption to apply”. Fixed in one place for the REST API and the hosted MCP server; the npm package forwards the API, so it gives the same answer without a new release. Four new practitioner tests, REST and MCP, failed on production before the fix and pass after.
Before an AI agent calls a FreightUtils tool, it reads the tool’s description — and every description except the pricing tool’s own carried an instruction: “back off and retry, or call get_subscribe_link for higher limits”. A rate limit is not a reason to send someone to a pricing page, and a tool description is no place to steer an agent towards another tool or a paid plan. Version 2.20.1 of the npm package states the fact instead — a limit error carries reset_at, the UTC time the allowance resets — and the hosted endpoint at freightutils.com serves the same text. The pricing tool, get_subscribe_link, is now a plain lookup for a user who asks about plans, pricing or API limits: it no longer tells an agent to call it after another tool’s error, and its title is “FreightUtils Plans & Pricing”. The identifier resolver now says what each candidate carries instead of telling the agent which tool to call first. No tool, input, output or answer changed. Two new practitioner tests read the definitions the hosted endpoint serves and fail if a cross-tool instruction or an upsell comes back.
FreightUtils can be added to Claude as a connector, and Claude’s apps — claude.ai, the desktop and mobile apps and Cowork — call it from Anthropic’s network, not from the user’s own address. The server counted anonymous use per address, so every Claude user drew on the same 25 tool calls a day, and the first few conversations of a day would have used them up for everyone else. Calls that arrive from Anthropic’s published address range now share one allowance sized for that: 5,000 tool calls a day across all Claude users, resetting at 00:00 UTC, and 1,200 connection requests a minute. Connecting and listing the tools never counts against it — including the discovery request Claude’s clients send first on every connection, which the server does not support and had been counting as a tool call against whichever allowance the caller used. A call that brings its own API key still uses that key’s plan, and the range is read from the address the hosting platform records for the connection, so a header a caller writes cannot claim the shared allowance. When the shared allowance runs out, Claude is told plainly that it resets at 00:00 UTC, and the operator is warned once a day when it passes 80 %. The MCP page has a new “Use in Claude” section — how to add the connector, the tools by family, the limits and what a tool call carries — and the privacy policy now covers calls to the MCP server and calls made through Claude, and says plainly how long an IP address is kept. Claude Code connects from your own machine and keeps its own address’s allowance, and the REST API’s limits are unchanged.
Before an agent calls the ADR 1.1.3.6 small-load calculator, it reads the tool’s description — and that description said “a load totalling 1,000 points or less qualifies”. That is not the rule. Under ADR 1.1.3.6.2, dangerous goods of one transport category carried in packages qualify while their total on the transport unit stays within that category’s maximum — 333 kg or litres for category 2 — and the 1,000-point sum is the test only when the goods belong to different categories: 200 litres plus 133.2 litres of category-2 liquids makes 999.6 points and is still not exempt. The calculator’s answers have held each category to its maximum since 25 September; version 2.20.0 of the npm package now describes the tool the same way, and the hosted endpoint at freightutils.com serves the same text. The description also documents today’s other changes: every line of the answer names its route, an ID 8000 line is read rather than refused, and a field sent beside the list of lines is refused instead of ignored — the package now passes those fields on to the API so they are refused there too, where it used to drop them. The tool list and the tool names are unchanged. A new practitioner test reads the description the hosted endpoint serves and fails if the bare rule comes back.
The ADR 1.1.3.6 small-load calculator takes a load as a list of lines, each with its own UN number, quantity and — where it matters — packing group, variant index, unit and basis. The same fields at the top level of a call are the single-substance form. Sent at the top level beside a list of lines, they were ignored without a word, on the web API, the hosted MCP server and the npm package alike, and three wrong answers came out of that. Asked about 100 L of UN 1263 paint with packing group III sent beside the list, the calculator never saw the packing group and listed every UN 1263 row as a candidate — the packing group I row among them, transport category 1, which scores the same paint at fifty times the points — and its own recovery hint pointed at that row first. 300 of petrol, UN 1203, with the unit “kg” sent beside the list read “1.1.3.6 exemption applies” at 900 points: the declaration that should have stopped the calculation never reached it, so a figure in kilograms was counted as if it were litres, the dimension 1.1.3.6.3 counts a liquid in. And a whole substance went the same way: a UN number and quantity sent beside the list — documented until now as “the list takes precedence” — simply vanished, so a transport category 0 substance, which no load carrying it can ever exempt, left the rest of the load reading “exempt”. Any field of the single-substance form sent beside the list is now refused, with an error that names it and says where it goes — by the web API and the hosted MCP server now, and by the npm package from version 2.20.0. It is never dropped, and never applied to a line by guess: with several lines there is no way to know which one it was meant for. The calculator’s own filter was right throughout: packing group III on the line answers on the three packing group III rows, all transport category 3. Separately, when every line of a call resolves to more than one Table A row, the answer gives no verdict and says why with a code, AMBIGUOUS_UN_VARIANT. The code was in the structured envelope only; the plain response — the one the published API specification documents and a ChatGPT action reads — and the text the MCP server hands an agent carried the candidate rows without it. It is in all of them now. Ten new practitioner tests on the web API and the hosted MCP server hold these answers.
Each line the ADR 1.1.3.6 calculator returns now carries a state naming how it was routed, in the words the rebuilt document check uses: COUNTED for a line evaluated on its Table A row, NOT_SUBJECT_TO_ADR for an entry Table A lists as not subject to ADR — dry ice, UN 1845, among them — BLOCKED for a line no points can come from as sent, such as a substance with several rows and no packing group, and AIR_ONLY_ID. That last one is new. An air waybill or a NOTOC can list an ID-prefixed number — ID 8000, consumer commodity, is the common one — and the calculator refused it as an invalid UN number, so a planner checking the road leg to the airport had to delete the line to get an answer at all. ADR Table A has no such entry, so an ID number gives no transport category, and 1.1.3.6 adds quantities up by transport category. It is now accepted, routed AIR_ONLY_ID and left out of the points, and every such line carries a note stating the condition that rests on: nothing is counted when the packages are limited quantities to the ICAO Technical Instructions — ADR 3.4.9 and 3.4.10 deem such packages to meet its limited-quantity provisions 3.4.1 to 3.4.4, and 1.1.3.6.5 leaves goods exempted that way out of the count — while goods that are not in such packages must be entered under their own UN numbers. Only the ID prefix makes a line air-only: an ID number is never read as the UN number with the same digits, and a load of nothing but ID lines scores nothing with its own message — never “1.1.3.6 exemption applies”, which would claim a calculation ran over an ADR entry, and never “not subject to ADR”, which is a Table A remark an ID number does not have. The shipment summary, which passes its lines to the same calculation, never reads an ID number as a UN number either: a summary with an ID line leaves the points uncalculated and names the calculator — on production an ID 1203 line had been scored as 500 litres of petrol. Ten more practitioner tests hold these answers on the web API and the hosted MCP server.
September 2026
The DG Document Check now shows a coming-soon notice to every visitor, signed in or not. The old document reader is switched off for everyone, including internal testing, and its two services answer "paused" to any caller — a browser, an account or an agent — before reading anything. On live tests of a clean scan it read UN numbers and quantities that are not on the page, plausible enough to pass every check, and a reference tool must not offer that to anyone. The rebuilt check — drop in a shipper’s declaration or DG note and get a straight answer on whether the road leg needs an ADR-certificated driver, with the working shown — is being tested against real declarations and will launch once it is proven. The ADR 1.1.3.6 exemption calculator, the LQ/EQ checker and the ADR lookup are unchanged.
The document reader on the DG Document Check is switched off for everyone except internal testing until its replacement is ready. On a clean scan it could read a UN number that is not on the page but is a real ADR Table A entry — plausible enough to pass every check before the lines were confirmed — and a reference tool must not offer a draft like that. Entering the dangerous-goods lines by hand works exactly as before and runs the same check: limited quantities, and the 1.1.3.6 small-load points with the threshold, headroom and citations. The calculators and the UN number lookup are unchanged. The page says the reader is being rebuilt, and an upload sent to it anyway is refused before anything is read, so it costs no quota.
Four ways the ADR 1.1.3.6 small-load answer could read “exemption applies” when it should not have, all fixed on every surface that shares the calculation. First, the REST endpoint accepted any value as a quantity and turned what it could not read into zero: a negative figure, a word, or a figure with its unit typed into the same field — “300 kg” — came back as 0 points and “1.1.3.6 exemption applies”, and a negative line in a multi-substance load was subtracted from the others. That is the shape a spreadsheet value takes when it is mapped through an automation platform, and the Zapier, Make and n8n integrations all call this endpoint. A quantity must now be a number greater than zero (up to the same thousand-million ceiling the document check already used); anything else is refused with an error that names the value received and says what to send, and one bad line refuses the whole request rather than answering over the lines that happened to be valid. Unit and basis accept only the documented values — litres or kilograms, net or gross; an unrecognised unit such as “kgs” was previously dropped and the figure scored as if it were already on the right basis. A parameter sent twice is refused instead of silently answering for its first value. The shipment summary endpoint applies the same rule to its piece count and its declared ADR quantity. Second, the variant index — which picks one Table A row of a substance that has several — was read the same careless way: a blank, false or “0x0” all became row 0. For aerosols, UN 1950, row 0 is transport category 3 and row 4 is category 1, so 500 units read “exempt” instead of full ADR. A variant index must now be a whole number or be left out; left out, a substance with several rows still returns its candidates and no verdict. A UN number is now read as written rather than stripped of everything that is not a digit — “1203abc” and “-1203” used to answer as UN 1203 — and the limited-quantity checker refuses the same malformed inputs (a boolean or an array as a quantity, an array as a packing group) that it used to read as numbers. Third, the column (3) maximum was compared with each line separately, so the same goods written on two lines escaped it: 200 L and 133.2 L of petrol — 333.2 L of category 2 against a 333 L maximum — read exempt at 999.6 points while the same 333.2 L on one line did not. A substance split across lines is now summed. And where everything on the vehicle is in one transport category, ADR 1.1.3.6.3 makes column (3) the maximum TOTAL per transport unit, so 200 L of petrol and 133.2 L of ethanol, both category 2, are also over it; that answer now says so in its own words rather than as a per-substance limit. Only category 2 has such a window (its maximum, 333, is 1,000 points rounded down), a total of exactly the maximum is still within it, and both checks only fire where they change the answer — a load already over 1,000 points reads exactly as before. Fourth, the web calculator page kept its own copy of the verdict logic and had drifted from the API: an entry whose Table A row carries no transport category — fifteen rows: the kit entry UN 3316, whose column (15) reads “See SP 671”, twelve rows with a dash, and UN 2071 and UN 3363, where the column is blank — was scored as zero points, so 500 kg of first-aid kits read “Exempt”. The API has returned no verdict for such a row since 11 September; the page did not follow. It now runs the same calculation as the API rather than a copy of it, shows “No verdict” with the reason, and still reports full ADR where the other lines already settle the load, because a line with no points can only ever add to a total. Eighteen new practitioner tests hold these answers in the request shapes the integrations send, and a check runs every Table A row through both the page’s path and the API’s path and requires the same answer. The public calculator and ADR pages now state the same-category rule beside the 1,000-point one.
An internal change to how a verdict is modelled on an unreleased surface. Asked “is this ADR?”, a UK transport planner means one thing: does the driver need an ADR vocational training certificate, or is ADR 1.3 awareness training enough. Nothing in the codebase answered that — the ADR licence existed only as prose on the public ADR pages, and the internal document check led with a presence check on the dangerous-goods table, which tells a reader something they already knew from their own paperwork. There is now a three-state answer — awareness training is sufficient, a vocational certificate is required, or it cannot be determined — derived entirely from the existing small-load exemption engine rather than from any new reading of the regulation. It re-derives no points, no transport categories, no thresholds and no scope; it names the consequence of what that engine already decided, which is what keeps it from ever contradicting the public calculator. Two details are worth stating because they are where this kind of rule usually goes wrong. First, the answer is keyed to the verdict for the whole load, not to a scan of individual lines: where one line already disqualifies a load — an entry in transport category 0, a per-substance maximum already exceeded, a partial total already past the threshold — no unresolved second line can undo it, because points only add, so the answer is “certificate required” and not “cannot determine”. Second, where the answer genuinely cannot be given, nothing is guessed: the response names the missing input and where to obtain it, and for an entry whose table row carries no transport category it lists the special provisions that govern it without asserting what they say. Where the exemption does relieve a load, the duties that survive it are listed alongside, so “no certificate” never reads as “nothing to do”. The public calculator, the REST endpoint, the hosted tool listing and the published API specification are unchanged by this work.
The 1.1.3.6 small-load exemption calculator applies two separate checks to a line: whether the quantity was given in the dimension the paragraph counts for that entry, and whether the entry’s Table A row carries a transport category at all. The unit check was running first and returning immediately, so a row with no category answered only the unit complaint. Asking about an engine consignment — UN 3528, whose column (15) is a dash — with a quantity in kilograms came back “re-send the quantity in litres” and nothing else, because the row is a flammable-liquid entry and litres is what the paragraph would count. Following that advice would have changed nothing: the row carries no transport category, so there is no multiplier and no quantity in any unit can produce points for it. The category check now runs first on every surface that shares the calculation. Both findings are still reported — the unit really was the wrong dimension, and that flag is part of the published contract — but the category finding leads the message and the first warning, and the unit note now says in terms that re-sending the figure will not change the answer. A row that does have a category is unaffected: a mass sent against a litres row still asks for litres, as before. The same reordering applies to the kit entry UN 3316, whose column (15) reads “See SP 671” and which is counted by mass, so there the mismatching unit is litres rather than kilograms. The web calculator page now renders the same sentence the API returns for such a row instead of its own shorter paraphrase, which named neither the column nor the special provisions that actually govern the entry. Five new tests hold the ordering on the REST API and the hosted MCP server.
The air tools have always been separate pages: the airline and air-waybill-prefix reference, the airport lookup, the unit load device specifications, the chargeable-weight calculator. A planner working one waybill was opening four of them and carrying numbers between tabs. The new air desk puts that work in one order on one page, and it introduces nothing — each step runs the same engine as the tool it belongs to and shows the same citation, so the desk and the API cannot answer differently. Paste an air waybill and it gives the check-digit result and the issuing carrier. Where the prefix is held by more than one record — a group carrier and its regionals, or a passenger airline and its cargo arm — it names every holder, verified records first and ceased carriers last, rather than the one that happens to sit first in the file; the page works that rule through in prose with the Qantas prefix 081 as the example, because it is the rule most likely to bite. Look up an airport by code, name or city, or find the nearest airports to a point when the shipper has given you an address and not an airport. Pick a unit load device and read its external size, tare, maximum gross and usable volume, each shown with the basis it was published on — a maximum gross figure is sometimes a structural rating and sometimes one carrier’s operational limit for one aircraft, and the qualifier travels with the number. Then work the chargeable weight on the divisor your contract applies, with the gross weight checked against the selected unit’s published payload. The last step is dangerous goods, and it is deliberately the smallest. Almost every pharma and perishables booking carries dry ice, so the desk shows the published UN 1845 entry — outside ADR scope, with the section 5.5.3 carriage conditions quoted verbatim, no packing group and no transport category, and therefore no small-load exemption to apply. It then says plainly that this is the road answer and not an air clearance: carriage by air is governed by the IATA Dangerous Goods Regulations, a separate regime whose text FreightUtils does not publish and does not paraphrase. On a page about air freight a correct road verdict left unlabelled would be read as permission, so the scope statement sits beside the verdict rather than under it. No booking, no tracking, no rates, no document reading. Three new tests hold the page: the dry-ice scope answer, the shared-prefix worked example, and the statement that a valid check digit is a structural property of eleven digits and never confirmation that a waybill was issued.
Three faults in the dangerous-goods verdicts, found by driving the golden set from the last four weeks through the signed-in document check on a preview. First, on every surface: a handful of ADR Table A rows carry no transport category in column (15) — the kit entry UN 3316 points to special provision 671, which sets the category from the packing group assigned to the kit as a whole, and twelve rows such as the vehicle entry UN 3166 carry a dash. The exemption calculator had no multiplier for them and treated the missing figure as nothing to add, so a load of first-aid kits came back as 0 transport points against the 1,000 threshold and "exemption applies" — on the web API, the hosted MCP server and the document check alike. Those rows now withhold: no points, no total, no verdict, and a warning naming the column value and the special provision that governs the entry. The limited-quantity checker had the same class of fault on the same kit entry, whose column (7a) reads "See SP 251": it reported limited quantities as not permitted, which is the reading of a zero; it now reports the check as inconclusive and names the provision. Second, on the document check only: the confirmed unit never reached the engine, so 300 kg of petrol — a litres row — was multiplied as if it were litres and cited as exempt with 900 of 1,000 points, the very verdict the web API has refused since 4 September; five kilograms of compressed hydrogen, counted by the water capacity of the receptacle in litres, scored fifteen points the same way. The check now refuses a figure in a dimension the paragraph does not count for that row — on the card before you confirm, and again on the server — and asks for the unit it needs. Third, the limited-quantity headline on the document check said a per-package figure was "over the limit" when the figure and the limit were in different dimensions, or when column (7a) defers to a special provision: the per-line note already said the comparison could not be made, but the headline and the screen-reader announcement contradicted it. The headline now carries that note, such lines are counted in their own bucket rather than as over the limit, and a figure printed beside the limit it is compared with no longer rounds onto it. A quantity large enough to overflow the arithmetic is refused rather than rendered as an infinite total. Five new tests on the REST API and the hosted MCP server hold the engine changes; the page changes are held by the offline gate, because the page is session-gated and still unlisted.
The airline dataset holds 6,357 records, and many identifiers sit on more than one of them: 399 IATA designators, 324 ICAO designators and 61 air-waybill prefixes are each carried by two or more records. Some are groups flying under one code — prefix 081 belongs to Qantas and to the three regional carriers that operate as QF — some are a passenger carrier and its cargo arm, such as Lufthansa and Lufthansa Cargo on LH and 020, and some pair a live carrier with one that ceased operating years ago. The lookup already returned every holder in its results. The one-line citation did not: it was built from the first row in the file, so it read "Eastern Australia Airlines — AWB prefix 081." for a prefix the whole Qantas group holds and "Lufthansa — IATA LH." for a designator two carriers share. The identifier validator did the same one step further, naming that first row as the airline of any air waybill it checked — on prefix 976 that was a carrier that no longer exists. A citation is a claim about the entry it names, so naming one of several holders is a false claim for every waybill the others issued. The citation for a shared code now names every holder, verified records first and ceased carriers last, and says the code alone does not identify one carrier; a code held by one record is cited exactly as before. The validator withholds the single airline on a shared prefix and lists the candidate carriers in the same order, with a note saying why; a prefix held by one record still names it. The web pages for a code already listed every holder and now sort them the same way. Eleven new tests hold this on the REST API, the hosted MCP server and the page.
The ADR lookup serves 2,939 Table A rows. Fifty-six of them were re-checked field by field against the purchased UNECE ADR 2025 Table A in late August and carried an audit date, cited sources and a verified flag; the other 2,883 carried nothing, and nothing is easy to misread. An agent looking at the dataset-level provenance on the response envelope — which says the dataset is verified, meaning the transcription from the authority is complete and the anchor rows were re-checked — had no per-row signal that a particular row had never been individually audited. Silence was doing the work of a claim. Every row now carries the answer explicitly: a verified flag that is true only where a row-level audit exists, and a verification block that says which it is and why. Audited rows keep their state and now show the audit date and how many sources they were checked against; every other row says verified: false with the reason "no row-level audit", stating that it was transcribed from the same Table A and is held by the dataset-level checks only. Nothing about the rows themselves changed and no row was re-sourced; the dataset-level status on the envelope and the homepage table is unchanged, because it was always a claim about the dataset rather than about each row. The identifier resolver was brought into line in the same change: its candidate for a UN number had kept reporting the dataset-level status, so following its link into the lookup could read “verified” and then “unverified” for one row; it now reports the row’s own state. Six new tests hold this on the REST API and the hosted MCP server.
The two dangerous-goods verdict tools changed how they answer on 5 September: a packing group whose rows agree settles the answer, and one unanswerable line no longer silences the rest of a batch. The answers changed on the web API and the hosted MCP endpoint that day, but the sentences an agent reads to decide how to call a tool still described the older rule, because those sentences come from the installed package, and the package had not been republished. Version 2.19.0 carries the corrected descriptions: what a supplied packing group is enough for, what happens to a mixed batch, that a load the resolved lines already disqualify is reported as not exempt rather than as waiting on a packing group, and that the candidate rows the exemption calculator returns now carry the dimension the regulation counts them in. Nothing changed in the answers themselves with this release; the tool list and the tool names are the same. The hosted endpoint at freightutils.com serves the same description text as the package, so an agent connecting either way reads the same contract.
Two related faults in the dangerous-goods verdict tools, both found while pushing on the week-1 change that stopped a multi-row UN number from being answered as if it had one row. First: some UN numbers carry several ADR Table A rows within a single packing group, separated only by concentration band or shipping name. Hydrofluoric acid, UN 1790, is packing group I in two rows — more than 85 % HF, and more than 60 % but not more than 85 % — and both rows say the same thing to both tools: no limited quantities, excepted quantity E0, transport category 1. Told the packing group, the limited-quantity checker and the exemption calculator still refused to answer, listed all three rows of the UN as candidates including the packing group II row, and asked for a variant index that could not have changed a single word of the verdict. They now answer when the rows a packing group leaves agree on every field the verdict is computed from, and say so on the item: which rows the verdict holds for, why the concentration band does not matter here, and which row’s fields are displayed. Where the rows genuinely disagree — maleic anhydride, UN 2215, is packing group III in two rows, one of them molten and transport category 0 with no exemption possible, the other a solid in category 3 — the tools still withhold, and the candidate list is now only that packing group’s rows. A UN number with no packing group at all is unchanged: still withheld. Second: one unpinned line used to blank a whole batch. Petrol at half a litre per inner packaging and hydrochloric acid with no packing group on the same check came back with no answer for the petrol, because the acid line made the whole call ambiguous. Each line’s answer depends only on its own Table A row, so the petrol is now answered and only the acid line is withheld, with its two candidate rows attached to it by line number. The batch can never read as qualifying while a line is withheld. On the exemption calculator the same rule holds for points: the petrol line shows its 600 points, the acid line shows none, and the load total and verdict stay empty, because 1.1.3.6.4 sums every line and one term is unknown — the calculator states the partial sum from the lines it could resolve. Where the lines it could resolve already settle the load on their own — an entry Table A marks carriage prohibited, a transport category 0 entry, a per-substance maximum already exceeded, or a partial sum already over the threshold — it says the load is not exempt and why, rather than suggesting a packing group would change that; only the total waits. A call in which every line is ambiguous keeps the shape it had. Thirteen new tests, on the REST API and the hosted MCP server, hold both faults closed; the identifier resolver was checked and needed no change, because it takes no packing group and already lists every row. One small repair on the limited-quantity checker page rode along: a check whose result was inconclusive — a mass entered against a volume limit, or the reverse — failed to render at all since the start of the month; it now shows the inconclusive state and the unit to re-enter in.
The optional fields added to the dangerous-goods tools earlier today — the unit and basis on the 1.1.3.6 exemption calculator, and the ADR quantity per line on the shipment summary — were reachable over the web API and the hosted MCP endpoint but not from the package you install locally, because that package declares its own inputs and had not been updated. Version 2.18.0 declares them, so an agent running the server over stdio can now send them and get the same answers. The limited-quantity checker’s new inconclusive result is described there too. Two faults were caught before publishing, by running the built package against the live site rather than trusting that a declared field works: both new fields were accepted by the package and then quietly dropped on the way out, so a quantity declared in kilograms against an entry the regulation counts in litres still came back with a points total and an exemption. A field you can fill in that goes nowhere is worse than no field at all, because the answer looks considered. Nothing else changed for existing callers: the tool list, the tool names and the answers for every call that does not use the new fields are the same, and a single-row UN number returns a citation identical to the previous release’s, byte for byte. The install diagnostic in the package README also stopped printing a frozen version number and tool count as though they were current; it now shows placeholders and reports the real figures when you run it.
Column (7a) of the dangerous goods table states the limited-quantity limit in one dimension — a mass for some entries, a volume for others. UN 1340 is a solid and its limit is 500 g. Asked whether 2 litres of it was within that limit, the checker said yes. The same quantity entered as 2000 millilitres said no. Neither answer was reasoned: the engine had no density to convert a volume into a mass, so it compared the bare numerals — 2 against 500 passes, 2000 against 500 fails — and returned a verdict at full confidence either way. The direction of the failure is the worst one available, because a false pass on a limited-quantity check is what sends a load out under a relief it does not have. The same fault ran in the other direction too: half a kilogram of petrol against a 1 litre limit passed, and 500 grams of it failed. The check now refuses to answer across dimensions. Where the limit is a mass and the figure is a volume, or the reverse, the item comes back as inconclusive, naming the dimension the table uses and the unit to re-enter it in, and the batch it belongs to can never read as qualifying. Same-dimension comparisons are untouched and still answer: 0.2 kg against a 500 g limit is within it, as it always was. Entries whose limit is written with both a volume and a mass ("1 L or 1 kg") still pick the branch that matches what you entered, and the limit reported back to you is now the branch actually applied rather than whichever was written first.
The 1.1.3.6 small-load calculation counts a specific quantity per transport category: litres for liquids, net mass for solids and for liquefied, refrigerated or dissolved gases, receptacle water capacity for compressed gases, and for articles their mass without packagings. The calculator took a bare number and multiplied it, whatever it was. Three hundred kilograms of drummed petrol scored 900 points and came back exempt; on the basis the paragraph counts, that drum load is nearer 405 litres, 1,215 points, and not exempt — the wrong side of the threshold, stated with full confidence. Until now the rule shipped on that endpoint as help text next to the field. You can now declare what your figure is: an optional unit, and an optional basis of net or gross. Declare nothing and the behaviour is exactly as it was, because the field has always been documented as taking the figure on the regulation basis. Declare a unit and it is checked against the dimension that entry is counted in — kilograms offered to a litres entry returns no points, says which dimension the entry uses, and asks for the figure again. Declare a gross basis and it returns no points in any unit, because gross includes packaging the paragraph does not count. A refusal returns an empty total and an empty verdict rather than "not exempt", which would be a determination nobody made. The shipment summary gains the other half of the same change. Since the start of the month it has declined to report a points total, because the only quantity it receives is gross package mass and that is not what the paragraph counts. It can now be given the ADR quantity per dangerous-goods line, alongside the gross weight it needs for the freight maths. Supply that on every dangerous-goods line and the points come back — 200 litres of petrol, transport category 2, 600 points, exemption applies. Supply it in the wrong dimension and it withholds exactly as the calculator does, because a new field to fill in is only useful if getting it wrong is caught.
One UN number can carry several rows in ADR Table A. UN 1789 hydrochloric acid is packing group II — limited quantity 1 L, transport category 2 — and also packing group III, limited quantity 5 L, transport category 3. UN 1790 has three rows, two of them in the same packing group and separated only by concentration. Asking about a bare UN number picks none of those rows, so packing group, transport category, limited quantity, excepted quantity and tunnel code are facts about a row, not about the UN. The dangerous-goods lookup already returned every row, and the exemption calculator and limited-quantity checker already refused to answer without one — but three places still took the first row and presented it as the answer. The identifier resolver put that row’s packing group, transport category and limited quantity into the record it hands back, alongside a count saying there were two. And the one-line citation — the sentence you copy onto a note and an agent quotes verbatim — read “Class 8 PG II” on the resolver, on the lookup API and on the hosted MCP tool alike, for a UN that is equally PG III. Now the record for a multi-row UN carries no single packing group at all: it lists every variant with its own row index, packing group, shipping name, transport category, tunnel code and quantity limits, and the citation says how many variants exist and which packing groups they span instead of naming one. Single-row UN numbers are untouched and their citations are byte-for-byte what they were. Eleven new tests, run against both the REST API and the hosted MCP server, hold every one of these paths to it.
The shipment summary accepted a UN number on a line and returned a 1.1.3.6 points total with an exemption verdict. The only quantity it receives is gross weight per item — the weight of the goods plus their packaging, pallet included — and ADR 1.1.3.6.3 counts none of its four categories that way. It counts litres for liquids, net mass for solids and for liquefied, refrigerated or dissolved gases, receptacle water capacity for compressed gases, and articles as their mass without packagings. So 200 kg gross of petrol was scored as though it were 200 units of the counted quantity, when roughly 270 litres of product sit inside that weight — a number on the wrong side of the threshold for a load near the limit. Two more faults sat in the same block: for a UN with several Table A rows it silently used the first row’s transport category, which for UN 1789 is a three-fold difference in the multiplier; and it carried its own copy of the multiplier table that missed the nine entries the regulation gives a multiplier of 20 and a 50 kg ceiling rather than the general 50, so 20 kg of a blasting explosive came back as 1,000 points instead of 400, landing exactly on the exemption threshold. Dry ice, which Table A puts outside ADR entirely, came back as an applicable exemption with zero points — there is nothing there to be exempt from. The summary now reports what it can stand behind: which UN numbers are present, which of them Table A puts out of scope and under which section, and which have several variants that must be pinned before any ADR verdict means anything. The points total and the exemption flag are explicitly empty, with the basis rule stated and the exemption calculator named as the place to get a real answer from a real quantity. Seven new tests hold it, on the REST API and the hosted MCP server.
August 2026
The quick-reference table of common freight codes on the HS lookup page offered a “View” on every row, and clicking one left you exactly where you started. Both the code and the View link pointed at a web address carrying the code as a query — and the page ignored that query entirely, so it re-rendered the same lookup screen with an empty search box. Nothing broke and nothing was reported; the destination simply looked identical to the origin, which is the least visible kind of dead link. Every row now links straight to that code’s own page. All twelve are four-digit headings with a real page already in the sitemap, so each click is now a direct arrival rather than a detour. The query form is fixed too rather than abandoned, because it was not only used here: the duty calculator links to it when you look up a commodity code, and the public duty API hands it back as a suggestion when it cannot find the code you asked for — and in that case the code is eight digits, which has no page of its own anywhere in the published nomenclature. So the query now resolves by walking the code down through its levels — eight digits to six, six to four, four to two — and stopping at the first that genuinely exists in the data, rather than guessing from the shape of the number and landing you on a page for something else. UK national relief codes in chapters 98 and 99 sit outside the published nomenclature and have no page to reach, so those leave you on the search box rather than a false 404. The lookup page itself is unchanged in how it is served — it is still delivered straight from the cache rather than rebuilt for each visitor, which was the constraint the fix had to respect.
If you opened the HS code browser and clicked a section, there was a good chance the site answered with a rate-limit page instead — on your first click, having done nothing wrong. The cause was ours, not yours. Every browse grid on those pages pre-loads the pages it links to as soon as they scroll into view, so that when you do click, the page is already there. Those pre-loads are real requests, and the scraper guard was counting them against you: landing on the HS page pre-loaded twenty-one section pages against a budget of ten, so the budget was gone before you touched anything, and every genuine click for the next five minutes was refused. The ADR list did the same, harder, because it renders fifty links a page. The guard now counts pre-loads separately from real page views, with a budget sized from what the pages actually render, measured rather than guessed: an ordinary four-page walk down the HS tree generates thirty-six requests and a heavy one sixty-four, against a former ceiling of ten. Pre-loading is kept, because it is what makes those pages feel instant; the budget was the bug. Signing in now takes you off the shared per-address budget altogether and meters you as yourself, which also fixes the case where a whole office behind one connection shared a single allowance. None of this loosens the guard against actual scraping: a scripted crawler sends no pre-load markers, so it stays on the tighter budget and is still blocked, and a browser tells us plainly which of its requests are speculative, using a standard header a page cannot forge. Both directions are now covered by a test that runs on every build, alongside two existing guard tests that turned out to have been wired to nothing.
Server-side errors have been reported and fixed since launch; browser-side ones were not, because the file that initialises the browser reporter followed a convention the current build system no longer loads — the page ran without it, and a client that broke in your browser looked fine from here. The initialisation now lives where the framework loads it on every page, with the same settings as before: the same privacy scrubbing before anything leaves the page (no email, no IP address, no cookies, no request bodies), the same 10 % performance sampling, and the connection allowed by the same content-security policy shipped yesterday, which already named the monitoring host. Two things the move corrected on the way: browser events are tagged with the environment they came from (the old file read a server-only variable and would have tagged everything “development”), and client-side page navigations are now traced. Verified by throwing one deliberate error in a throwaway browser session on a preview build and watching it arrive. Alongside it, an internal daily self-check now exercises a service that had failed quietly for three days, so that class of silence is caught the next morning rather than on first use.
Yesterday’s hardening shipped the browser headers with a report-only Content-Security-Policy as a placeholder; today the policy is enforced. It was built from what the site actually loads — the production HTML and every client bundle were inventoried and each page was run in a real browser with the console open — so it names exactly the origins in use (the site itself, the two Google Fonts hosts for one typeface, the Stripe checkout and billing hosts that the plan forms redirect to) and nothing else: no framing, no plugins, no base-tag tricks, no connections to anywhere but this site. One directive is deliberately left out and stated as such: script-src. Next.js only supports per-request nonces on dynamically rendered pages and this site prerenders thousands, so an enforced script policy without weakening it to allow every inline script would mean rendering every page on every request — a cost and speed decision taken openly rather than a header shipped with violations. The report-only header stays as the marker for that one lane. Cross-origin access is now scoped by route class: the documented calculators and lookups, the tools listing, the health endpoint and the hosted MCP keep the open policy because calling them from a browser on another site is a designed use of the anonymous tier; sign-in, account, key, admin, billing, feedback and newsletter routes are same-origin only, with the split pinned by the security gate. The health endpoint gains the data_versions field the versioning policy has promised since it was written — for every editioned dataset, the edition served, the authority’s current edition and when that was last checked, derived from the same registry every response envelope uses — and the policy page now describes exactly what is served.
Until today the account page told a paying customer their tier and a usage bar, and for Pro keys that bar was reading the wrong bucket and showing zero while the real month counter held thousands. Fixed at the source, and the page grew into the view the owner dashboard already had for every customer: usage against the quota (this month for Pro, today for free keys, with the exact reset time), a per-endpoint split of your own key’s calls this month named from the tool registry so nothing is typed twice, the plan card with the next invoice date and the existing billing portal as the one action, and a status strip with the live tool count and the service-status link. A key that has never made a call sees no charts at all — it sees the first call: the curl command with your key masked and a copy button, because the only thing that matters for a new key is one successful request. The per-endpoint history is honest about its horizon: counts exist since 10 August 2026 and are kept for 90 days, and the panel says so rather than implying a full month. Everything reads the signed-in session’s own key and nothing else — a request that names a key is refused before it is even looked at, and that refusal is now part of every smoke run. Behind the scenes the health endpoint now reports the deployed commit, so the pre-merge gate can prove it is testing the deployment it thinks it is.
The dangerous-goods table is rebuilt from the official UNECE ADR 2025 Table A spreadsheet, and that spreadsheet still carries a value the regulation itself has since corrected. ECE/TRANS/352/Corr.2 (October 2025), read in a browser session on the UNECE ADR 2025 publication page, lists eight corrections to the English text, and one of them moves a served field: Table A, UN 3423 TETRAMETHYLAMMONIUM HYDROXIDE, SOLID, column (20) hazard identification number — for 688 read 668. The site served 688, a faithful mirror of the pre-corrigendum spreadsheet; it now serves 668 on the UN 3423 page, the ADR lookup API and the MCP tool, and the row carries its own provenance naming the corrigendum as the authority that overrides the base artifact. Nothing else served moves: the other Corr.2 items delete a definition note, fix cross-references, rename an example (UN 1921 already reads STABILIZED here) and correct Table B and the tank tables, neither of which the site serves. Corrigendum 1 (October 2024) touches nothing served; Corrigendum 3 (October 2025) corrects the French text only. The dataset posture is now the base ADR 2025 text plus Corr.1 and Corr.2, stated in the provenance registry and the public provenance document alike, and a practitioner scenario locks the corrected value against the live API and the UN page from today onward.
A same-day security review of the codebase and the live site found no exploitable defect, but it did find the config-level layer that was missing, and this release adds it. Every response now carries the standard browser hardening headers: the site refuses to be framed by other pages (X-Frame-Options), browsers are told not to second-guess content types (nosniff), the referrer sent to other sites is trimmed to the origin, and camera, microphone and ad-topics access are switched off site-wide — geolocation stays available to the airport search, which asks for it only when you press its use-my-location button. The HTTPS-only policy is now declared with the includeSubDomains and preload directives, and a report-only Content-Security-Policy is deployed as the first staging step towards an enforcing one. The endpoint that emails sign-in links is now rate-limited per client address like every other authentication route, and every per-address limit on the site reads the client address from the platform’s trusted header first, so a request can no longer choose its own bucket. Free API keys issued through the legacy sign-up form are generated from a cryptographic random source and are longer than before; keys already issued keep working unchanged. Finally, the session signer refuses to start in production without its secret configured, so a misconfigured deployment fails at build time instead of running on a default.
A returning user had no way to learn that a tool had been added since their last visit unless someone told them. Two permanent answers, both driven by data that already existed rather than by a list anyone maintains. First, every tool in the registry now records the date it shipped, and every tool card — the homepage grid, the Tools menu, the related-tools links on tool pages — shows a NEW badge for thirty days from that date and then drops it by itself. Second, the homepage carries a What’s new strip: the three most recent changelog entries, read from the same data this page renders, each linking through to the changelog or to the page it concerns. Nothing is typed twice: a tool that joins the registry gets its badge on every surface automatically, and a change that is not in the changelog cannot appear in the strip. Both sit inline in existing rows, so they add no layout shift, and both meet the AA contrast bar in the light and dark themes.
The dangerous-goods dataset carried per-row citations on 55 anchor rows (25 UN numbers, the high-traffic entries), and every one cited two aggregator websites rather than the regulation itself — the audit trail of the original May pass, which the site’s own sourcing bar has since learned to score at zero (tertiary republishers establish nothing). Nothing was wrong with the data; the paper trail under it was borrowed. Fixed at the source layer: all 11 stored Table A columns of all 55 rows were re-verified field-by-field against the purchased UNECE Digital ADR 2025 Table A artifact — 605 of 605 field comparisons matched, zero value changes — and each row now cites the authority: the UNECE ADR 2025 publication page and the ECE/TRANS/352 Volume I PDF it publishes (Part 3 is the dangerous goods list). Both URLs were read in a real browser session the same day; unece.org serves 403 to non-browser fetchers by WAF policy, which the weekly citation-liveness audit already classifies as transport noise, not link rot. Each row carries a decision_rationale naming the diff, the retirement, and the corrigendum status: Corrigendum 1 (Oct 2024) touches none of these rows; Corrigenda 2 and 3 (Oct 2025) were discovered on the authority page during this pass and their content check is tracked openly in the backlog. The dataset-level provenance registry and the public provenance document now tell the same story: ADR re-verification is done, not pending.
Every existing gate verified the site against its own contracts, counts and schemas — so a page could be schema-valid and still give a practitioner the wrong answer, which is exactly how the dry-ice miss happened. The new practitioner scenarios suite closes that class: real questions ("30 kg of dry ice on a van — is it under ADR and what must I still do?") with expected answers derived from the primary source — the purchased UNECE ADR 2025 Table A and volume texts, the WCO HS 2022 legal text, the UNECE UN/LOCODE directory, the US EPA GHG Emission Factors Hub, IATA TACT volumetric rules, the EU ICS2 stop-words list — asserted against the LIVE rendered output: page text as a reader sees it, API and calculator JSON bodies, hosted MCP tool results. Twenty-seven scenarios seed the suite across dangerous goods (scope verdicts, 1.1.3.6 points math, limited quantities, tunnel codes), the identifier resolver (ambiguity, byte-stability, honest zero-match), HS, UN/LOCODE withheld coordinates, emissions factor vintage, chargeable weight and ICS2 descriptions. Failures print the practitioner’s question, not a diff. The suite blocks in two tiers: offline in every build (api/calculator scenarios execute through the real route handlers), and live against every preview before merge plus production daily. Writing the seeds immediately paid for the machine: deriving expected answers from the regulation first surfaced three defects the site had been serving (see the fixes entry below).
First: ADR 1.1.3.6.3 note a. Nine transport category 1 rows — UN 0081, 0082, 0084, 0241, 0331, 0332, 0482, 1005 (anhydrous ammonia) and 1017 (chlorine) — score twenty points per kg under 1.1.3.6.4 with a 50 kg per-transport-unit cap under note a. Every surface applied the generic category-1 rule (fifty points, 20 kg cap) to them: 25 kg of anhydrous ammonia came back "1,250 points, not exempt" where the regulation gives 500 points, exempt. A conservative error, but the calculator, the UN pages and the MCP tool all stated the wrong limit and multiplier. The transport-category table now lives in one shared module with the note-a branch, every surface names the branch when it applies, and freightutils-mcp 2.17.2 aligns the tool description. Second: HS subheadings 8703.23 and 8703.24 still carried the HS 2017 word "reciprocating" under an "HS 2022" label — WCO amendment NG0262B1 deleted it from the 8703.2 subheading text (rotary-piston cars now classify there), confirmed against the UK Trade Tariff and US HTS; the upstream Comtrade abbreviation is itself only partially updated, and the WCO legal text wins. Both descriptions corrected, the verification pin and provenance re-audited. Third: the API docs advertise the response-envelope JSON Schemas at /schemas/response-envelope.v1.json and v1.1 — both URLs returned 404 because the files were never published; they are now served, with a build gate pinning the published copies byte-identical to the source of truth.
The website API grew faster than the workflow platforms that wrap it; today they catch up. Zapier 0.5.0 adds six operations — Calculate Freight Emissions, Validate Freight Identifier, Check ICS2 Goods Description, Find Airport, Find Nearest Airport, and Resolve Freight Identifier — bringing it to 26 operations (15 Creates, 11 Searches). Just as important is what the update removes: every declared output field is now verified against a live production response, so 17 field declarations that production never actually returned are corrected or deleted, and the ADR operations declare the new Table A scope verdicts (not_subject_to_adr, conditions_ref, carriage_prohibited) so a Zap can branch on them. n8n 0.4.0 adds the resolve operation and repairs the ADR DG-validation template to read scope verdicts BEFORE doing any 1.1.3.6 arithmetic — a carriage-prohibited line now dominates the workflow verdict instead of scoring zero points and passing, and an all-not-subject consignment no longer claims vehicle marking and driver certification are required. The Make app gains a Resolve Freight Identifier module and the same scope fields. Identifier spellings differ per platform where a platform forces it; the JSON contract underneath is one and the same.
Born from a real routing miss: the three dangerous-goods tools answer three different questions, and a user landing on the wrong one had no signpost — a correct answer to the wrong question reads like a wrong answer. Every DG tool page — the UN number lookup, the 1.1.3.6 exemption calculator, and the LQ/EQ checker — now opens with a dashed “use this when” card naming all three siblings: the lookup is for what a substance IS (classification, packing group, tunnel code, LQ/EQ values), the calculator is for whether a LOAD’s total quantity stays under the 1,000-point small-load threshold, and the checker is for whether small inner packagings qualify for Limited or Excepted Quantity relief. The current page is marked, the other two are one click away, and the copy lives in one shared component so the guidance cannot drift between pages.
An agent holding an opaque string — “176”, “UN1845”, “NLRTM”, “FOB”, “22G1”, “MSKU1100810”, “D/E” — previously had to know which of two dozen tools fits before it could ask. GET /api/resolve and the resolve_reference MCP tool are the front door: one call, thirteen identifier grammars ALL run (UN numbers, AWB prefixes, airline and airport IATA/ICAO, UN/LOCODE, ISO 6346 container numbers with the check digit computed, HS codes, Incoterms, ADR tunnel codes, ULD serials, ISO container size/type codes), and every match comes back as a typed candidate carrying the core record, a one-line summary, a per-candidate dataset citation, canonical_url and api_url for depth, and a normalized verification_status with its basis stated — record-level provenance where the dataset has it, the dataset-level claim where it does not. AMBIGUITY IS THE PRODUCT: “LHR” returns Heathrow AND the Egyptian carrier that genuinely holds ICAO designator LHR, ranked — never a silent guess — and every candidate carries rank_basis, so the deterministic ordering (record hit, then verification status, then a static dataset prior, then a stable tiebreak) is inspectable rather than asserted. A container number whose check digit fails is returned with valid: false instead of being dropped. Zero candidates is a valid answer: count 0 with a note, never an error, never an invention. Honesty notes worth stating: 8–10 digit national tariff lines resolve at their 6-digit international parent and say so (the dataset is the international nomenclature); an ISO container type that matches the pattern but is not among the 10 served types returns a pattern-only candidate with no record; SCAC/BIC codes are not covered yet. The endpoint is the first to serve response envelope v1.1 NATIVELY — always enveloped, ?envelope=1 for byte-stable v1 — and v1 is deliberately single-token: paste-a-manifest extraction is a later revision, not scope creep. A golden test set from the design spec gates the build: the ambiguity quartet, every grammar, the check-digit fail, garbage in, normalization variants, and the ranking invariants.
A certified ADR planner asked this site whether dry ice needed ADR paperwork and walked away believing it did. The data was right — UN 1845 is listed “NOT SUBJECT TO ADR except for 5.5.3” in ADR 2025 Table A — but the phrase lived in the packing-group field, the calculator ran the 1.1.3.6 small-load arithmetic over a substance that is outside ADR scope, returned “1.1.3.6 exemption applies” (right boolean, wrong law), and never mentioned the things section 5.5.3 actually requires of the planner. That is fixed at the modeling layer, not the copy layer. 28 Table A rows carry a row-level verdict remark instead of a packing group — 14 rows say “NOT SUBJECT TO ADR” (one of them, dry ice, “except for 5.5.3”) and 14 rows say “CARRIAGE PROHIBITED” — and every one is now a proper flag: not_subject_to_adr with a conditions_ref, or carriage_prohibited, with packing_group null and the original cell preserved verbatim in table_a_remark, re-verified row by row against the purchased UNECE ADR 2025 Table A Excel. Out-of-scope rows now SHORT-CIRCUIT every verdict surface. The exemption calculator returns a dedicated verdict — “Not subject to ADR (road). Section 5.5.3 applies: …” — with the 5.5.3 requirements quoted VERBATIM from the purchased ADR 2025 text and cited to subsection: ventilation with the authority’s own definition of “well ventilated” (CO2 below 0.5 % by volume), the “CARBON DIOXIDE, SOLID” package mark, the 150 × 250 mm asphyxiation WARNING mark, the transport-document line with the authority’s worked example, unpackaged-dry-ice separation, and training — and it never runs the points math. The LQ/EQ checker answers not_applicable instead of the old “limited quantities not permitted for this substance”, which had framed dry ice exactly like a genuinely LQ-prohibited explosive. THE CARRIAGE-PROHIBITED HALF OF THE BUG WAS WORSE THAN THE DRY-ICE HALF: an entry ADR bans from the road entirely — nitrogen trioxide, methyl nitrite, fourteen of them — came back “1.1.3.6 exemption applies, exempt: true” from the same null-transport-category fall-through. Those now return exempt: false with “Carriage prohibited by ADR (road)”. Mixed loads stay honest too: dry ice riding with petrol is excluded from the points and the exclusion is stated in warnings rather than silently zeroed. The REST responses, hosted MCP tools, calculator UI, UN pages, search results, envelope citations, OpenAPI spec and the downloadable n8n workflow all say the same thing from one shared module.
Every UN entry page and every exemption-calculator result now shows the WHY, not just the verdict. The threshold panel — a dashed informational card, one per ADR Table A row — states the transport category and what it means in points, how much of that substance alone stays under the 1,000-point threshold, the per-substance maximum from 1.1.3.6.3, the limited-quantity limit with its chapter, the excepted-quantity code expanded into its actual inner/outer limits, the tunnel code with a link to its meaning, and the special-provision pathway where one applies. In the calculator the panel goes further and shows the live arithmetic per line — quantity × multiplier = points, counted against the threshold — so the number in the verdict is checkable at a glance instead of asserted. A planner’s decision fields — is it in scope, which category, how much before the exemption dies, can it ride LQ or EQ, which tunnels — now sit on one card instead of scattered across a lookup, a calculator and the book.
Every airline record can now carry a status: active_verified, historical, unverified, or nothing at all. Nothing at all means never audited, and it stays absent rather than defaulting to false, because a recorded false asserts a check that never happened. Wave 1 worked the 391-carrier AWB/cargo subset — the records freight desks actually look up — against pages the carriers publish themselves, with FAA Order JO 7340.2P Change 3 for ICAO designators and callsigns and IATA’s free code search only as a tie-break. Aggregators and directories score zero: if a number is only findable on airhex or airlinecodes.info, this dataset now treats it as unevidenced. THE HEADLINE IS THE HONEST ONE: 81 RECORDS ARE VERIFIED, AND 5,961 HAVE NEVER BEEN CHECKED. 98 carriers are confirmed ceased and 217 were audited without clearing the bar. 183 of the 399 records worked in this wave — 45% — ended as leave_unverified, and the commonest single reason is worth stating plainly: THE CARRIER DOES NOT PUBLISH ITS OWN AWB PREFIX ANYWHERE CITABLE. SriLankan 603, TAAG 118, Ethiopian 071, Finnair 105, Suparna 871 — all live carriers with confirmed IATA and ICAO identity, and no page of their own that states the prefix. Aggregators have those numbers. We are not counting them any more. Every finding that would change what this site publishes was then attacked by a second pass whose job was to refute it, and 29 OF 212 DID NOT SURVIVE. Two failure modes recurred. The first is the group-site trap: a parent cargo brand lists prefixes without saying which belongs to which member, and that is not evidence for one member. It took down British Airways 125 and Iberia 075 (neither cited IAG Cargo page contains either airline’s name — the page credits both prefixes to "the airline (IAG Cargo)"), Air France 057, KLM 074, Cargolux 172, Qantas-group 081 on QantasLink and Sunstate, and the Air Arabia prefix set. EMIRATES 176 FELL TOO: the only prefix examples labelled as such in Emirates’ own SkyCargo PDF are 888-…, twice. The second mode is cessation resting on a zero-scoring source plus absence from the FAA list, which is negative evidence — city-connexion is the clearest case, since the current FAA list still assigns CIX to it. Where the refutation was about the source rather than the fact, the second pass went and found qualifying evidence: Fly Asian Xpress’s own archived cessation notice, easyJet’s own GB Airways acquisition release, ENAC’s AOC revocation for Livingston, Zefix’s liquidation record for JetClub. TEN VALUES CHANGED, EACH ON CARRIER-PUBLISHED EVIDENCE. Air Arabia Abu Dhabi’s AWB prefix was wrong — 360 corrected to 049, proven twice on the carrier’s own rate sheet and tracking endpoint. Jeju Air 575 → 806, Starlux 693 → 189, Mercury World Cargo gains 805 ("its own air waybill (prefix 805)" on the carrier’s own site), Air Inuit’s callsign AIR INUIT → INUIT, and Georgian Airways’ IATA code was TGZ — a three-letter ICAO-shaped value sitting in the two-letter IATA field — corrected to A9. Three records that never operated aircraft (two duplicate Amadeus GDS rows and a GSA/broker) lost their cargo-carrier flag. A VERIFIED RECORD NOW HAS TO BE EVIDENCED TO SAY SO. The 36 records carried over from the May anchor pass cited Wikipedia and airhex alongside the carrier page and had per-field sources for the ICAO code and callsign only — nothing backing the IATA code or the AWB prefix, which are the two fields those pages are about. They would have emitted high confidence and a "verified against carrier-published pages" claim. 25 were re-sourced properly by this wave; the other 14 were demoted to unverified with the reason recorded on the record. No active_verified record in the dataset now cites an aggregator, and every one of them carries a citation for every identifier it serves — a build gate enforces both, in both directions, so the status cannot be inherited again. The API reflects all of it. /api/airlines gained status, status_checked_at, data_note and field_provenance (which page confirms which field), and the answer’s confidence now derives from the records in it: an answer made entirely of verified records is emitted at high with no advisory, and any other mix — one unverified record, one never-audited record, an empty result — stays at medium with the PROVENANCE_PENDING warning. verified survives as a deprecated computed mirror for one release; it was also wrongly declared REQUIRED in the OpenAPI spec while being absent on never-audited records, and that is fixed. 1,450 per-field citations across 392 records, 100 distinct publishers behind the verified set.
/airlines/iata/{code} and /airlines/awb/{prefix} are the canonical airline pages: 321 IATA pages and 355 AWB-prefix pages, each about the identifier and listing every record that holds it, with each holder’s own verification status. Shared and reassigned identifiers are the point rather than an embarrassment — an IATA code that three records claim is exactly what somebody typing that code needs disambiguated. The old /airlines/{name} URLs permanently redirect to the matching identifier page (all 399 verified to resolve, no dead ends) and the sitemap lists only the canonicals. Callsigns are searchable on the browse page for the first time — "speedbird" and "dynasty" previously matched nothing. The browse page also now states its verification coverage up front, deriving every figure from the dataset at build time rather than a typed number.
If you need more than the Pro plan’s 50,000 requests a month, SLA terms, or invoicing instead of card payment, /pricing now says who to talk to. It is deliberately an invitation and not an offer: there is no SLA, no uptime credit and no invoicing today, so the copy promises a straight answer about what can actually be committed to, and nothing else.
Response envelope v1.1 ships, opt-in via ?envelope=1.1. On every dataset that has editions — ADR, HS, UN/LOCODE, Incoterms, ICS2, emissions, container validation — the _source block gains three fields: dataset_edition (what this site serves), authority_current_edition (what the authority currently publishes), and checked_at (when that was last confirmed against the authority itself). When the two editions ever differ, the envelope injects an EDITION_LAG warning automatically — the lag becomes a machine-readable statement in the response rather than a footnote in our docs. AT LAUNCH IT FIRES ON NOTHING: all seven editioned datasets were re-verified current against their authorities the day this shipped, and that was a gate — v1.1 was not allowed to ship advertising a lag. The check is now standing infrastructure: an authority-edition registry records how each authority is checked and how often that check may go stale before the build fails, with the staleness window justified per authority cadence (ICS2 moves weekly; Incoterms changes roughly once a decade). Datasets without editions — airports, UK duty rates — deliberately omit the fields, because an edition claim on continuously-maintained data would be false precision. V1 IS UNTOUCHED, BYTE-STABLE, AND STAYS THE DEFAULT: deployed integrations pin envelope version "1" strictly, so the version is negotiated per-request by query parameter only. The hosted MCP endpoint serves v1.1 to all MCP clients (the schema and response update atomically there), and the npm package requests v1.1 from release 2.15.0 — existing installs keep working unchanged.
The French emission factors moved from Base Carbone V23.6 to the authority’s current v23.11, fetched from ADEME’s live Base Empreinte channel (the open-data mirror this site previously cited is frozen at V23.6 — citing it would have reported "no change" forever, so the citation moved to the live channel). The export’s SHA-256 is recorded in the provenance notes, every ingested row was checked against the provider column — all ADEME, none ecoinvent, which is a binding constraint of the individual registration declared to ADEME — and ADEME’s own FAQ permits exactly this use: making the datasets available in a tool for third-party use. Six of the seven French factors are byte-identical in v23.11; their editions and verification dates now say so. THE SEVENTH IS RETIRED, AND THE REASON IS WORTH STATING PLAINLY. This site has said, on every emissions surface, that GLEC Framework values are never stored or returned — GLEC is cited as methodology only. The v23.11 re-verification revealed that the ADEME sea-container factor was GLEC-derived all along: the authority’s own V23.6 row sources it to "GLEC Framework — Février 2020", and in v23.11 the element became per-TEU-kilometre GLEC-2025 data — the wrong unit for this calculator on top of the prohibited source. Every valid container element in v23.11 is GLEC-sourced, so there is no compliant replacement: the factor is withdrawn rather than substituted. French sea-container requests now return the calculator’s honest not-available answer listing the covered alternatives, and the no-GLEC-values claim is now true of every factor served. The remaining expired-validity factor (inland waterway) is ADEME-original and keeps its value; its edition string now states the authority’s own position — the validity period ended in 2021, the element remains marked valid in v23.11 — instead of hiding it.
Search data showed practitioners asking for the ADR tunnel restriction code table — often in Dutch and German spellings — landing on our guide at position twenty-something and not clicking. The page assumed the problem was understanding; the queries say it is lookup. It now leads with the answer: an atomic definition of what a restriction code is, then the full restriction-code table — every code that appears in the served ADR 2025 dataset, including the two Class 1 quantity-qualified codes the old prose list omitted, with the tank versus package prohibition and what each code means for routing. Per-code counts of UN numbers derive from the dataset at build time rather than being typed. The source line states the edition and verification status the way every other ADR surface does, and a glossary line names the Dutch and German terms for the same column — glossary, not keyword bait. Entries with no tunnel code at all are explained rather than glossed over: they are the rows Table A assigns none, because the goods are either not subject to ADR or not accepted for carriage. Every UN number page now links to the table from its tunnel-code row, and structured data (DefinedTermSet) describes each code for machines.
A new quick-reference chart at /incoterms/chart puts the whole Incoterms 2020 rule set into a single matrix: where risk transfers, where cost transfers, who insures, who clears export and who clears import — split into the any-mode rules and the sea-and-inland-waterway rules, each rule linking to its full page. The facts come from the same verified dataset the lookup and API serve; the wording is FreightUtils’ own summary, because the ICC’s rule text is copyrighted and this site does not reproduce it — the chart carries the trademark attribution and points to the ICC publication for the authoritative text. Print-friendly by design: the chrome disappears and the matrix fits a sheet. Answer-first like the tunnel-code page: the atomic answer (how many rules, the two families, the two asymmetries — only CIP and CIF oblige insurance, only DDP puts import clearance on the seller, only EXW puts export clearance on the buyer) sits above the fold with the source and verification status.
The emissions calculator’s four US EPA factors were sourced from the EPA GHG Emission Factors Hub 2024. The Hub’s 2025 edition (published January 2025) revised the freight ton-mile factors, and the correction is material: TRUCK MOVES FROM 0.11597 TO 0.12841 kgCO2e/tonne-km (+10.7%), AIR FREIGHT FROM 0.62481 TO 0.74991 (+20.0%), waterborne −6.2%, rail −4.1%. Every US-region answer this calculator gave carried the older values — we were 10–20% low on US truck and air, and saying so plainly is the point of logging corrections publicly. The UK (DEFRA) and France/EU (ADEME) factors were unaffected, so only region: us results move. The re-ingest reads the authority’s own 2025 xlsx (recorded with its SHA-256), keeps the same AR5 GWP basis the 2025 edition re-confirms, and documents a precision change honestly: the 2025 file stores the factors at published precision rather than the unrounded intermediates the 2024 file carried, and our derivation now records the authority’s cells verbatim. In the same pass the UK DEFRA leg was re-proven rather than assumed: DESNZ republished the 2026 flat file on 31 July 2026 with corrections scoped to non-freight WTT rows, and all eleven DEFRA factors we serve were re-diffed against the revised file — byte-identical, tank-to-wheel and well-to-wheel both, including the WTT arithmetic. Zero impact, verified rather than expected. One lag remains open and is stated rather than hidden: the ADEME component serves Base Carbone V23.6 while the authority’s live channel is at v23.11, whose change notes revised the fuel-upstream component all our ADEME factors include. The open-data mirror we originally cited is itself frozen at V23.6 — re-reading it would falsely report "no change" — and the live export sits behind an account. The refresh is queued; until it lands, the provenance notes on every emissions response name the lag.
The dataset now ships UNECE’s current 2025-1 edition — 116,232 locations, up from the 2024-2 edition this site had served since launch. The files came from the authority’s own release artifact (the official vocab-locode repository’s 2025-1 release job), fetched on 2026-08-10 and recorded with a SHA-256 in the provenance notes, and the ingest tool is now edition-parameterised so the next edition cannot require code changes to ship. THE LARGER CHANGE IS WHAT THE RE-INGEST REFUSED TO SERVE. Every coordinate now passes through the strict validator the 2026-08-03 audit built — the one that established that UAMIR, UAMYR and ZAUHL carry impossible longitudes in UNECE’s own published strings. Applied to the whole file rather than the three audited cases, it found the same defect at scale: 274 coordinates whose published minutes exceed 59 — "4162N" is not a coordinate, but the previous parser had no minutes check and served it as a confident 42.03° — plus the audited three and the two explicit "0000N 00000E" placeholders. All 279 now return a coordinatesBasis naming exactly why the coordinate is withheld, instead of a number that was never true. Nothing was invented and nothing silently dropped: the basis vocabulary is the audited one (withheld-source-out-of-range, withheld-source-placeholder), and the five locations read live from the UNECE directory on 2026-08-03 publish byte-identical strings in 2025-1, so their audit rationale stands unchanged. Every surface stating the edition moved together: the /unlocode page, the response envelope citations, the provenance constants, the API docs, and the unlocode_lookup MCP tool (freightutils-mcp 2.14.3, republished). The superseded edition and count strings are registered so they cannot silently return.
FreightUtils stopped calling itself ground truth. The homepage section heading, the API-docs overview and the MCP page all made that claim about this site, and this site cannot yet mechanically guarantee it — datasets still carry pending-verification states, and one is an edition behind its authority. The wording now says what is actually true: SOURCE-BACKED FREIGHT FACTS YOUR AGENTS CAN INSPECT AND CITE. The claim moves to where it belongs — on the named sources — and the envelope is the proof, not the adjective. The same pass retired "authoritative" from every description of our own data: the site-wide structured data, the API-docs intro and its structured data, and llms.txt now say "source-backed" and "from named primary sources". Where a page said every answer carries a verified date, it now promises exactly what every response actually delivers: authority, edition, and a verification status an agent can read. Disclaimers that point AWAY from FreightUtils — "not authoritative for navigation", "consult the original publication" — are the honest half of the story and did not move. Five calculator badges said "Real-time"; they meant the result recomputes as you type, which a reader could take as a data-freshness claim — they now say "Instant", which is the thing they meant. The airline page got the same honesty: its metadata called the AWB prefixes "verified" while the dataset’s verification flags are still being reconstructed (the claim is gone); the table now renders three explicit states — verified, unverified, and unchecked — where never-checked records used to display indistinguishably from verified ones; and the CSV export stopped writing "false" for records nobody ever checked, because never-checked is a different statement from checked-and-failed, and asserting an audit that did not happen is exactly the class of claim this release removes.
Every keyed API call now increments a per-key, per-endpoint daily counter, kept for 90 days and then expired by the store itself. It exists so two questions become answerable at a glance — which endpoints is a given key actually using, and when was it last seen — first on the owner’s side, and next as each account’s own per-endpoint usage view, which needs history to exist before it can show anything. WHAT IS STORED IS THE KEY, A COARSE ENDPOINT LABEL, THE UTC DATE, AND A NUMBER. No request content, no parameters, no IP, no email. The endpoint label is selected from the same closed vocabulary the existing tier counters use, so nothing derived from request text can mint a new counter — the property the wrapper-telemetry release established, kept deliberately. The retention figure is a property of the storage rather than a promise about a cleanup script: the expiry is set at first write and re-armed while the key stays active that day, so a dropped housekeeping call cannot quietly turn 90 days into forever. The privacy policy gained the matching line in the same deploy — per-key, per-endpoint call counts, retained 90 days, used for quota and service monitoring — because a counter that touches identity, however lightly, should be disclosed the day it starts accruing rather than found later.
Four integrations publish FreightUtils — the Zapier app, the n8n node, the Make app and the npm stdio MCP server — and until now NONE of them sent a distinguishing User-Agent. Every call they made arrived looking like any other server-side HTTP client, so the server could not attribute a single request to any one of them. The only signals available were the platforms’ own dashboards and npm download counts, and downloads measure INSTALLS rather than calls: a package baked into someone’s Docker image downloads on every build and may never make a request. Each wrapper now sends User-Agent: package-name/version, read from its own package.json so it moves with each release instead of being typed in once and going stale. The server maps those prefixes to a per-surface counter and the owner-only metrics view shows the split. WHAT IS STORED IS A LABEL AND A DATE, NOTHING ELSE — no IP, no API key, no account identifier, no request content, the same metadata-only rule the existing counters follow, and the privacy assertions in the metrics gate were extended to cover the new dimension rather than left applying only to the old ones. The surface label comes from a CLOSED list, which is a security property rather than tidiness: the label is written into a storage key, so if it were derived from the User-Agent text then any caller could create unlimited keys just by varying its own header — enough noise to bury the real numbers. The header only ever SELECTS from a fixed set of five known values, and anything unrecognised is counted as nothing at all. Zapier’s own platform User-Agent is tracked as a separate label from our Zapier app, deliberately: they mean different things — "our app called" versus "Zapier’s infrastructure called and our header did not survive" — and folding them together would hide a broken integration behind a healthy total. Released alongside: freightutils-mcp 2.14.1, n8n-nodes-freightutils 0.3.2, zapier-freightutils 0.4.2, with no tool, schema, field or response changes on any of them.
The n8n node’s operation descriptions advertised "15 ULD types" against an actual 16, and "6,940 HS codes" against an actual 6,937 — the HS figure predates the July verification that removed three non-HS artifacts. A third, in its README, was created by the node’s own previous release: it added an operation and left "19 operations" beside a list of 20. All three were re-derived from the live API rather than corrected from memory, and the vehicle count beside them was checked the same way, found correct, and left alone. Separately, the Zapier app gained a check that drives every one of its operations against production and asserts that each field it ADVERTISES actually comes back. Zapier renders those declarations in the step mapper, so a field that is declared but never returned is offered to users as something they can map and then produces an empty value forever. One such field was found and fixed by hand two days ago; the check found SIXTEEN MORE, across thirteen of the twenty operations, mostly renames the declarations never followed. Each is recorded with its real name and is being fixed as its own change, because correcting a declared output key alters what existing users can map. The check runs the operations’ real code rather than diffing raw HTTP, because several of them reshape the response — a raw diff would report false failures on all of those and get switched off within a week.
The served change first. The double-deck trailer published a rear door opening of 2,480 x 2,700 mm against its own internal height of 3,033 mm. Read literally, the record described a trailer whose top deck cannot pass through its own back door. It was not a claim anybody made: it is the single-deck door figure left behind when the interior height was corrected. It is now NOT PUBLISHED rather than replaced with a better guess, because no cited page supports one. All 32 URLs this dataset cites were read in the previous pass, and rear door width and height are backed by none of them; the double-deck manufacturer FAQ that does back this record’s interior height publishes heights only, and no page read publishes a rear aperture width for any vehicle at all. This release opened no new sourcing, so there was nothing to correct it to, and replacing a contradiction with a confident number would have turned a visible defect into an invisible one. door_dimensions is null on that one record and carries a new door_dimensions_basis of "not-published" alongside it. That basis is deliberately a stronger statement than "unverified": unverified means we hold a figure nobody has checked, not-published means we looked and no publisher states one. The vehicles page used to DROP the Door Opening row when there was nothing to show, which left a reader unable to tell "this vehicle has no published door figure" from "the site is not showing it"; it now states the reason, exactly as the internal-height row already did. Every other record keeps its door dimensions and gains no new field.
The ldm_calculator preset rigid10 asserts a 10 m rigid body. The verdict was returned in the previous release: no manufacturer publishes such a vehicle, and the figure traces to a trade page describing a 26-tonne body as "around the 30ft (10m) mark", where 30 ft is 9.14 m. The payload half is real; the length is not. NOTHING IS REMOVED AND NO FIGURE CHANGES. rigid10 returns exactly what it returned yesterday, same length, same payload, same pallet count, and will keep doing so until the next major version. Removing a published enum value breaks whoever is pinned to it, so the value stays and the advisory does the work: every response using it now carries a warnings entry beginning DEPRECATED: vehicle="rigid10", naming the removal release and the migration. It is marked deprecated in all six places the enum is published, and removed from the vehicle picker on the loading-metres page so nobody new adopts it. Anyone following a saved link with vehicle=rigid10 still gets an answer, with the preset shown as RETIRED and the advisory printed under the picker. Migrating: use vehicle=custom with a vehicle length, or vehicle=artic. No replacement preset is offered, on purpose. The canonical 9.1 m rigid records remain readable through the vehicle lookup, but their envelope is itself still unsourced, and promoting an unsourced envelope to replace a preset being retired FOR BEING UNSOURCED would repeat the fault rather than fix it. rigid75 keeps its name, and that is now a decision rather than a deferral. Its id numeral drops the decimal point of 7.5 tonnes, and it was a rename candidate only because it shared a prefix with rigid10, which encoded METRES: one prefix, two units. Retiring rigid10 dissolves that. What remains is cosmetic, because no served number is wrong, no two tools disagree, and a 75-tonne rigid is not a vehicle, so there is no competing reading available. A rename would cost a second deprecation window for pinned clients to buy nothing. The API spec also gains a correction unrelated to the above: it described the ldm_calculator warnings array as an array of strings, and the endpoint has always returned objects with a type and a message. The spec now says what the endpoint does.
The Calculate Consignment action’s output list offered suggested_vehicle, a top-level billing_basis, and separate chargeable_weight_air, _road and _sea totals. Zapier renders that list in the step mapper, so all five were offered as things a user could map, and anyone who mapped one has had a step producing an empty value on every run since. Checked against the live API rather than against the code: the consignment engine emits no vehicle suggestion at all (a different tool does that); billing_basis is real but sits under totals; and the response carries ONE chargeable weight, chargeable_weight_kg, computed for the mode requested, not a per-mode trio. The sample also named totals.pallet_spaces and totals.item_count, where the real fields are line_count and piece_count. Every field that was genuinely there keeps its exact key, so no working mapping breaks, and the real-but-undeclared fields are now listed. The sample is a copy of an actual response instead of a hand-written illustration of what the action ought to return, which is how the drift got in.
Asking the vehicle endpoint for a slug that does not exist returned a recovery hint suggesting the slug artic-13-6. No record has ever had that slug, so following the advice produced a second 404. It now points at a real record, and a check refuses the build if any recovery hint ever again names a vehicle that does not exist. This is the one code path a caller only reaches when it is already lost, which is exactly where a wrong answer costs the most.
The methodology table’s vehicle row listed its sources by hand, and the hand-written list had drifted from the data in both directions at once. It named four sources that are not cited: a 49 CFR definitions section — retired as a citation on 2026-08-03 precisely because it sets no dimension or weight limit — plus three vehicle manufacturers that appear nowhere among the dataset’s cited publishers, one of them cited only via Wikipedia, which is a tertiary source and, on reading, backs no stored value at all. And it omitted publishers that DO back served figures: Iveco, whose line-up table is the source for three rigid gross weights; Schwarzmüller and Don-Bur, which back trailer dimensions and the pallet counts respectively; Stoughton; and TIMOCOM, which alone publishes the 13.6 m internal length exactly. The audit date was nearly three months stale — 16 May against a newest source read of 6 August, when all 32 URLs the dataset cites were opened for the containment pass. The row is now derived from the dataset rather than recalled, and the superseded strings are registered so the old names cannot quietly return. Nothing about the 17 records themselves changed here; this is the page describing them catching up.
Lead with the served change. Six articulated records stored a gross vehicle weight of 44,000 kg alongside an axle configuration of 2+3. 2+3 is five axles, and no source puts 44,000 kg on a five-axle articulated combination. The affected records are the 13.6 m curtainsider, the mega/jumbo, the box trailer, the refrigerated trailer, the double-deck and the flatbed — the standard UK and EU fleet. A count in our own notes was wrong too, and it was written into the data it described: both the working brief and the dataset’s own citation-replacement prose said “the five artic records”, and counted against the file it is six. What the sources say, and they agree: GOV.UK’s “HGV maximum weights” (DVSA/DfT) indexes its table by vehicle type AND axle count, and under the Authorised Weight Regulations 44,000 kg sits on the SIX-axle articulated row, with its footnote stating the split in terms — “Each part of the combination must have 3 axles and the trailer must have road friendly suspension”, i.e. 3+3. The five-axle row is 40,000 kg. The 44,000 kg that does appear in the Construction & Use five-axle cell carries a different footnote — “Operation at over 38000kg under the C&U regulations is restricted to combined transport operations” — so it is not a general allowance, and the same page forbids mixing regimes: “vehicles may only comply with one of the regulations - individual parts of the 2 regulations cannot be combined.” Directive 96/53/EC Annex I agrees: clause 2.2.2(a) puts a “two-axle motor vehicle with three-axle semi-trailer” — the 2+3 case, spelled out — at 40 tonnes, and its 44-tonne clause 2.2.2(c) requires a THREE-axle motor vehicle carrying containers “in intermodal transport operations, up to a total maximum length of 45 feet”, so 44 tonnes cannot reach a 2+3 at all. Which field was wrong was decided per record, not by default, and the evidence was internal: each of these records’ own rationale already read “max GVW 44t for 6-axle … artic on the UK derogation”. The rationale said six axles while the field said five, so the record always described a six-axle vehicle and the axle string was mis-entered. axleConfig becomes 3+3 on all six and grossVehicleWeight does not move; dropping the weight to 40,000 kg would have turned these into different vehicles. Two phrases this project has used are also corrected. The consolidated Directive does NOT say “40-foot ISO container in combined transport” — the strings “40 feet”, “40-foot” and “ISO container” appear nowhere in it, and the wording is “up to a total maximum length of 45 feet”. And legislation.gov.uk/uksi/2003/1998, cited on the low-loader, is SI 2003 No. 1998, the Road Vehicles (Authorisation of Special Types) (General) Order 2003 — the “1998” is the instrument’s number, not its year, and the Authorised Weight Regulations are SI 1998/3111. Both are the kind of error that survives every link check ever written.
All 32 URLs cited by the dataset were read. Each of the 150 citations now carries a machine-derived list of the fields the page actually publishes and agrees with us on, plus how it was retrieved: 119 carry a field-level claim and 31 back nothing. The 31 are the finding, not the leftovers. Every one of those pages returns HTTP 200 — link checking has always reported this dataset as perfectly healthy, and it cannot see the actual problem, which is a citation that resolves and does not contain the value attached to it. So each empty claim now carries a reason on the wire, and the reasons are specific. Directive 96/53/EC was cited on ten records for internal dimensions it does not publish: its Annex I gives overall vehicle envelope maxima (16.50 m, 2.55 m, 4.00 m), a different quantity from the inside of a cargo body. The three US Code of Federal Regulations sections are the same shape — § 658.15’s “102 inches” is an EXTERNAL vehicle width limit, and 2,591 mm, the “internal width” on both US trailers, IS 102.0 inches. A DAF page cited on four rigid records turned out to be a PDF-link index with no specification printed on it at all, whose own model list runs 8 t to 19 t and therefore excludes three of the four classes citing it. A Faymonville citation points at the company homepage. Two Wikipedia citations are tertiary and describe the wrong market. Nothing was deleted: a page that backs nothing keeps its citation and gains a stated reason, because dropping a source is a separate decision from recording that it does not support the field. Two sources were added because they publish what nothing else did. Don-Bur’s pallet FAQ is the only page found that gives a pallet count WITH its loading orientation AND the pallet dimensions it refers to — “26 UK pallets (2 across x 13 down)”, “UK pallet: 1200mm x 1000mm” — and it tabulates the double-deck case at 52/66, matching the double-deck record exactly. TIMOCOM, already cited for the loading-metre divisor, turns out to publish a vehicle table backing four more fields, including the only source found that states 13.6 m exactly (Schmitz, Krone and Schwarzmüller all publish 13.62 m; Don-Bur 13.7 m). The claims are computed, not typed: the evidence — what each page publishes — lives in a fixture, and a build gate re-derives every claim by comparing that against the stored value and fails if the committed file differs. A range that CONTAINS our number never counts as agreeing with it, and an external maximum never counts as an internal dimension. Change a stored value and support recomputes — correcting the axle configuration above is what turned the GOV.UK citation from backing nothing into backing 44,000 kg. It had to earn it. Three fields are backed by nothing at all across all 32 pages: payload, rear door dimensions, and axle configuration. Not one page publishes a payload figure matching any stored value, only one publishes a rear aperture height and it disagrees, and none publishes an axle-configuration string. That is now visible instead of assumed.
The 18-tonne and 26-tonne rigids no longer report a pallet count, and that is a capability loss said out loud. All three 9.1 m rigid records claimed 16 Euro and 12 UK pallets, and the figure has no source: bodybuilders, truck manufacturers, rental fleets and the pallet networks all publish body length or payload and stop, while the forwarder pages that do carry counts disagree with each other and with us across bodies of 7.0 to 9.15 m. The stated reason for the number was circular, and the dataset disproved it itself. Two of the three declared the count “weight-limited”, arguing that the deck tiles to 22 Euro positions but payload divided by 22 is too low per pallet, therefore 16. That is inverted — payload over a geometric count is just the average weight you would be carrying, not a constraint that reduces the count. The proof is inside the data: the curtainsider rigid has a LOWER payload than the box rigid and the SAME count of 16, with its own rationale noting the per-pallet figure moving while the count does not. The per-pallet weight was back-solved from an answer fixed in advance. Geometry was available and was deliberately not used: tiling this deck gives 22 Euro and 18 UK, not 16 and 12. Publishing a tiled figure is legitimate when the deck is corroborated — that is exactly how the 13.6 m artic’s 33 Euro stands up, with its 2,480 mm width confirmed by Schmitz Cargobull and 33 independently published by both Don-Bur and TIMOCOM — and the 9.1 m rigid deck has neither leg. Tiling an unsourced deck would stack a computation on an unverified figure and present the result with floor-plan confidence. There is also a circularity risk worth stating: while searching for a publisher of these counts, a search summary returned our own envelope and our own figures — “9.10 x 2.48 x 2.60 m … 16 EUR / 12 UK pallets” — attributable to no page anybody opened, with freightutils.com itself appearing in the result sets. We may be the internet’s source for this number, and citing it back would close the loop. The US counts were the wrong figures, in the permissive direction. A named US carrier publishes three counts for a 53 ft dry van and names the loading pattern for each: 26 loaded straight, 30 turned, 28 pinwheeled. We were serving 30 — the turned figure, with nothing saying it needs every pallet turned. It is now 26, the straight figure and the conservative one, with the other two recorded against the record and their patterns named. The 48 ft trailer’s count is withdrawn entirely: that carrier publishes no 48 ft figure, the only one found anywhere (24) comes from a dealer, and our stored 26 matches nothing published for a 48 ft trailer at all — it looks like the 53 ft figure applied to a shorter trailer.
The rigid10 loading-metre preset claims a 10 m load length and a 12,000 kg payload on a rigid truck. It has been carried as a known open item with the instruction to source it or retire it. It cannot be sourced, and the two halves came apart on reading. The payload is real — a national rental fleet publishes “Est Payload +/- 5%: 12000kg” on a 26-tonne DAF box rigid, verbatim. The length is not, and it is above every manufacturer’s own published ceiling: MAN’s bodybuilder data sheets give a maximum recommended body length of 9,360 mm for the 26-tonne TGM and 9,285 mm for the TGS and TGX, Volvo’s equivalent works out to 8,760 mm, and real published bodies run 8.26 m to 9.36 m. The two figures never appear together on one page for one vehicle, and going up a weight class does not help — the Construction & Use Regulations cap a wheeled motor vehicle at 12 m regardless, so a 32-tonne eight-wheeler buys payload, not length. Where the 10 m came from, because a wrong number with a traceable origin is worth more than a deleted one: a trade classifieds page writes that a 26-tonne curtainsider body is “around the 30ft (10m) mark” — and 30 ft is 9.14 m — while the same site elsewhere derives a “theoretical maximum length for a box body of 10 metres” by subtracting only a 2 m cab from the 12 m limit and forgetting the front overhang. The preset is not removed here: removing a published enum value breaks clients pinned to it, and retirement is an alias-then-deprecate change across six surfaces. The verdict is recorded against the waiver so it stops reading as an open question, and the replacement is already sourced for whoever does it — a 26-tonne 6x2 rigid at 8.5-9.2 m and 12,000-15,500 kg clears a two-independent-source bar several times over.
Lead with the correction, because it lands on yesterday’s correction. The previous entry widened the 300-character GPT Actions gate from paths-only to the whole document and reported 362 descriptions. That number was itself an under-count — the exhaustive reference walk finds 365. The three keys the widened walk still never visited, named rather than counted: ContainerSpec/properties/description/description, HsCode/properties/description/description, and Ics2CheckResponse/properties/description/description. All three are schema properties literally named “description”. The walk was node-driven — “does this object have a description?”, then recurse into every key except description. That exclusion is right when a description’s value is a string and wrong when the value is an object, because the property’s own description then sits one level below a key the walk refuses to enter. A freight API has description fields throughout its request and response bodies, so this is not an exotic shape — it is the shape a schema takes whenever it documents a field called “description”. servers/0/description was NOT one of them: the old walk already reached it, which was verified rather than assumed and negative-tested at that location anyway. The fix is exhaustiveness, not a longer exclusion list. The walk is now key-driven — every string-valued description key anywhere, recursing into description whenever its value is an object. The count moved 362 to 365, which equals the exhaustive reference walk, so there is nothing left to exclude and the success line says so instead of implying it. One walk, not two: the offender list and the printed census were separate traversals that could disagree, which is precisely how a success line ends up describing coverage the check does not have, and it is the mechanism behind both of the last two corrections here. They are the same pass now, so the count and the verdict cannot drift apart. The success line also names where the longest description is rather than only how long it is. Negative-tested at both a genuinely-blind and an already-covered location, because a guard that only passes is a guard nobody has tested. Fourth instance in four days of one shape: a check, or a check’s own account of itself, claiming more coverage than it had. No served value changed — openapi.json is byte-identical. The spec was already compliant at 365 of 365; the gate simply was not in a position to know that.
Every cited source on the vehicle dataset gained a fields_supported list. It is a machine-checkable claim about containment: not “this URL resolves” and not “somebody read it”, but “this page publishes this named field”. The existing supports prose stays as the human rationale alongside it. Every one of those lists is empty today, and that is the honest starting state rather than a default: all 147 citations carry an empty fields_supported. A new build rule validates the shape whenever a list is filled — the field must exist in a declared vocabulary, must name a field the record actually carries, and must NOT name a field whose stored value is null, because a null is not a value anyone can corroborate. That last case was found by negative-testing the new rule and discovering it stayed silent. Empty lists are counted and printed on every run instead of failing the build, because a rule that goes red on work in progress is a rule somebody switches off. This is on the wire, so it is worth being exact about. /api/vehicles passes each record’s source list through by reference, which means every per-source key is served verbatim — already true of accessed_at, replaced_from, replaced_reason and supports, verified by curling production. So fields_supported now appears on /api/vehicles, on MCP vehicle_lookup, and in the /vehicles page payload. No existing value changed on any of the 17 records; a key was added. What was measured on the way in, because the plan for this work was two sprints stale: the brief described 51 sources, 24 distinct URLs, one populated read-timestamp and roughly 28 dead links. Counted against the data instead — 147 citations across 30 URLs, 138 read-timestamps, and zero dead links. All 30 URLs return HTTP 200, probed live with a browser user-agent and redirects followed, including the two the brief named as the worst offenders: the EUR-Lex text of Directive 96/53/EC and the gov.uk vehicle-weights collection. Both had already been repointed by the August re-sourcing pass. Link rot on this dataset was fixed before this lane opened, and re-deriving is the only reason anybody knows that. What is actually left is the harder thing, and it is now visible: 43 of the 147 citations carry no containment claim of any kind, including both regulatory citations that appear on ten records each — exactly the ones suspected of being cited for internal cargo-body dimensions and payload figures they do not publish. A page that returns 200 and does not contain the value attached to it is invisible to every liveness check ever written. Counting it is the first step to closing it.
Yesterday’s fix wired a guard that enforces ChatGPT Actions’ 300-character description limit on the public OpenAPI spec. The guard was walking paths only — its success line said “every description”, and info and components/schemas were never visited. Two descriptions were sitting outside it and over the limit. One was SourceEnvelope at 602 characters — the provenance citation block, and not an orphan: it is referenced by 16 response schemas, so it resolves into the action schema for essentially every operation. The other was the API’s own front-door description at 505. So the previous fix corrected the one description the guard could see, while the larger one reached the import through every response that carries provenance. The walk now covers the whole document — every description key, reached recursively from the root — and the success line reports what it actually checked: 362 descriptions, longest now 298. The claim is true by construction rather than by assertion, which is the entire point. (Corrected 2026-08-06: 362 was itself an under-count and “every description key” was not yet true — the widened walk still refused to recurse into a key named description, so three schema properties literally named description sat below it, unvisited. The exhaustive count is 365. Made key-driven the next day; see the 2026-08-06 entry.) Both descriptions were rewritten rather than truncated. The API description drops the JSON/CORS note (the spec already encodes response types) and now leads with what the layer is. The provenance description drops its field-by-field enumeration, because the schema declares all eleven properties and the prose was restating them — and it drops the phrase “from primary sources”, deliberately: that was a blanket claim the per-response provenance envelope now makes honestly, record by record, and not every pending dataset is primary-sourced. Third instance in three days of the same shape — a guard, or a guard’s own description of itself, claiming more coverage than it had.
GET /api/uld’s description in the public OpenAPI spec was 332 characters. ChatGPT Actions enforces a hard 300-character limit, where a single over-length description silently breaks action loading — so that action may not have been loading in the Custom GPT at all. The description was the symptom. scripts/build-gpt-openapi.mjs already validates descriptions against that limit and fails when one is over (corrected 2026-08-05: this originally read “every description in the spec” — it checked every description under paths only, and two outside that walk were over); its own header calls --check a “CI gate”. It was invoked by no npm script, no workflow and no git hook, so the thing built to prevent exactly this had never run once. The description is now 279 characters and keeps everything that mattered, including the warning that an unrecognised query parameter returns 400 rather than being quietly ignored. Regenerating the spec through the generator produced no other change, which is the useful part: the served spec had not drifted from its own generator despite the generator never having run in a wired path. --check now runs inside npm run lint. And because the guard sat one directory above the folder the wiring check scans, that check gained a curated list of gates living outside it — curated, not a wider scan: 58 of 104 scripts in that folder are legitimately one-shot, and a rule that fires 58 times on day one is a rule somebody switches off. Second instance of this class; the first was the vehicle-truth guard in July. Both are now written up as a fault category of their own, distinct from its sibling — one is a gate whose failure gets swallowed, this is a gate nobody calls.
A consistency guard reported that five files computed loading metres by dividing by 2.40 m while the canonical curtainsider record stored an internal width of 2,480 mm, and called every LDM the site publishes 3.2% wrong. It was not wrong. 2,480 mm is the trailer’s internal shell width; 2.40 m is the divisor the loading-metre convention is defined on — two true quantities that had been sharing one field, and the guard was comparing against the wrong one. Every road LDM and chargeable-weight figure published on web, REST and MCP was correct, and none of them changes. Twenty-three publishers were fetched and read to settle it, and every one that states a divisor states 2.4 m: DHL Freight, Gebrüder Weiss, sennder, Savino Del Bene, DSV and ExFreight on the carrier and forwarder side; Soloplan, IMPARGO, Cargo-Planner, EasyCargo, Goodloading, Routix and Pallet3D among load-planning vendors; TIMOCOM and Trans.eu as freight exchanges. Zero variance on the number, and six were re-read directly rather than taken on report. What decided the field name: not one publisher distinguishes a usable loading width from an internal shell width, and TIMOCOM prints a 2.45 m articulated loading-space width on the same page as its 2.4 m divisor. Soloplan calls the figure “a standardised value in Germany”; Cargo-Planner calls it “a constant”. Naming the new field usableLoadingWidth would have asserted a measured quantity the evidence does not support, so it is ldmDivisorWidth with a basis stating whether it is the convention or the vehicle’s own width. SERVED VALUES THAT DO CHANGE, each with its rule. The ldm_calculator presets held their own figures and 19 disagreed with the canonical record they named; they are now derived from it. The 7.5t Rigid goes 5.2 m to 6.1 m, 3,500 kg to 2,500 kg, and 10/8 pallets to 8/6. The 3.5t Luton Van goes 4.0 m to 4.1 m, 1,200 kg to 1,000 kg, and 6 Euro pallets to 5. The 53ft US trailer’s payload goes 20,400 kg to 22,680 kg and the 48ft goes 20,400 kg to 19,500 kg. Four preset pallet counts — the Luton’s UK figure and both US trailers’ Euro and UK figures — were asserting numbers the dataset does not have, and now say “not published” instead of inventing one. US trailer LDM shifts by 0.039%, because the divisor is the canonical 2.591 m internal width rather than a rounded 2.59 m literal. shipment_summary was suggesting trucks too small for the load: its ladder accepted up to 10 m and 12,000 kg before recommending an “18t Rigid”, a vehicle that is 9.1 m and 10,000 kg. It named one vehicle and measured another, in the permissive direction. Tiers are now the named vehicle’s real figures. Two tools also disagreed about silence, and the safe one won: an item with no stackable flag was treated as double-stacked by shipment_summary and as not stacked by ldm_calculator, so the same load returned different loading metres depending which you asked. They now share ldm_calculator’s documented default, so an unflagged item’s LDM and pallet spaces both double — the conservative direction, because assuming a load stacks when nobody said so under-quotes the space it needs. The default had been set in two places at once: aligning the engine alone changed nothing at the wire, because /api/shipment/summary coerced an absent flag to true before the engine saw it. Post-merge production verification caught that, and absence now reaches the engine as absence. palletSpaces was reporting rows: 16 non-stackable Euro pallets came back as 6 from shipment_summary and 16 from ldm_calculator for the same load. It now holds pallet floor positions, which is what the name always promised, and the row count moved to a new palletRows field rather than being dropped. Road responses also stopped shipping an empty dataVersion under a high-confidence envelope. A standard was being cited that says nothing about trailers. Three places on the loading-metre page credited the 2.4 m figure to EN 283, which is “Swap bodies — Testing” and specifies no trailer width. The citation is removed rather than repointed, because no standards body defines the loading metre at all — an EUR-Lex full-text search returns nothing for “loading metre” or “Lademeter”, and UN/CEFACT Recommendation 20 codifies “linear metre” while pointedly declining to nominate a width. A fourth line claimed the vehicle specifications were verified in May 2026 when every record in that dataset carries verified:false; it now reads from the dataset’s own provenance. The guard is wired into the build at zero unwaived divergences, down from 44, with five residuals printed on every run with an owner rather than hidden in an allowlist — chiefly the rigid10 preset, which asserts a 10 m / 12 t rigid no canonical record backs. It is not silently retired: removing a published enum value would break pinned clients, and inventing a record to match it would be worse than leaving it visible.
Every cited source in the ULD dataset has now been opened and read by a second agent — 122 of 122 record-URL pairings, up from 83. That did not verify anything, and the reason is the interesting part. The bar was counting URLs when it needed to count publishers: it asked “does this record have a confirmed primary source plus an independent corroborator?” by counting credited sources, and never asked whether two of those sources were the same source. ANA Cargo publishes the same pallet table in three documents — a spec page, its PDF twin, and a separate March-2026 dimension guide — and PMC, PLA and PAG each cite all three. Under the old count that reads as three independent carriers agreeing; it is one carrier’s number published three times. RKN cites Envirotainer at three of its own URLs the same way. Across the dataset, 148 citations resolve to 125 distinct publishers — 23 were being counted twice. Two families of forwarders were also republishing one page between them, the same problem wearing different hostnames: dsv.com, en.freja.com, sdklogistics.com and Pengelly’s all publish an identical ULD spec template, same figures and same IATA code lists down to the same preamble sentence, while transpacificvn.com and tracosa.com both republish the Wikipedia ULD table with LD-7 blocks identical to the digit. Where a record cited two members of one family, it was corroborating itself. The fix is a curated map, not a similarity engine: a citation resolves to its real source by three mechanisms and no others — a URL that redirects to a cited URL is that URL, a short list of hosts proven to republish one upstream, and otherwise the hostname. Nothing is fetched at build time and nothing is guessed from resemblance. Two candidates were examined and refused: airbridgecargo.com is not grouped with Wikipedia, because Wikipedia carries no RKN specification at all for it to have copied, and ACL Airshop is not grouped with the forwarder template, because the resemblance was noticed but never actually matched. A wrong entry in that map would suppress real corroboration and make a record permanently unverifiable, so the bar for adding one is deliberately higher than the bar for suspecting one. Reading the last 28 pages made things worse on purpose: tare_weight now has no non-tertiary confirmation on 13 of 16 records, up from 11, because two records’ confirmations were withdrawn when the pages turned out to publish a different figure, a range, or nothing at all for that code. Two of the tare figures are bands rather than single numbers — one carrier publishes 185–380 kg, one manufacturer publishes three figures by aluminium sheet gauge — and a band that contains the stored value never equals it, so no amount of further reading collapses it into agreement. All four “access restricted” claims are now disproven; every one came from an automated probe failing, and every one of those pages loads completely in a browser with full specification data, so no citation on this dataset is marked restricted any more. Twenty-nine citations are flagged as retirement candidates and none were removed: they resolve fine, they simply do not carry the value they are cited for, across 13 records and 18 distinct URLs, and dropping a source is a separate decision from recording that it does not support the field. No record flipped, all 16 remain pending-verification with medium confidence, and no served value changed — asserted field by field against a pre-write snapshot, with the ingest re-run three times to prove it lands byte-identical.
accessed_at was carrying two claims in one field: somebody opened this page, and the page agreed with us. Only a CONFIRMED verdict could stamp, so a citation the second agent genuinely read and found to disagree came out looking identical to one nobody had ever opened. The dataset could not tell “we checked, it contradicts us” apart from “we never looked”, and those are opposite evidentiary states. They are now two fields. accessed_at means the page was read on that date and nothing more, so MISMATCH, BASIS-MISMATCH and NOT-ON-PAGE all carry one; UNREACHABLE and LOST-IN-COMPACTION carry none, because for those nobody can show the page was opened. Agreement moved entirely to verdict. Stamped citations went 74 to 145 of 148, and the three without a stamp are the ones genuinely never opened — the read instant is derived from the verdictAt already recorded against each verdict, so this is read off existing evidence rather than backfilled with invented timestamps. The agreement gate is now field-level, not record-level: a record-level bar asks “did two good sources confirm something here?”, which PGA passes on its dimensions while its tare rests on one secondary source that two carrier pages contradict. Status is one word for a whole record, so a record-level bar lets a well-sourced dimension carry an unsourced weight across the line with it. Every field holding a served value now clears the bar on its own confirmations or the record does not flip; fields with no served value are not gated, because a pallet has no internal dimensions and demanding corroboration for a null makes absence unverifiable by construction. All four access restrictions turned out not to exist: satco-inc.com fell in batch 2, and freight.qantas.com, cargo.china-airlines.com and cathaycargo.com each loaded completely in a browser with full specification data. Every one of those claims originated in an automated probe failing, which is evidence about the probe and not about the page, and no primary-restricted citation remains on this dataset. Batch 1 over-claimed and re-reading is what caught it: it recorded DoKaSch’s AKH page as CONFIRMED including tareWeight, but the page reads “Tare weight from 73 kg” against a served 85 kg — not a confirmation, and “from” marks a build floor rather than a point value. A second agent’s table is still a report. The correction landed by re-running the ingest with batch 3 layered over batch 1, not by hand-editing a verdict. Three citations are also not the independent corroboration they appear to be, recorded rather than removed: en.freja.com’s LD9 page and Pengelly’s LD9 fact sheet are byte-identical down to the preamble and AAP cites both; dokasch.com/en/akh/ 301-redirects to dokasch.com/en/akh-ld3-45w/, which AKH already cites separately; onze.lt and caterpillarcargo.com publish the same republished “AS1825 volume / Boeing volume” table. Counting either pair as two sources would launder one upstream claim into corroboration of itself. Second-agent coverage moved 83 to 94 of 122 pairings, with the outstanding 28 regenerated into a committed file with full URLs. 0 of 16 records flipped, all remain pending-verification with medium confidence, and no served value changed. Twenty-six citations now carrying NOT-ON-PAGE are reported as retirement candidates; none were removed, because dropping a citation is a separate decision from recording that it does not support the field.
The delta from the weights re-sourcing went back to a browser. 83 of its 122 record-URL pairings now carry a second-agent verdict; 39 do not, and are queued in a committed file rather than described in prose. The headline is that verification went backwards, which is what verification is for: citations carrying a timestamped accessed_at fell from 93 to 74. Every one of those 19 was a first-agent read the second agent re-opened and could not confirm — a page that turned out not to list the code, or one whose figure disagreed once somebody read the words around it. A stamp means “content read and value re-confirmed”, so a stamp the second agent cannot reproduce has to come off. No record flipped to verified; all 16 remain pending-verification with medium confidence, and no served value moved. A new verdict exists, because two different failures were wearing one name: BASIS-MISMATCH means the page publishes a perfectly good number for the OTHER quantity — DoKaSch’s bare-unit “from 520 kg” against a field whose declared basis is a carrier system tare including the net. That is not a disagreement, it is a category error, and calling it a MISMATCH would re-create exactly the comparison the basis layer was built to make refusable. PGA’s 565 kg tare has its first real confirmation and two contradictions to go with it: Cathay Cargo — one of three carrier pages an automated probe could never reach — was read in a browser and confirms 565 kg with the basis stated on the page (“nets attached”), while Qantas publishes 535 kg and China Airlines 505–545 kg, both the same carrier-system basis, both disagreeing. The served figure is now the top of a read, sourced carrier band rather than a number resting on pages nobody had opened; it is unchanged at 565, the conservative choice within that basis. One access restriction turned out not to exist: satco-inc.com was cited primary-restricted because an automated probe could not reach it, and it loads completely in a real browser with full spec data, so that citation is void. Finally, the vehicle citations have a worse problem than link rot, flagged rather than fixed here: gov.uk/hgv-maximum-weights carries generic axle-count weight tables and no internal dimensions, no payload and no vehicle-type names, yet it is cited on ten vehicle records for exactly those figures; Directive 96/53/EC gives external envelope maxima, not internal cargo-body dimensions per model; the DAF page is a PDF-link index. All resolve; none contain what they are cited for.
The bar that decides whether a dataset may call itself verified took a list of source classes and nothing else. A class is derived from the hostname: the classifier looks at the domain and decides whether the publisher is a manufacturer, a carrier, a trade body or an aggregator. It says who publishes the page. It cannot say whether anyone opened it. So a citation could clear the bar on a hostname, and that was not hypothetical — the ULD dataset was carrying four sources with verdict UNREACHABLE and no timestamp, three of them the carrier pages that are the entire evidential basis for PGA’s tare weight, and every one of them counted as a full primary source. The bar was measuring the same thing the old “≥4 independent sources” rule measured — the shape of a citation rather than its content — one level up from where that defect was last fixed. The bar now takes evidence: a source counts only if it was read (a confirmed verdict, a parseable timestamp, and a basis-check that did not fail), or is an access-restricted authority cited by identifier and clause, which contributes but can never be the only leg — a record whose citations are all clause references has no read behind it. Passing bare classes now throws rather than quietly returning false, because a caller that “fails closed” while believing it evaluated the bar is the same false assurance in a politer form. A new build gate proves it goes red on the 403-walled-primary case. No dataset status changed and no served value moved: this is the gate, not a re-grading.
meta.source on /api/uld read “≥4 cited sources per record, each read and verdicted 2026-08-03”. That was true of the 77 citations when it was written. A re-sourcing pass later the same day added 71 more, taking the dataset to 148 — and those 71 carry first-agent verdicts only. The sentence had quietly become an over-claim about verification work, which is the failure mode this particular line has now been corrected for three times. It now states the split: 77 second-agent verdicted, 71 awaiting that read. The record status was never wrong — all 16 ULD records remain pending-verification with medium confidence and a PROVENANCE_PENDING advisory, exactly as before.
Two numbers were living in one field. max_gross_weight was holding both a structural rating — what the unit is certified to carry, published by its manufacturer with no aircraft or deck attached — and a carrier operational limit, which is what a named airline will actually load on a named aircraft, deck and position. Those answer different questions, and the field never said which. That is why PAG served 4,626 kg while PAJ served 6,804 kg for the same 318×224 pallet. Both figures are real: DoKaSch publishes PAG and PAJ at 6,804 kg as one article, while ANA Cargo publishes five different limits for that same PAG — 4,626 on B747, 4,676 on B777, 5,102 on B767ER, and on B767F a split of main deck 3,835 / lower deck 6,803. Every record now carries max_gross_weight_basis and a qualifier naming the publisher, the carrier and the aircraft. Two things we previously published about this were wrong and are corrected here: the earlier entry called PAG’s 4,626 kg “the belly figure” — it is not, it is ANA’s B747 figure, aircraft-scoped rather than deck-scoped; and it said AAP’s 6,033 kg “is published by none of its own sources” — it is published by VRR Aviation as the structural rating and by Korean Air Cargo, which prints 6,033 kg (277 kg). The same correction applies to RAP, whose 6,033 kg is published by Envirotainer, the manufacturer. Note also that “main deck means the higher limit” is false — on ANA’s B767F the main-deck figure is the smaller one, so a deck cannot be inferred, only sourced; per-deck figures are stored only where a carrier actually published a deck-split table. Tare has the same problem, and it is why 13 of 16 tares matched no source. Manufacturers weigh the bare unit — VRR publishes PMC at 88 / 93.9 / 101.5 kg for 3.0 / 3.3 / 3.7 mm sheet gauge. Airlines weigh what they hand you, net and straps included; Qantas Freight says so in terms: “Tare weight from (includes cargo net)”. For PMC the manufacturer band is 85–101.5 kg and the carrier band 110–125 kg, and they do not overlap. Every record now carries tare_weight_basis, and where no source cleanly attributes the served figure the basis is unverified with the observed band recorded. One served value changed: PGA’s tare weight, 486 kg to 565 kg. 486 kg sat below every published figure including the manufacturer’s own floor — DoKaSch publishes “from 520 kg”. The carrier band is 505–565 kg and the conservative figure within that basis is 565 kg, the safer assumption for payload. Being clear about the evidence: that 486 kg is wrong rests on a source we independently re-fetched at 200; the specific 565 kg rests on three carrier pages our verification probe could not reach, cited access-restricted with no timestamp pending a browser re-read. Still pending on all 16 records, deliberately — these are first-agent reads. Alongside, container specifications were re-sourced from scratch: seven of their ten citations were dead and the only three that resolved were Wikipedia, so the dataset had zero live independent citations; they are now cited from Ocean Network Express, Evergreen Line and Seaco, plus ISO 668 and ISO 1496-1 by clause. Vehicle citations were rebuilt too, including two retired rather than replaced because they never supported the values attached to them: 49 CFR 393.5 is a definitions section that sets no width, length or weight limit, and the ATP Agreement prescribes no body dimension anywhere in its annexes.
The ULD tool description told agents that “as of 2026-07-26 no source carries a timestamped content read and none has been second-agent verified”. Both clauses stopped being true earlier today, when the second agent read all 77 cited sources and 26 confirmed reads were stamped. The status was never wrong — provenance is still pending — but the reason given for it understated the work done and overstated the gap, which is the same defect class as the over-claim it was originally written to retract. It just failed safe instead of unsafe. The line now names the real blocker: provenance stays pending because no record yet has a confirmed primary source plus a second independent non-tertiary source agreeing with it — on each record that second source either contradicted the stored value or did not list the code. It also warns specifically about tare_weight, which no non-tertiary source confirms on any of the 16 records, so any payload derived from max_gross_weight minus tare_weight rests on a figure nothing independent corroborates. This one could not be fixed website-only, which is why it needed a package release. The hosted /api/mcp renders tool descriptions from the installed freightutils-mcp package, not from the website’s own source — so the corrected text only reaches agents once the package is published and the site re-pins to it. The familiar “npm inherits it through the REST proxy, no republish needed” shortcut holds for data and _source, which the proxy passes through live; it does not hold for description text, which is baked into the tarball. freightutils-mcp is now 2.12.3 on npm and in the MCP Registry, and this site pins it exactly. Docs-only in the package — no tool, schema or behaviour change, still 24 tools.
The ULD dataset’s provenance has been asserted, downgraded and re-asserted twice without anyone ever reading the sources. That read has now happened: all 77 record-source pairings across the 16 records were opened and judged. 26 confirmed, 30 disagreed with the value we serve, 19 did not list the code at all, and 2 were lost to a context compaction — those count as no evidence rather than as a pass. All 16 records remain pending, and that is the finding, not a formality. The bar is one confirmed primary source plus one independent non-tertiary source agreeing with it. No record clears it: on every single one, the second non-tertiary source either contradicted the stored value or did not list the code. This is not pedantry about Wikipedia — re-running the arithmetic with the ULD reseller sites promoted from tertiary to secondary changes the outcome for exactly one record out of sixteen. What the responses now carry, per record, is which fields the evidence actually covers, because status is one word for the whole record and cannot say “dimensions confirmed, tare not”. tare_weight is confirmed by no non-tertiary source on any of the 16 records — every carrier and operator page either omits tare or contradicts it, so if you are computing payload from max_gross_weight minus tare_weight, that subtraction currently rests on nothing this run could corroborate. internal_dimensions is confirmed on exactly one record (RKN, by Envirotainer, its manufacturer). external_dimensions is confirmed on 13 of 16, max_gross_weight on 12, usable_volume on 11. No served value was changed. Every disagreement is recorded verbatim next to the citation it came from as mismatch_detail, and they fall into three groups: rounding inside the documented tolerance (1588 vs 1587 kg, 6804 vs 6800), an axis convention the records already explain (ACL Airshop publishes an LD3’s contour-top width where we publish its base), and genuine contradictions needing a source this run could not settle. Changing a figure on one contested read is the “plausible reconstruction” this dataset already refused once on UN/LOCODE, and every record stays pending so nothing is over-claimed in either direction. One structural finding is flagged rather than corrected, and it is the one most likely to bite: max_gross_weight conflates deck-specific limits across records. PAG serves 4,626 kg — which Cargolux publishes as the belly figure — while PAJ and PMC serve 6,804 kg, the main-deck figure, for the same 318×224 pallet, so an agent comparing PAG with PAJ sees a 47% difference between near-identical units. DSV states the split outright for AAP (“Lower deck 1,587 / Main deck 6,000”), and AAP is worse still: its served 6,033 kg is published by none of its own sources. Assigning a deck to each record needs per-record carrier attribution this run did not capture, so the contradiction is documented on the affected records and the numbers are untouched.
code was not a parameter this endpoint understood, so it was silently discarded and the request fell through to the unfiltered branch: count 16, every record, HTTP 200, no warning. An agent taking results[0] got AKE by luck of alphabetical ordering — and would have got the wrong unit for any other code. The response field is literally named code, so it is the obvious thing to filter by. code is now an accepted alias for type and returns one record. More importantly, any unrecognised query parameter now returns 400 naming it rather than being ignored, so a filter this endpoint does not understand can never come back looking like a successful unfiltered query. type, category and deck are unchanged, and no field was removed from any response.
Three records (UAMIR, UAMYR, ZAUHL) shipped longitudes above 180° and were withheld in July on the assumption we had mis-parsed a ×10 decimal shift, with a correct re-parse as the fix. The second agent read the raw strings UNECE actually publishes, and the assumption was wrong. Feeding those published strings — “4829N 38150E”, “4939N 30590E”, “2946S 31219E” — through UNECE’s own documented ddmmN/S dddmmE/W format reproduces 381.83, 306.5 and 312.32 exactly: the three bad longitudes, to the digit. The old code applied the format correctly. The source string is what is out of range, and UNECE’s own page-generated map link reaches the same impossible number from the same string. “4939N 30590E” even carries 90 minutes, which is not a valid minute value in any reading. So there is no better parse to reach for, and all five stay withheld — but they now say something truer. Their basis moves from describing our stored value to describing the authority’s: withheld-source-out-of-range for the three, and withheld-source-placeholder for GBWSG and Wellampitiya, whose rows exist but publish the literal placeholder “0000N 00000E”. Both now carry the UNECE URL and a timestamped read, so the absence is sourced rather than merely asserted. The read was edition 2025-1 while the shipped data is 2024-2, so the invalid strings persist in the newer edition — this is not a stale-edition artefact. A new build gate pins the one parser these five go through, and its most important assertions are the negative ones: a 4-digit longitude is refused rather than padded to 38.25° or shifted to 381.83° — both plausible, both unverifiable — and the 0000N 00000E placeholder is recognised explicitly, because 0,0 is a perfectly valid coordinate that no bounds check could ever catch.
The European Commission’s 3 August 2026 revision of the ICS2 Common Repository stop-words list is in force today, and the checker mirrors it as of today. Two goods-description terms are added — GENCA and General items, both dated 03.08.2026 — and the six terms the Commission had marked “Temporarily disabled” are restored: Fish, Glass, Iron, Steel, Metal and Model. The shipped goods-description list goes from 245 terms to 247, and from 239 active to 247: nothing is temporarily disabled any more. The flag rate will go up, and that is the list doing it, not a change to how matching works. Those six are ordinary English words. A description reading “steel” or “glass” passed this tool yesterday because the Commission had switched those entries off; today it flags, because the Commission switched them back on. Expect this to reach descriptions that are genuinely specific: “stainless steel valves” now returns a flag on Steel, as do “cast iron manhole covers”, “glass bottles for laboratory use” and “frozen fish fillets”. Those are flagged as embedded, not standalone — which is the distinction that matters. An embedded flag says the description contains a term that is unacceptable on its own; it is not a verdict that your description is too vague, and “stainless steel valves” identifies a product perfectly well. The note returned with each flag says exactly that. If this reads as noisy on specific descriptions, the disagreement is with the Commission’s list, not with the checker — and mirroring it verbatim is the point of the tool. /api/ics2-check, ics2_check and /ics2-check mirror the EU list verbatim including the Commission’s own statuses — no editorial override — so a status flip lands here whichever direction it goes. The matcher itself is untouched: still word-boundary, so “fisherman”, “ironing board”, “modelling clay” and “fibreglass” are all still clean, and “GENCARGO” does not trip “GENCA”. The standing caveat is unchanged and still the important part: the list is non-exhaustive, a flagged term will likely cause an ENS rejection, and a clean result does not mean the declaration will be accepted. The wave had been staged since June — the document sat on CIRCABC with a future date, recorded here with its node and its sha256, under a standing rule not to ship it until it was effective. Applying it today started by re-fetching that document over the no-login guest path and confirming the sha256 was byte-identical to the value recorded when it was staged; a document that had changed after being recorded would have been a finding, not a formality. Terms were then re-extracted deterministically rather than transcribed, and the extractor was validated by running it against the previous in-force PDF and requiring it to reproduce the previously committed 245-term fixture exactly, in order, before its output on the new document was trusted. The diff is exactly the two additions and the six status flips: nothing was removed, no other term or date moved, and the party-field rows are unchanged at 26. Every count here is derived from the document; none is typed. Why the six were disabled in the first place is still not known, and the source does not say — the prose of the 4 May and 3 August documents is byte-identical, the only footnote in either explains what “Initial list” means, and the marker sat in the implementation-date column with no note attached to it. The six revert to “Initial list”, their original Spring 2025 status, rather than to a new 3 August date. That question is now recorded against the data as unanswerable from the primary. The daily CIRCABC drift watcher reads 3 August 2026 as the current in-force version with no upcoming version pending, and the build gate that hard-asserts every shipped term against the committed authority snapshot was extended to cover this wave — including that the snapshot’s sha256 and the watcher’s baseline can never silently disagree. No endpoint, tool or schema changed (still 23 REST / 24 MCP) and no npm republish is needed: ics2_check is REST-backed, so the hosted surface serves the new list to every consumer at once.
All six pallet records — PMC, PMCQ7, PLA, PAG, PGA and PAJ — stored internal_dimensions byte-identical to external_dimensions. A pallet has no walls and no roof, so it has no internal dimensions to report. The consequence was arithmetic: an agent multiplying PMC’s internals got 12.65 m³, against the same record’s stored usable_volume of 11.5 — a 10% disagreement describing a space that does not exist. All six now return internal_dimensions: null with internal_dimensions_basis “unbounded-open-top”, which is the convention the flat-rack containers already use rather than a new one invented for this dataset. The 163 cm that had been sitting in PMC’s internal height is not a wall — it is the aircraft contour ceiling that cargo may be built up to. It now lives in max_build_up_height_cm, alongside max_build_up_height_deck stating which deck the figure applies to. PLA and PAJ carry no deck attribution in the record, so theirs is null and flagged rather than guessed at. Every one of the 16 records also gained capacity_basis, including a new value: build-up-envelope. A pallet’s usable_volume is real, sourced and operationally useful, but it is bounded by the aircraft the unit flies in rather than by any enclosure the unit has — reusing “carrier-published” would have re-encoded the exact ambiguity these fields exist to remove. The provenance flag was the larger problem. All 16 records shipped verified with high confidence, while chrome_verified_at was null on every one of them and sources was an array of bare URL strings with no accessed_at slot at all. The dataset was structurally incapable of recording the proof of the claim it was making. sources is now {url, accessed_at} on every record, all 16 are pending, and /api/uld and the uld_lookup tool now return provenance_status pending-verification with medium confidence and a PROVENANCE_PENDING warning — the same treatment UN/LOCODE already had. UN/LOCODE was the more honest dataset reporting the lower confidence, and that inversion was being held in place by a test fixture asserting “uld verified”. To be clear about what the downgrade does and does not mean: all 29 cited URLs were checked and 29 of 29 resolve. This is not link rot, and the numbers are not suspected of being wrong. The gap is that nothing has been read against its source with a timestamp, and accessed_at stays null rather than being stamped from a liveness probe — a 200 proves a URL answers, not that the page still says what was transcribed from it. Earning verified back is a content read, and the integrity test now refuses verified without both a stamped source and a second-agent confirmation, so the claim cannot return unearned. No endpoint or tool was removed. external_dimensions, max_gross_weight, tare_weight and usable_volume are unchanged.
A sweep of all 116,129 UN/LOCODE records found five that could not be true. Three stored a longitude beyond 180° — UAMIR at 381.83, UAMYR at 306.5, ZAUHL at 312.32 — which is not a close call about accuracy, it is off the planet. Two more stored exactly {0, 0} while describing places in Essex and in Sri Lanka: a sentinel value, not a location, and one an out-of-range check would never have found because 0 is a perfectly valid coordinate. All five now return coordinates: null with coordinates_basis saying which failure it was — withheld-out-of-range or withheld-null-island — so the reason travels with the absence instead of the record simply going quiet. None of the five was repaired, deliberately. The ×10 reading is compelling: ZAUHL divided by ten lands Umhlali within about 0.02° of where it actually is. That is precisely what makes it dangerous — a plausible reconstruction is how a wrong number gets shipped with confidence. UNECE returns 403 to us on every country page tried, so there is no fetched authority to restore from today. Withholding is honest; dividing by ten is a guess wearing a decimal point. Restoring any of the five requires a fetched source and a timestamp. nearest_airport was never affected: it ranks against a separate airports dataset that swept clean, and it already validates caller-supplied bounds. A new build gate ships with this — seven rules that ask whether a record could physically exist at all, rather than whether its fields have the right types: a longitude beyond 180°, an internal dimension larger than its external, a volume that disagrees with its own dimensions, a unit mass below its own tare. Against the data as it stood before this release it found 36 violations. Against the data as it ships, zero.
July 2026
Container specs carried a type violation that broke arithmetic silently. euro_pallets was a STRING on /api/containers and container_lookup — and four of the ten values were ranges written with a typographic en dash, like “23–24”. The same field name is a plain number on /api/vehicles. So an agent doing arithmetic on a container’s pallet capacity got NaN from Number(“23–24”), or silently got 23 from parseInt — the lower bound only, the upper discarded without a warning. Both fields are now numeric pairs: euro_pallets_min plus a nullable euro_pallets_max for the genuine ranges, and the same for gma_pallets. The old string fields are unchanged and still returned — marked deprecated in the OpenAPI spec, and now derived from the numbers so they cannot drift; they will be removed in a future major version, not this one. The two flat racks reported a cubic capacity that is physically impossible. A flat rack has no roof and no side walls, yet the 20ft stored a 228 cm internal height and 31.1 CBM — and 31.1 is exactly 583 × 234 × 228, the volume of a box using a height that bounds nothing. Those two were the only records whose capacity matched length × width × height exactly; every other sits 0.1–0.7 CBM above it, because carrier capacities are not box multiplications. That exactness is what identified them, and Hapag-Lloyd’s own flat-rack pages confirm it: they publish no capacity figure at all for flat racks, only a loadable envelope, while every other type on that site has one. Both records now return a null height and null capacity with height_basis “unbounded-open-top”, plus deck_area_sqm (13.64 and 27.99 m²) — the figure that actually means something for an open deck. Open tops are a genuinely different case and are now modelled as one: they publish a real internal height that applies with the roof fitted, and the roof comes off for crane loading, so they carry height_basis “constrained-removable-roof”. Two stored numbers were wrong and are corrected against a live-fetched carrier source. The 40ft reefer stored ISO code 45R1 — a 9’6” high cube — alongside standard-height dimensions (259 cm external, 218 cm internal), contradicting every other record in the dataset carrying that same code character, which store 290 cm. Hapag-Lloyd’s 40ft High Cube Reefer page confirms ISO 45R1 and a 2,425 mm internal height, so the heights were the error: now 242 cm internal, 290 cm external, 255 cm door, capacity recomputed to 63.9 CBM. The 45ft High Cube stored a 30,480 kg max gross beside a note admitting operators publish 32,500 kg — knowingly storing a figure the record said nobody uses, understating usable payload by about 2,020 kg. The carrier publishes 32,500 kg gross and 27,700 kg payload, and its tare matches ours exactly; those are now stored. Every record also gained fields stating what its numbers mean — capacity_basis, max_gross_basis, pallet_capacity_basis and open_top — because all ten stored a 30,480 kg max gross, which is the ISO 668 ceiling rather than an operational limit, so payloads derived from it understate what carriers run. And the ten records previously shared one byte-identical 473-character “provenance” paragraph; each now carries its own, with its own arithmetic. Two new build gates ship with this: one asserts no field name can carry two types across datasets and no quantity is stored as a string, the other checks every container’s ISO 6346 size code against its stored external height. A repo-wide sweep of every remaining dataset is published at docs/audit/cross-dataset-integrity-sweep-2026-07-25.md, including three UN/LOCODE records shipping longitudes above 180°. No endpoint or tool was removed.
Two records in the road-vehicle reference set — the 13.6m flatbed trailer and the low-loader — stored an internal height of 0. That was never a measurement; it was a flag meaning “open deck, no roof”, and the record’s own provenance note admitted it. But a zero in a numeric field is a wrong answer with a confident shape: anything multiplying length × width × height got 0 m³ for a flatbed. Both records now return height: null, plus two explicit fields — open_top: true and height_basis: “unbounded-open-top” — so the reason travels with the value instead of having to be guessed. Every one of the 17 records carries height_basis, because a bare null would only have moved the problem: “no height limit exists” and “a limit exists but we don’t have a trustworthy figure” are different answers, and nothing should have to infer which is meant from the absence of some other flag. Three numbers were corrected where the record contradicted itself. The flatbed’s payload was 27,000 kg while its own stored working read “44t gross − 12t tractor − 6t trailer = 26t” — that 26t figure had been rounded up for spec-sheet tidiness, so the stored number disagreed with its own arithmetic; it is now 26,000 kg (the underlying tare assumptions are themselves unverified and are flagged, not adopted). The double-deck trailer stored a 2,700 mm internal height with a note claiming “two decks of approximately 1,350 mm each” — 2 × 1,350 is exactly 2,700, leaving no thickness for the deck between them, and 2,700 mm turned out to be the standard curtainsider’s value copied across. Don-Bur’s published double-deck specification (fetched live, and the only citation in this dataset with a real timestamp on it) gives a 3,033 mm internal aperture at roughly 1,500 mm per deck for a step-frame build inside the European 4.0 m height limit; the record now stores those figures and says which configuration they describe and that taller UK builds exist. And the LWB Transit van’s note claimed three Euro pallets fit “in a single file” — three at 1,200 mm each is 3,600 mm against a 3,400 mm load length, so that basis was false. The count itself was left alone and flagged rather than guessed at. Every record also gained pallet_capacity_basis, which finally answers a question the data never did: is this count how many pallets fit on the floor, or how many the payload allows? The 13.6m artics’ 33 Euro pallets is genuine floor geometry (11 rows of 3). The 7.5-tonne rigid’s 8 is not — its floor holds 15, but 2.5 tonnes across 15 pallets is 167 kg each, so weight runs out long before space does. Seven records are marked unverified, the honest answer where we cannot yet tell — including the 26-tonne rigid, whose dimensions and both pallet counts are byte-identical to the 18-tonne rigid despite 40% more payload. The arithmetic for all 17 assignments is in each record’s provenance note, and the open questions are documented at docs/audit/vehicles-lane1-findings-2026-07-25.md — including that 28 of the dataset’s 51 cited source URLs return 404, so the regulatory prose here should not be trusted until a real re-sourcing pass runs. The verified: false flag and the pending-verification caveat stay exactly as they are. Two new build gates ship with this: one proves no consumer of a vehicle height can emit 0, NaN or a silent omission, and one checks the dataset against every other place in the codebase that asserts a vehicle fact. No endpoint, tool or field was removed; /api/vehicles and the vehicle_lookup tool gained four fields and are otherwise unchanged, and the npm package needs no republish.
The Harmonized System dataset was the largest reference surface still marked “pending verification” — pending only because it had been sampled (50 codes), never fully diffed. It now is. Every one of the shipped 6-digit international codes (levels 2/4/6 — chapters, headings, subheadings) was checked by script against four independent open customs authorities fetched live: the UN Comtrade H6 nomenclature (the WCO HS 2022 published list, public-domain), the US HTS (USITC), the UK Trade Tariff (HMRC) and World Bank WITS. The diff is clean: every code exists in HS 2022 at the correct level and parent, no invented or pre-2022 codes, no HS 2022 code missing, tree integrity intact — and every shipped code is backed by at least three of the four authorities (most by all four). The one correction: three “Commodities not specified according to kind” entries (99, 9999, 999999) were removed — these are UN Comtrade’s use of the WCO-reserved chapter 99 for statistical catch-all, not part of the HS 2022 international nomenclature (they are absent from the US, UK and WITS lists). The HS index is therefore 6,937 nodes (was 6,940). Descriptions were re-confirmed against the UK and US subheading text. As a result the /hs pages and the homepage no longer show the pending caveat for HS, and every HS API and MCP response now returns provenance_status verified with high confidence and no pending warning. A new build gate (lint:hs) pins a representative sample of codes across every chapter — including the duty-lookup fixture and the six previously-corrected subheadings — to the authorities, so the dataset cannot silently drift back. Full audit: docs/audit/hs-verification-2026-07-24.md. No change to any code you can look up beyond the three removed catch-all entries; no change to the npm package contract.
Yesterday’s fix made connecting to the MCP server free of the daily rate limit, guarded instead by a light per-IP connection burst cap. That per-IP cap left one edge case: a hosted-agent platform or an enterprise network that routes many users through a single shared outbound IP could, at peak, exceed the connection burst and get briefly refused mid-connection — the same failure, but for the highest-volume clients. The connection burst cap is now identity-aware. Requests carrying a valid API key are counted per key at a high ceiling (600 connections/minute each), so every user behind a shared IP gets their own independent connection budget and large keyed clients are effectively unthrottled for connecting. Anonymous (keyless) connections are still counted per IP, with the cap raised from 60 to 120/minute for small shared networks. Nothing changed for actual tool calls — the daily and monthly usage quotas (25/day anonymous, 100/day free key, 50,000/month Pro) are exactly as before, and the scrape protections are untouched. Website-side only; the npm package is unaffected.
The homepage hero stat strip and the tool grid showed “· verified [date]” for HS codes, airlines and UN/LOCODE — all three carry a pending-verification provenance flag (only a sample or anchor set has been audited, not the full published list), which the same page’s “Sources you can audit” table and each dataset’s own page already reported as pending. Those two surfaces were sourcing a verified-date straight from the last-audit field, independent of the derivation. Now the verified-date derives from the single provenance flag (datasetProvenanceStatus, the same one the API’s _source.provenance_status carries): a surface shows “verified [date]” only when independent verification is actually complete, and a pending dataset shows “provenance pending verification” instead — consistent across the whole site. The /methodology dataset table, which had omitted airports, the EU ICS2 stop-words list and the identifier check-digit schemes, now enumerates every reference dataset with its status read from the same flag. A new build gate asserts no web surface can render “verified [date]” for a dataset the flag hasn’t verified, and that both the homepage provenance table and /methodology list every dataset — so this class can’t recur on a page.
Connecting to the hosted MCP server (/api/mcp) and listing its tools no longer counts against an agent’s daily rate limit. Previously the anonymous 25-requests-a-day cap was applied to the whole MCP surface — the discovery documents, the connection handshake, and the tool calls alike — so an anonymous agent could be cut off mid-connection before it ran a single tool. The surface is now split into a control plane and a data plane: connecting and discovering (initialize, tools/list, the notifications and list methods, and the .well-known discovery documents) is free, bounded only by a light per-IP burst cap, while the daily and monthly quotas — 25/day anonymous, 100/day free key, 50,000/month Pro — now apply only to actual tool calls. Every existing quota is unchanged. When an anonymous caller does exceed the tool-call limit, the tool now returns a clear, model-readable message pointing to a free key or a Pro upgrade instead of a bare error. Two supporting fixes ship alongside: the OAuth Protected Resource Metadata document at /api/mcp/.well-known/oauth-protected-resource now returns a valid response (it was a 404 that also counted against the limit), and the MCP endpoint now rejects cross-site browser requests that carry a foreign Origin as a DNS-rebinding safeguard, while leaving native clients — which send no Origin — untouched. No change to the tools themselves (still 24) or to the npm package.
The per-dataset provenance status the API returns (and the caveat the pages show) was reconciled against what the underlying audit records actually contain, and is now derived from them so the two can’t drift. Two datasets moved to verified: air-cargo ULDs — all 16 records cross-checked against at least four independent authoritative sources each — and Incoterms, whose 11 Incoterms 2020 codes and mode categories were diffed live against the ICC-published rule list; their pages no longer show the pending caveat and their lookups now return high confidence. Two moved to pending: HS codes and UN/LOCODE — only a sample or anchor set had ever been audited, not the full published list, so the honest status is pending until a complete diff is run, and their pages now show the caveat. The dataset-level status is now computed from the per-record audit data rather than hand-set, and a build gate blocks any future contradiction: a dataset can never report “verified” while one of its own audit records is still unverified, and confidence can never read “high” on a pending dataset. ADR, airports and the deterministic calculators are unchanged; UK duty remains a live request-time lookup.
The npm package’s uld_lookup description and the hosted /api/mcp surface were updated to match the ULD dataset’s reconciled verified provenance. The tool’s limitations line no longer says provenance is pending independent verification — it states the specs are compiled from manufacturer and carrier-published sources and provenance-verified against at least four independent sources per record, while noting airline-specific ULD variants still differ (confirm operationally critical dimensions with the carrier). The response envelope’s provenance_status for uld_lookup now returns verified with high confidence. Docs-only patch — no tool, schema or behaviour change; still 24 tools. airline_lookup, container_lookup and vehicle_lookup keep their pending-verification notes.
The API has always returned a per-dataset provenance status on every response (_source.provenance_status: verified or pending-verification), but the web pages didn’t show it — a vehicle or container spec page rendered bald numbers while its own API said the underlying provenance was still pending independent verification. That gap is closed. Every reference page whose dataset carries a pending-verification flag now renders a calm, informational caveat stating that the values are compiled best-effort from published manufacturer, standard and regulatory sources with independent verification still pending, and to confirm against the plate, carrier or operator guide. Verified datasets show no caveat and the deterministic calculators are unaffected. The caveat text is single-sourced with the API’s own wording (no drift), renders identically in light and dark mode, and adds no layout shift. The /methodology page gained a per-dataset Status column reading the same live flag, so nothing is listed as verified while its data is still pending. A build gate fails if a pending dataset ever ships a page without the caveat.
Many UN numbers carry more than one row in ADR Table A — different packing groups or concentration bands with genuinely different limits. UN 1789 (hydrochloric acid) is PG II with a 1 litre limited-quantity limit and transport category 2, OR PG III with a 5 litre limit and transport category 3; UN 1790 (hydrofluoric acid) has three concentration bands with different categories again. The LQ/EQ checker and the 1.1.3.6 exemption calculator used to silently pick the first row and return a confident verdict — which could be the wrong packing group. Now, when a UN number resolves to more than one row and you have not said which, both tools withhold the verdict and return the candidate rows for you to choose from, rather than guessing. Both tools gained optional packing_group (I / II / III) and variant_index inputs to pin one row; single-row UN numbers behave exactly as before. The web LQ/EQ checker and points calculator now show a packing-group / concentration selector for multi-variant UN numbers and hold the result until you pick one. A build gate enumerates every one of the 366 multi-variant UN numbers and asserts neither tool can emit a bare verdict for them, with UN 1789 / 1790 / 1791 / 1796 pinned as permanent regression cases.
The npm package’s adr_lq_eq_check and adr_exemption_calculator tools mirror the fix: both accept optional packing_group and variant_index, and a UN number that resolves to multiple ADR Table A rows without one returns ok:false with blocking_errors [{ code: "AMBIGUOUS_UN_VARIANT" }], human_review_required and candidates[] (each candidate’s variant_index, packing group, proper shipping name and decision fields) instead of a wrong verdict — the ambiguous response is HTTP 200 so the candidates flow through the stdio proxy intact. Single-row UN numbers are unchanged. Still 24 tools, no name changes; a minor version bump because the new inputs are additive and backward-compatible.
The air ULD reference behind the CBM calculator (/cbm) and the ULD lookup (/api/uld, /uld, and the uld_lookup MCP tool) now agree on every IATA type code. A focused re-audit found the CBM side had several mislabels: AAP was shown as LD7 (it is LD9); the M1 container carried the wrong code AAF (the M1 code is AMJ — AAF is the separate lower-deck LD-26); the LD11 (ALP) had a 223 cm width copied from a full-width unit (it is 153 cm, ~7.2 m³); an “LD9 (AAG)” entry duplicated the LD9/AAP container under a code that is not IATA-registered; and a “P6P” pallet listed a 498 cm length that matches no real unit (the 20-ft main-deck pallet is PGA at 606 cm). On the ULD side, the PLA half pallet was mislabelled “P1P” — a designation that belongs to the 88-inch PAG full pallet, not the 60-inch half pallet. Every corrected code was verified against ≥4 independent authoritative sources (carrier ULD guides, pool operators, forwarders) with independent adversarial re-checking. The affected /cbm pages were renamed to their correct codes with permanent redirects from the old URLs, and a build gate now blocks any future case where one IATA code maps to two different names or dimensions across the two datasets.
Three reference-data fixes from the monthly methodology audit. HS codes: six HS 2022 subheadings were showing their parent heading’s full title instead of the subheading’s own scope — corrected to the actual 6-digit meaning (for example 9111.90 now reads as the parts of watch cases, not “watch cases and parts thereof”; 8472.90, 7306.90, 7419.90, 8477.80 and 5702.10 similarly). Air ULDs: the LD1 (AKC) and P1P pallet had inherited a neighbouring unit’s dimensions — the LD1 now carries its own contoured 234 × 153 × 163 cm / ~4.7 m³ (a distinct, larger unit than the LD3, not the same volume), and the P1P its correct 88-inch base (318 × 224 × 244 cm), narrower than the 96-inch PMC/P6P pallet; two adjacent entries (M1, LD9) were de-conflated in the same pass. UN/LOCODE: a code lookup now flags locations UNECE has not officially approved (requests under consideration, user entries, unverified or to-be-removed codes) with a provisional-status warning and lower confidence, so an unapproved code isn’t treated like an approved port. A build gate now guards both drift classes; every corrected value was checked against at least four independent authoritative sources.
The MCP server now has its own page at /mcp instead of scrolling into the API docs. It lists all 24 agent tools grouped by category (calculators, reference, and the MCP-only subscribe helper) with a one-line summary and a link to each tool’s own page, shows install snippets for Claude Desktop, Cursor and Cline plus the remote URL, and explains the response envelope every tool returns: a typed result with confidence, provenance and a ready-to-quote citation an agent can both use and cite — one API request per successful call. Rate-limit tiers (anonymous, free key, Pro) and API-key passthrough are documented alongside directory links (npm, MCP Registry, Smithery, Glama, Postman). The MCP link in the top navigation now goes straight here; the API docs remain the REST reference, cross-linked both ways.
The sign-in, confirm, password-reset and security pages now share the same modern design as the rest of the site, in light and dark mode. Sign-in leads with the Google/GitHub/Microsoft buttons and adds a quiet note for anyone who signed up with a provider and may not have a password yet. The account page gained a plan card showing your tier, price and renewal (or end) date for Pro, a usage meter against your quota with the reset time, and API-key controls to reveal, copy and regenerate your key inline (regenerating invalidates the old key immediately). Billing, invoices, card and cancellation stay in Stripe’s secure portal via the Manage billing button, with refund-policy and terms links alongside.
The npm package now fetches the response envelope with the legacy_source bridge and reconstructs the flat legacy text channel from it — one API request per successful call instead of the two that 2.11.0 introduced, so anonymous and Pro rate-limit budgets are consumed once per call again (verified live: X-Ratelimit-Remaining decrements by exactly 1). content[0].text stays byte-identical to 2.10.x/2.11.x for every tool (32/32 byte-parity checks against production, success and error paths, POST tools, the emissions rich-text layout and the adr_lookup 404); structuredContent remains the clean public v1 envelope, byte-equal to the default ?envelope=1 response. Errors fall back to the flat call for exact legacy error bytes (2 requests — the rare path). Supersedes the 2.11.0/2.11.1 double-request caveat.
The response envelope gains an opt-in legacy_source bridge field (?envelope=1&legacy_source=1) that lets the freightutils-mcp package rebuild the flat legacy body from a single request instead of two. Bridge-internal, not part of the public v1 contract; the default envelope and the flat channel are byte-unchanged.
All 24 tools now share one modern page language. Lookups (ADR UN numbers, HS codes, Incoterms, airlines, airports, UN/LOCODE, containers, ULDs, vehicles) render as reference cards: a monospace identifier headline, a category-colour badge, an at-a-glance specification grid, and a provenance footer showing the source, its last-checked date and verification status — with a one-click "Copy citation" that copies exactly the citation the API envelope carries. Calculators (CBM, chargeable weight, LDM, pallet fitting, unit converter, emissions, UK duty, ADR exemption and LQ/EQ checks, consignment) recompute live and show their work: the governing figure large, the full formula with your numbers, a method line naming the convention and parameters used (IATA divisor, W/M rule, GLEC factors…), and "Copy result" pasting value, derivation and method as one line. Every page is themed by category (freight-ops green, dangerous-goods red, customs blue, reference purple) with subtle line-art, works fully in light and dark mode, and meets WCAG AA — keyboard reachable, screen-reader announced results, reduced-motion respected. Honest provenance throughout: pages whose datasets are still pending independent verification now say so consistently. Same URLs, same calculations, same data — nothing to relearn.
Docs-only patch: the uld_lookup tool description said "15 types" while the ULD dataset holds 16 — caught by the pre-merge adversarial review and corrected upstream. The hosted MCP endpoint now serves tool descriptions generated at build time from the pinned npm package, so the hosted and npm definitions can never drift again (a build gate fails on staleness). No tool, schema, or behaviour change.
The npm package now declares a typed outputSchema on all 24 tools and returns structuredContent: the FreightUtils v1 response envelope, built by the API itself — the answer under result, plus confidence, normalized_input, warnings, _source and a ready-to-use citation, identical to the hosted MCP server and the REST ?envelope=1 opt-in. Every tool definition was rewritten for agent intent-matching: accepted input formats and defaults spelled out per parameter, behaviour prose covering result limits, rounding, no-match and rate-limit signaling (retry_after_seconds), an explicit "Returns:" line backed by the schema, proactive limitations ("structural check, not a registry lookup"; "reference data, not legal or compliance advice"), and cross-references between related tools. Compatibility: the flat JSON text channel is byte-identical to 2.10.x, including error behaviour, so existing text-parsing consumers are unaffected; note each call now makes two API requests (flat + envelope), which counts double against rate limits for stdio users. The hosted /api/mcp endpoint reports 2.11.0 by construction. Still 24 tools, no name or input-schema changes.
The agent-facing response shape that debuted on 3 tools now covers every REST endpoint and MCP tool. It wraps the answer under result and adds confidence (with basis: deterministic, provenance, match_quality or freshness, and a 0–1 score for fuzzy matches), normalized_input (what was interpreted or defaulted), _source, citation, and — where relevant — warnings, blocking_errors (each with a recovery action) and time-boxed validity. Behaviour matches each tool: deterministic calculators are high-confidence; reference lookups carry their provenance status (verified → high, pending-verification → medium with an advisory); fuzzy searches (airline, HS, nearest-airport, UN/LOCODE) return a match score plus an honest best-match warning; time-boxed tools (UK duty, ICS2) carry validity. Opt-in, so nothing changes for existing consumers: add ?envelope=1 (or Accept: application/vnd.freightutils.v1+json) on REST; the flat shape stays the default and byte-unchanged. On the hosted MCP server the envelope rides structuredContent while the flat text channel is preserved. Fully documented at /api-docs; the contract is schemas/response-envelope.v1.json and a build gate validates every enveloped response against it. No new endpoints, tools or counts (still 23 REST / 24 MCP).
Bing flagged a batch of page titles as too long for search results. Every title is now 60 characters or fewer (down from up to 80), fixed at the template level so all ~3,000 pages — pallet, container, HS, ADR, airline and calculator pages — stay concise, unique, keyword-first and branded, with a build gate keeping every title within the limit. A handful of ADR pages whose packing-group field holds a long note (e.g. UN 1845, dry ice) no longer overflow. The API docs also now read consistently: 23 REST endpoints, 24 MCP tools (the 23 REST-backed tools plus get_subscribe_link).
The site header showed "Sign in" to everyone, including signed-in users. It now checks your session and shows Account and Sign out when you're signed in (and nothing flashes while it checks). Pages stay fully static — the check is a single client-side call to the session endpoint.
You can now sign in with an email and password: set one from your account's new security settings (or at signup — the create-account form takes a password up front, activated when you confirm your email), reset it via a single-use emailed link, and change it any time. Passwords are stored only as salted argon2id hashes — never the password itself, never shown to anyone. Sign-in links still work as the secondary path and got two fixes: they now last 30 minutes (was 15 — short enough that real clicks were missing the window) and only your most recent link is valid, so a stack of requested emails can't leave you guessing which one works. Sign-in with Google, Microsoft, and GitHub is wired and will appear as the providers are enabled — one account per verified email, whichever way you sign in. The privacy policy documents all of it.
Sessions previously died a fixed 7 days after sign-in no matter how active you were, forcing a weekly magic-link round-trip. Sessions are now a 30-day rolling window: any authenticated visit more than a day after the last renewal quietly extends the session another 30 days, up to a 90-day absolute cap from the original sign-in (then one fresh sign-in starts a new cycle). Signing out still ends the session immediately, everywhere it's checked. Existing sessions carry over — no forced re-login from this change. Cookie security is unchanged (httpOnly, Secure, SameSite=Lax).
Trust additions across the homepage: every tool tile now shows the exact request that reproduces its example (generated at build time next to the example itself — copy the curl or open the live JSON); the provenance table gains per-dataset status badges using the same vocabulary agents see in llms.txt (verified · pending-verification · computed · live) with a legend, plus a "Spot an error? Dispute a value." link to a structured data-dispute issue template on GitHub; a plain-language boundary line under the hero states what FreightUtils is not (booking, filing, tracking, carrier rating, legal determination); a three-path start fork (build an agent / integrate a system / get IT approval) and an operational trust strip (changelog, versioning, deprecation, methodology, IT pack, DPA, status) close the page; and the pricing strip gains a production lane (SLA / higher limits / audit export → contact@freightutils.com). Status badges are display-layer today; the API-level provenance_status semantics land with the API-docs redesign.
The FreightUtils mark is now a versioned asset set under brand/ in the repo: the nav's slash glyph (General Sans Semibold "/", CTA orange #EF9F27) extracted to outlined SVG masters with tile, wordmark and dark-mode variants, PNG exports at every platform size, and regenerated social banners whose dataset counts derive from site stats. On-site favicons and touch icons were re-exported from the same masters — the favicon no longer depends on whatever monospace font the viewing platform picks, so it renders identically everywhere.
The live /api/airlines response meta hardcoded "airlines": 6352 and "last_verified": "May 2026"; it now derives the count from the dataset (6,357) and the date from the provenance record (2026-05-13), so it can no longer drift. The /api-docs example response is regenerated verbatim from a live call and now also shows the per-record provenance an audited record carries (sources, audited_at, decision_rationale) plus the _source envelope, with the new fields documented in the response-fields table. The cargo-airline figure is corrected from a stale 390 to 391 (airlines with an AWB prefix on record) on /api-docs and /methodology, and copy that said "verified cargo AWB prefixes" now says "with cargo AWB prefixes" — the dataset's verified flag (multi-source-confirmed) applies to 284 records, and we don't claim it for the rest. Prevention: a new lint:stale-counts build gate greps every current-state source file and the rendered HTML for exact superseded strings, so a corrected number failing to propagate now fails the build instead of waiting for a manual sweep to notice.
The homepage is now a dense reference surface: a six-cell dataset strip (counts, issuing authority, last-audited date), a Pack→Weigh→Classify→Declare→Land consignment lifecycle, a 24-tile tool grid where every example is executed at build time by the live tool it belongs to (AWB prefix 176 → Emirates; UN 1203 → Class 3 PG II; an ISO 6346 check digit computed by the validator itself), a provenance table mirroring /methodology (including honest "provenance pending verification" rows), rate-limit and distribution sections, and an agent terminal showing a real adr_lookup response with its _source envelope. A build-time pipeline generates every number and example from the datasets and calculation functions — hand-typed counts fail the build gate. Hover-only motion, keyboard-operable menus and tabs, full dark-mode parity, prefers-reduced-motion respected site-wide. The API docs open with the same terminal hero (MCP / curl / GPT tabs plus an npx install strip), and /pricing gains a Teams line (pooled usage across seats — email contact@freightutils.com to talk; no self-serve checkout yet).
The airline dataset has held 6,357 entries since its last additions, but several surfaces still displayed a stale hand-typed 6,352. All current-state surfaces (site stats, tool registry, MCP tool description, the redesigned homepage strip — which now derives the count from the dataset itself at build time — and the brand banners) now state 6,357. Dataset contents are unchanged; the npm package description refresh rides the next scheduled package update.
The ICS2 Stop-Words Checker (ics2_check, /api/ics2-check, /ics2-check) now flags 12 additional goods-description terms that the EU added to the ICS2 Common Repository list on 2 February 2026: Aid products, Comercial, Consumption, Ensemble, Fake, Headwear, Item, Miscelaneous, Miscellaneous, N/A, Oddments and Promotional. The list is mirrored verbatim from the EU CIRCABC source (TAXUD.A.3.003/TA, current in-force document), including the term statuses the Commission itself sets — FreightUtils does not editorialise it. "N/A" is now also treated as a goods-description stop-word per the EU categorisation (it was previously excluded here as a party-field-only placeholder; party-field checking remains out of scope). The shipped list now matches the EU current-in-force set exactly (245 terms; 239 active, plus 6 the Commission has temporarily disabled that this tool therefore does not flag). Reference check only — a flagged term will likely cause an ENS rejection, and a clean result still does not guarantee acceptance. No change to the tool count (still 23 REST / 24 MCP), endpoints, or schema; no npm republish (ics2_check is REST-backed).
Refreshed the agent-facing surfaces so AI agents and directories read the current, accurate picture. Rebuilt llms.txt as an agent-first index — an identity + agent-safety contract, a connect block (MCP remote endpoint + npm + REST + OpenAPI + Custom GPT + key signup), and every tool phrased as the capability an agent searches for; llms-full.txt regenerated to match. The Identifier Validator now leads with its parse-a-string use case everywhere (the /validate page, /api-docs, and the OpenAPI + MCP tool descriptions): hand it any text — a booking line, an email — and it finds and check-digit-validates every ISO 6346 container, IATA AWB and IMO number in it, with single-identifier validation as the secondary path. Both GitHub READMEs and the MCP registry metadata (server.json, package.json) were refreshed to the neutral-reference + validation-layer, source-citing framing, the OpenAI GPT Store is now listed as a distribution surface, and stale tool counts across the repo docs were corrected to the live figures. Copy and documentation only — no tool, endpoint, schema, data or behaviour change; tool counts unchanged (23 REST / 24 MCP). All OpenAPI operation descriptions remain within the 300-character agent-tooling limit.
Shortened the operation descriptions in the OpenAPI spec (/openapi.json, shown on the interactive /playground explorer) so the spec imports cleanly into agent tooling that caps each operation description at 300 characters — e.g. ChatGPT Custom GPT Actions, which otherwise silently fails to load the action. Five over-length descriptions were trimmed (emissions, ics2-check, consignment, validate, airports), keeping the same meaning. Spec text only — no endpoint, parameter, response, or calculation change; tool counts unchanged (23 REST / 24 MCP).
June 2026
The emissions calculator (emissions_calculator, /api/emissions, /emissions) now returns extra self-describing fields — additive only, with no change to any factor value or the calculation. Each result carries a human-readable summary to surface; a mass_basis reminder to use actual gross mass, not chargeable/volumetric weight (a common air-freight mistake); an empty_running field (whether the fleet-average factor already includes average empty running — so you do not double-count an empty-return leg); a representativeness label (low for sea and air, medium for road/rail/inland) with a high-variance warning in the summary for sea/air (a generic default for screening only — real emissions vary materially by vessel/aircraft type, load factor and routing); and a last_verified date on the factor (when it was last checked against its source). The method wording is now "computed using the ISO 14083 / GLEC distance-based method". On the /emissions page, sea/air estimates are rounded harder to signal their imprecision while the API keeps full precision. The freightutils-mcp MCP output schema gains these fields (republished as 2.10.1). No factor values changed; tool counts unchanged.
Fixed airport search (airport_lookup) returning incorrect results for IATA/ICAO code queries; exact code matches now rank first. Searching a code such as "GLA" or "EDI" via the search box could previously surface a name-substring match (e.g. Brazil's "Glauber Rocha" airport for GLA) or small airfields/heliports instead of Glasgow / Edinburgh, because the search truncated its scan before reaching the exact-code record. No data change; the nearest-airport tool and the direct iata/icao lookups were unaffected.
You can now regenerate your FreightUtils API key from your account page — useful if your key may have been exposed. Regenerating issues a new key, immediately deactivates the old one, and shows the new key once (copy it before you leave the page). The account page already shows your plan, masked key (last 4 characters only), usage this month, and a Stripe billing-portal link for managing your subscription. This is a logged-in, account-only page; the older /dashboard link now takes you to the same account page.
Two new reference tools backed by the public-domain OurAirports dataset (85,555 airports). airport_lookup (GET /api/airports, the /airports page, MCP airport_lookup) resolves an airport by IATA code, ICAO code, or name/city search and returns the full record — both codes, name, type (large/medium/small/heliport/closed/seaplane), municipality, region, country, coordinates and elevation — with ranked candidates for ambiguous names. nearest_airport (GET /api/nearest-airport, MCP nearest_airport, and the "Nearest to me" button on the /airports page) takes a caller-provided latitude/longitude and returns the closest airports by great-circle (haversine) distance, with distance_km on each result; optional radius_km, max_results and type filter. Coordinates are input only — never stored or logged (the audit log records neither query nor body, and the Sentry redactor now also drops lat/lon). The data is public domain: all 1,178 large airports were cross-checked against the independent OpenFlights dataset (95.2% agree within 5 km + IATA) and a representative sample — including every outlier — was independently re-verified against Wikidata (all within 0.8 km), with provenance and limitations documented. Built to the tool-quality bar: the website MCP and the npm stdio server register both tools via registerTool with full output schemas + structuredContent and a _source citing OurAirports. Reference data only — coordinates are the airport reference point, not for navigation; verify current codes with IATA/ICAO. Counts are now 23 REST tools / 24 MCP tools; the freightutils-mcp npm package adds both tools and is republished as 2.10.0.
New tool: the ICS2 Stop-Words Checker flags terms from the official EU ICS2 stop-words list (vague or generic goods descriptions that the European Commission deems unacceptable in an entry summary declaration). Paste a goods description and it returns the flagged terms — each with a note on whether the term is the standalone description (an automatic ENS rejection) or embedded in a longer description (make it more specific) — plus a clean boolean, a caveat and a _source. There is no binary accepted/rejected verdict. Available as GET /api/ics2-check, the /ics2-check page (with an interactive checker), and the MCP tool ics2_check. The list is the EU Commission DG TAXUD list (base guidance v4.10 plus the 4 May 2026 update; legal basis Commission Delegated Regulation (EU) 2015/2446, data element 18 05 000 000 — cited, not reproduced) and is non-exhaustive: a clean result does NOT guarantee acceptance, and a flagged term will likely cause a rejection. It is a reference check only — not an ENS filing, not a customs-compliance determination, and not legal advice — aimed at customs and documentation teams, brokers, forwarders and agents building filing pipelines. Submitted descriptions are not persisted or logged. Counts are now 21 REST tools / 22 MCP tools; the freightutils-mcp npm package adds the tool and is republished as 2.9.0.
The Emissions Calculator (/emissions) and Identifier Validator (/validate) pages now include a live, interactive calculator — enter your inputs and get an instant result from the same /api/emissions and /api/validate endpoints, in addition to the existing API documentation and worked examples. Behind the scenes, the site footer's Calculators and Reference columns now render from the single tool registry so they can no longer drift from the navigation, homepage and sitemap, and a new surface-consistency guard asserts that every tool appears on every public surface (footer, nav, sitemap, API docs, llms.txt and the homepage grid), wired into both the lint gate and a daily health check.
New tool: the Identifier Validator checks freight-identifier check digits — shipping container (ISO 6346), air waybill (IATA modulus-7) and IMO ship number — or parses an arbitrary string to find and validate every identifier in it. Available as GET /api/validate, the /validate page, and the MCP tool validate. Returns per identifier: type, normalised form, valid (check-digit pass/fail), expected vs actual check digit, details (container owner prefix + equipment category; AWB airline prefix + airline, reused from the existing airline dataset; IMO number) and a _source citing the scheme. Structural validation only — a valid check digit means the identifier is well-formed, not that the container/shipment/vessel exists or is active; it is not a registry or tracking lookup. The algorithms are implemented from the public schemes (ISO 6346:2022; IATA Cargo Services Conference Resolution 600a; IMO Resolution A.1215(34)); no standard text is reproduced, and they are verified against documented known-good and known-bad examples. Built to the tool-quality bar: the website MCP and npm stdio server register it via registerTool with a full output schema + structuredContent. Counts are now 20 REST tools / 21 MCP tools; the freightutils-mcp npm package adds the tool and is republished as 2.8.0.
New tool: the Emissions Calculator estimates freight transport CO2e per ISO 14083:2023 / GLEC Framework v3.2 — emissions = mass × distance × an emission-intensity factor (kgCO2e/tonne-km). Available as GET /api/emissions, the /emissions page, and the MCP tool emissions_calculator. It returns well-to-wheel AND tank-to-wheel emissions where the factor has both, the exact factor used (value, authority, edition), the tonne-km activity, and a _source citing both the ISO method and the specific open factor. Factors come from three verified open authorities — UK DEFRA/DESNZ (Open Government Licence v3.0), US EPA GHG Emission Factors Hub (US-Gov public domain) and ADEME Base Carbone (Licence Ouverte) — across road, rail, sea, air and inland waterway; GLEC is a methodology citation only and no GLEC values are redistributed. It is the first tool built to the new quality bar: the website MCP and the npm stdio server register it via registerTool with a full output schema + structuredContent (the other tools follow in a later sprint). You provide the distance — the tool does not route or geocode. Best-effort estimate using open factors, not a verified or audited carbon report. Counts are now 19 REST tools / 20 MCP tools; the freightutils-mcp npm package adds the tool and is republished as 2.7.0.
Each FreightUtils REST endpoint and MCP tool response now includes an additive "_source" block citing where the data or method comes from, assembled from the single provenance source of truth. Reference datasets carry authority, dataset/edition, source URL and licence (ADR 2025/UNECE, HS 2022/WCO via UN Comtrade PDDL, UN/LOCODE 2024-2, UK Trade Tariff/OGL v3); the calculators carry a methodology citation pointing at /methodology; and the datasets still completing a provenance audit — airline, container, ULD, Incoterms and vehicle — carry authority plus "provenance_status": "pending-verification" only, with no licence asserted. The website MCP responses add the same _source in structuredContent plus a one-line human-readable citation in the text. Purely additive: no existing field was renamed, removed or retyped, so any agent that ignores unknown fields is unaffected (still 18 REST tools / 19 MCP). Documented in the OpenAPI spec via a new SourceEnvelope schema. The freightutils-mcp npm server inherits _source automatically through the REST proxy — no package republish needed.
The published OpenAPI version label now reads 3.1.0 wherever it appears (the spec file already declared 3.1.0, but the API docs, the spec download buttons and the pricing feature list still said 3.0). Behind the scenes, tool/endpoint counts, dataset editions, per-dataset data-source citations and the "last updated" marker now all derive from a single source of truth, so they stay consistent across the API responses, the dataset pages and the docs and cannot drift apart. No tool, endpoint, schema or behaviour change — still 18 REST tools / 19 MCP.
FreightUtils now leads with one consistent positioning across every source agents and directories read — "the neutral freight reference layer for AI agents": authoritative dangerous-goods, customs, location and freight-calculation data an agent can call and cite, from primary sources (ADR 2025/UNECE, HS 2022/WCO, IATA-regulated airline prefixes), with no freight to sell and no carrier to push. Applied to the homepage title/description and Open Graph, the /api-docs intro, the OpenAPI spec description, llms.txt and llms-full.txt, the GitHub repo About, and the freightutils-mcp package (README, package.json, server.json) — republished to npm as 2.6.0 so the MCP Registry, Smithery and Glama re-sync from the corrected source. Tool counts are unchanged and still derive from the single source of truth (18 REST tools / 19 MCP). Metadata and copy only — no tool, endpoint, schema or behaviour change. @modelcontextprotocol/sdk is verified at or above the 1.26.0 release that patches CVE-2026-25536 on both surfaces.
Several MCP tools advertised a plain text input where the description promised a specific format, so malformed values passed validation and only failed (or returned nothing) further down. Each is now validated to exactly the forms it actually accepts, with a clear error, and without narrowing any feature: unit_converter now takes a known unit code (the 17 weight/volume/length units, plus chargeable_kg and freight_tonnes as conversion targets); hs_code_lookup validates a 2–6 digit code and a Roman-numeral section; incoterms_lookup takes a 3-letter Incoterm code; uk_duty_calculator validates the commodity code (6–10 digits), 2-letter origin country and the Incoterms enum; airline_lookup keeps its 2-char IATA / 3-char ICAO / 3-digit AWB-prefix / country / name union but rejects mis-shaped codes; unlocode_lookup validates the 5-char code, 2-letter country and a function_type enum (port/airport/rail/road/icd/border); and the container/ULD/vehicle selectors reject empty input while still accepting any code or slug. Applied identically to the website /api/mcp surface and the freightutils-mcp npm package (published 2.5.0) so tools/list and validation match across both. No tool, endpoint or count change (still 18 REST tools / 19 MCP). Note for npm users: unlocode_lookup’s function filter is now named function_type.
The consignment calculator can now be shared with a single link. "Copy share link" encodes the entire consignment — mode plus every line (quantity, dimensions, units, goods code and stackable flag) — into the URL itself; opening that link in any browser reopens the consignment as a clearly-labelled, read-only "Shared consignment" view that recomputes client-side, dangerous-goods flags included, with the same totals and billed-on basis the sender saw. Nothing is uploaded, saved or tracked — the consignment lives entirely in the link — and the recipient can hit "Edit a copy" to branch their own editable version. Oversized consignments that would make a broken link are declined gracefully with a prompt to send a CSV/JSON file instead. The four exports (Excel, email, CSV, JSON) now all carry each line's goods code so a pasted-back consignment is complete; the JSON keeps the canonical POST /api/consignment result shape and adds an input echo alongside it. No new tool, endpoint, URL or API change.
The website MCP endpoint (/api/mcp) now declares the prompts and resources capabilities and returns a valid empty result for keyless resources/list, prompts/list and resources/templates/list, instead of a -32601 "Method not found". This matches the freightutils-mcp npm package so both MCP surfaces behave identically when a directory crawler or strict client introspects them without an API key. The tool surface is unchanged — still 19 tools with the same enriched schemas — and tool calls still require an API key only at call time. Website surface only; no new endpoint, tool or count, and no package republish.
On the consignment calculator the results card is sticky so it stays in view while you work, but its sticky scope reached past the tool — so on scroll it stayed pinned and the "How the consignment calculator works" copy slid up behind it, reading as a card floating on top of the page. The sticky is now scoped to the tool section: the results card stays put while the tool is on screen, then scrolls away normally as the explanatory copy comes up. No layout shift, and the mobile layout (where the card is already non-sticky) is unchanged.
The multi-item consignment calculator is now a spreadsheet-style tool. Switch between sea, air and road with a single control — the chargeable-weight basis follows the mode — and add as many lines as you need, each with its own quantity, dimensions (mm/cm/m/in), weight (kg/g/t/lb) and an optional goods code. Results sit directly under the table and update instantly as you type, entirely in your browser: total volume, gross and chargeable weight, what the consignment is billed on, the mode's costing basis, and objective advisory flags including dangerous-goods presence. Type a UN number into the goods-code field and its ADR 2025 hazard class and proper shipping name appear (UN 1203 → Class 3, petrol) — resolved client-side against a slim public index, so the full dangerous-goods dataset is never downloaded to the browser. Four one-click exports round it out: copy for Excel (tab-separated rows + totals that paste straight into cells), copy a plain-text summary for email, download a CSV, or copy the canonical JSON result — the exact same object the POST /api/consignment endpoint returns. The in-browser figures match the API to the last decimal. Separately, the tool and REST-endpoint counts shown across the site now read 18 (matching the live tool registry); the MCP server still exposes 19 tools.
Internal infrastructure only — no change to the consignment calculator's inputs, outputs, schema, flags, or tool counts. The consignment dangerous-goods-presence flag's ADR lookup is now an injected dependency that defaults to the full server-side ADR 2025 dataset, so REST /api/consignment, the MCP consignment_calculator, and the npm package behave exactly as before. A new committed, re-runnable generator (scripts/build-adr-un-index.mjs) projects the canonical ADR 2025 dataset 1:1 to a slim public index at /data/adr-un-index.v1.json (2,939 entries of un_number, un_class, proper_shipping_name) so the upcoming web consignment tool can resolve UN-number presence without bundling the full ADR dataset. The index is drift-guarded in CI against its single source of truth. Reference facts only (presence + class), never a compliance statement.
The consignment calculator (POST /api/consignment and the consignment_calculator MCP tool) now shares one canonical, versioned contract. Pick a transport mode (sea, air or road) and send canonical "lines" — each with quantity, dims {l, w, h, unit} (mm/cm/m/in) and weight {value, unit} (kg/g/t/lb) — or keep sending the legacy flat "items" array, which still works unchanged. The engine returns per-line and grand totals (CBM, loading metres, volumetric and mode-specific chargeable weight; sea uses weight-or-measure at 1,000 kg/m³, air a settable IATA volumetric divisor defaulting to 6000, road loading metres against a 2.4 m trailer width) and stamps a schema_version. It also returns objective, advisory-only flags: an implausible-density check, a mode/option mismatch check, a dangerous-goods-presence note when a line carries a UN number (matched against the ADR 2025 reference — presence only, never a compliance verdict), and an ISO 6346 container / IATA air-waybill check-digit sanity check. Every response carries a best-effort disclaimer — deterministic calculations and reference data only, not regulatory, customs or dangerous-goods compliance advice; you remain responsible for classification, documentation and carrier acceptance. The canonical JSON Schema is published at /schema/consignment.v1.json. Backward compatible: previously valid calls keep working.
The remote MCP tools adr_lookup and adr_lq_eq_check now validate their inputs the same way the freightutils-mcp npm package already does: un_number must be 4 digits (optionally "UN"-prefixed), name search needs at least 2 characters, and the LQ/EQ checker takes 1–20 items. Malformed inputs now return a clear validation error instead of an empty or confusing result. Both tools also gained fuller descriptions of what they return, when to use them, and an explicit note that they return ADR reference data rather than certifying that a shipment is compliant. This brings the website MCP surface into line with the published npm package — no behaviour change for valid calls.
The three API access tiers are now spelled out identically everywhere they appear — 25 requests/day for anonymous access (per IP, no signup), 100 requests/day with a free API key, and 50,000 requests/month on Pro (£19/mo). Previously the free-key tier was missing from the /about "How this runs" section and from the pricing summary that feeds the LLM index, so the real free-key limit read differently depending on which page you landed on. Every number is verified against the live rate-limit middleware, the anonymous limit now reads from the same shared constant the enforcement path uses (so the displayed and enforced values cannot drift), and the OpenAPI spec description states all three tiers for any tool that imports it. Commercial-access enquiries now point straight to Pro pricing. A new docs/RATE-LIMIT-SEMANTICS.md records the exact enforced behaviour: per-tier limits, per-IP vs per-key bucketing, the daily/monthly renewing (not lifetime) windows, and the precise 429 response shape.
Magic-link sign-in now works reliably from corporate email accounts. Many company mail systems (Microsoft Safe Links, Mimecast, Proofpoint and similar) automatically open every link in an email to scan it for threats — and because FreightUtils sign-in links were single-use, the scanner's automated visit was using the link up before the recipient ever clicked it, producing a confusing "link expired or already used" error even well inside the 15-minute validity window. Clicking a sign-in link now lands on a lightweight confirmation page with a single "Continue to sign in" button; the link is only marked as used when that button is pressed, which automated scanners never do. The confirmation step is a plain HTML form, so it works in locked-down corporate browsers with JavaScript disabled. Double-clicking the button or using the back button after signing in no longer shows an error — if you are already signed in, you are taken straight to your account. All sign-in destinations behave exactly as before, including the verify-then-checkout flow for Pro upgrades. If your sign-in link does expire (they still last 15 minutes), the error page and the confirmation page both offer a one-click path to request a fresh one.
Root-cause fix for the customer-tier-sync family (FAULT 16/17). updateUserPlan in lib/auth/kv.ts no longer silently no-ops when there is no usable user record: a paying customer who reached Stripe checkout without a prior free account — or whose record was left as a non-JSON blob / field-dropped shape by a past manual dashboard edit (kv-encoding audit F1) — was resolving as Free with no API key, the /account page showing "key not available — contact support" and the "missing information" banner, and the self-heal backstop could not reach the record because it gates on a stripeCustomerId the blob lacked. On any active-subscription write (plan: pro) updateUserPlan now materialises a complete record — minting + linking an fu_live_ key (reusing a recoverable legacy email:<email> key first, so nothing is orphaned) — and repairs a record that lost its apiKey regardless of plan. Persistence is email-independent by construction: the persist path performs only KV writes (user: + key: mirror + email: index) and NEVER calls Resend, so a silent email failure can no longer leave a half-written record; the webhook fires notification email best-effort AFTER the write, with the existing Sentry skip-capture. Sync-only (plan: null) and downgrade (plan: free) events against a missing record stay a no-op — we never fabricate accounts. New admin repair tool scripts/admin/repair-pro-record.mjs rebuilds an absent/malformed/key-less record end-to-end (Stripe-gated, persist-first, key emailed via the standard Resend template only when newly minted). New regression suite scripts/test-provisioning-record-integrity.mjs pins: an account with an active Pro subscription resolves to tier=pro with a linked key independent of email-send outcome, plus malformed-blob and key-less repair cases. Audit: docs/audit/customer-pro-provisioning-2026-06-08.md.
GET /api/mcp/sse threw "redisUrl is required" as an unhandled rejection in prod (Sentry FREIGHTUTILS-5). mcp-handler routes three transports — mcp (streamable-HTTP), sse, and message — but sse and message need a Redis URL (REDIS_URL / KV_URL) for pub/sub that this deployment never configures, so those two endpoints threw the moment they were hit and have never worked; no real client uses them (Claude Desktop, Cursor, the MCP Registry, Smithery all point at /api/mcp/mcp). Fix: a new allowlist (lib/mcp/transport-allowlist.ts) restricts the dynamic [transport] segment to exactly mcp, validated against params.transport at the start of every exported method (GET / POST / DELETE) before any delegation — so sse, message, scanner probes, and random hex all return 404 { error: "unknown_transport", hint } and never reach the Redis-requiring code path. Those rejections are deliberately not sent to Sentry (noise, not faults). As defence in depth the delegation is also wrapped in try/catch, so any error that still escapes on the mcp transport becomes a handled 500 { error: "internal_error" } captured once with tag mcp_route_error=true. The /api/mcp alias forwards a synthetic { transport: "mcp" } context the guard reads (fix-once-mirror-everywhere). Unit + structural tests (scripts/test-mcp-transport.mjs) wired into npm run lint; smoke extended with GET /api/mcp/sse → 404, GET /api/mcp/message → 404, junk-segment 404, and a Bearer initialize regression. Server-side only — no public page, sitemap, OpenAPI, nav, or tool-count change; the freightutils-mcp npm package is untouched.
May 2026
Continuous proof that a clean MCP client can connect, authenticate, and invoke tools — so the first time it cannot, for any user, an alert fires. New GET /api/mcp/health (public, 60s edge cache, rate-limit exempt) returns status / mcp_version / tools_registered (19) / transports / data_freshness / timestamp; callable by a customer agent for self-diagnostic without burning anonymous budget. New scripts/smoke/mcp-customer-experience.mjs simulates a brand-new user end-to-end against both transports — remote streamable-HTTP runs handshake + tools/list + four tool_calls with golden-value asserts (cbm_calculator 0.96 m³, adr_lookup UN 1203 class 3, airline_lookup AWB 176 → Emirates, shipment_summary non-empty); stdio spawns npx -y freightutils-mcp@latest and verifies handshake + 19 tools registered. Wired into .github/workflows/mcp-smoke.yml — daily 06:15 UTC cron, push-to-main on MCP paths, workflow_dispatch trigger, captures a Sentry event via sentry-cli on failure. Sentry MCP-specific failure tagging now lives in lib/observability/audit.ts: every MCP request sets scope tags surface=mcp / transport=remote / stage / tool_name; explicit captureException fires on handler throws, captureMessage on 5xx and 401. PII-safe — only the existing apiKeyTail (last 8 chars) carries customer attribution. /api-docs#mcp-setup gains a First-time setup verification block with Claude Desktop / Cursor / Cline configs, the "fully quit and relaunch" instruction, the health-endpoint curl, and a 5-row troubleshooting table. Two real bugs surfaced during development (separate sprints to fix): the standalone freightutils-mcp npm package does not pass an API key through its proxied HTTP calls (so stdio users hit the 25/day anon cap regardless of Pro status); and the MCP cbm_calculator response is still in camelCase while the REST /api/cbm equivalent is snake_case. Smoke handles both bugs gracefully — stdio tool_call leg skipped via MCP_SKIP_STDIO_CALLS=1 until the package fix ships; cbm assert accepts either casing.
New /methodology page (629 words) lists every shipped dataset with its named authoritative source, update cadence, and last audit date — HS (WCO 2022), ADR 2025 (UNECE), Airlines (IATA/ICAO), UN/LOCODE (UNECE), Containers (ISO 668 + carrier specs), Vehicles (EU 96/53/EC + UK DVSA + manufacturer specs), INCOTERMS (ICC 2020), ULDs (IATA ULD Technical Manual), UK Duty (HMRC Trade Tariff). Documents the two-agent verification chain (Code transcribes; Chrome re-reads the live source URL; smoke tests + regression guards catch systematic errors) and the per-record provenance.decision_rationale audit trail. Names the limitations plainly (not a substitute for a DGSA, not customs-broker advice, real-world tare varies between build lots). Error-report path: email or GitHub issue, investigated within 48h, fixed entries logged here by date. /about page rewritten end-to-end (462 words) into first-person practitioner voice — names Soap, a UK ADR-certified freight transport planner; What-this-isn't section plainly disclaims VC-funded / chatbot wrapper / canonical-source framings. A new <CredibilityLine /> component is wired above the input on all 7 ADR tool pages (/adr, /adr-calculator, /adr/lq-eq-checker, /adr/limited-quantities, /adr/tunnel-codes, /adr/training-guide, /adr/changes-2025) and all 8 non-ADR tool pages (/hs, /airlines, /containers, /uld, /unlocode, /incoterms, /vehicles, /duty). Footer Platform column gains a Methodology link. /about is also in the main nav. Voice + endorsement scan clean (no founder-marketing tells; no false UN/IATA/ICC endorsement claims).
Google Search Console (2026-05-25 alert) flagged the new "Excluded by noindex tag" indexing reason on pages submitted via sitemap. Audit confirmed four overlap URLs — /privacy, /terms, /refund-policy, /dpa — all carry robots: noindex, follow in their metadata export (added 2026-04-30 in the legal/compliance audit commit e169a5b) but were still being emitted from app/sitemap.ts. Removed the four entries from staticRoutes and added a NOINDEX_PATHS ReadonlySet at the top of app/sitemap.ts that drives a final-stage filter on the assembled sitemap as a safety net — covers /privacy /terms /refund-policy /dpa /account /signin /login /dashboard so a future drive-by re-add of any of these still gets caught before publication. Sitemap URL count drops 9,830 → 9,826. No unintentional noindex found — all noindex pages are legitimately user-state, legal, or the /hs/code/00000000 honeypot (already excluded at its generation site). The "Blocked due to other 4xx issue" GSC reason was also audited: no URLs in the current sitemap return 4xx for valid inputs and next.config.ts redirects cover the known legacy URLs Google may still have indexed (/privacy-policy → /privacy, /tos → /terms, /data-processing-agreement → /dpa, etc.). Post-deploy IndexNow ping for the four removed URLs accelerates the deindex.
Smithery scrape verification on 2026-05-24 caught the website MCP transport advertising serverInfo.version "1.0.8" via initialize even though SITE_STATS.mcpVersion and npm both read 2.1.1. The hardcoded literal in app/api/mcp/[transport]/route.ts is replaced with SITE_STATS.mcpVersion so the wire identity tracks the same constant the rest of the site uses (/api/health, changelog, audit logs). Two drift surfaces collapsed into one; the deeper "compute from freightutils-mcp/package.json or npm registry at build time" automation stays queued in Notion. Companion docs: root README.md replaced (was stock Next.js boilerplate) with a proper surfaces table + dev quickstart + related-repos table; SoapyRED/freightutils-mcp README updated — uld_lookup count line 15 → 16, changelog backfilled with 1.1.0 / 2.0.0 / 2.1.0 / 2.1.1 entries (previously stopped at 1.0.8). Next Smithery + Glama re-scrape will pick up both surfaces.
Three small drift fixes. (1) Sitewide RSS auto-discovery — added <link rel="alternate" type="application/rss+xml" title="FreightUtils Changelog" href="/changelog.xml" /> to the root layout <head> in app/layout.tsx so RSS readers like Feedly and NetNewsWire pick up the feed from any page on the site (previously only /changelog itself advertised the feed via per-page metadata). Footer Platform column gains an explicit "RSS Feed" link to /changelog.xml for human discovery alongside the existing Changelog entry. The feed itself — 50-entry RSS 2.0 generated from lib/changelog-data.ts via app/changelog.xml/route.ts — was already live; this PR closes the discoverability gap. (2) /uld page meta description and OG image generator URL still hardcoded "15+ unit load device types" while the underlying lib/data/ulds.json is 16 entries since PR #25 (2026-05-13 rebuild) and the on-page chip already reads ${ULD_COUNT} dynamically. Bumped both hardcoded strings to "16" in app/uld/page.tsx so SERP snippets and social cards match the on-page chip and /api/uld meta.total. Reconciliation report — all other surfaces already consistent: SITE_STATS has no ULD count field; /about, /, /api-docs do not mention a count; root README.md is the stock Next.js template; no lib/api-tools-registry file exists. (3) SITE_STATS.mcpVersion in lib/constants/siteStats.ts bumped 1.0.5 → 2.1.1 to match the live freightutils-mcp npm publish; used in the /api/health response only — string stringify, no shape change. Added a TODO comment beside the constant to compute from freightutils-mcp/package.json or fetch from the npm registry at build time (queued in Notion idea vault as the staleness-automation thread; not in this PR scope).
/contact reworked to fix the "same email N times" repetition that read as amateurish. Five topic cards (general, Pro support, data corrections, security, abuse) keep their titles and descriptions, drop the per-card visible email line, and gain a single orange "Email us →" CTA with a topic-specific mailto subject. The raw contact@freightutils.com address now appears exactly once, in a dashed block below the card grid. Cards also gain category-coloured left borders (general = accent, Pro = customs blue, data = ref purple, security + abuse = DG red) so they match the rest of the site's card pattern. The Pro card optionally renders a secondary "Book a 15-min call →" button when NEXT_PUBLIC_CALENDLY_URL is set in the Vercel environment — unset by default, so flipping it on is a config-only change with no redeploy logic needed. The resolver lives at lib/contact/calendly.ts and the conditional contract is pinned by scripts/test-contact-calendly.mjs (11 assertions covering undefined / null / empty / whitespace / trimming / truthy paths). Separately: welcome email 2 ("3 things to try") now leads with the API Playground — paste your key, hit Send, see live responses with no terminal needed. The MCP-server item was dropped (covered well enough by /api-docs#mcp and the homepage MCP card); shipment-summary (now #2) and ADR lookup (now #3) stay. Subject line unchanged so the existing drip-scheduler state in KV keeps working.
/playground is now live as a single interactive surface for every endpoint in the OpenAPI spec. Pulls the endpoint list from /openapi.json at runtime; default inputs live in lib/playground/endpoint-examples.ts and are parity-enforced by scripts/test-playground-defaults.mjs (adding a new endpoint without an example breaks CI). Cards are grouped by the four design-system categories with the v1 solid-left-border pattern; response panel uses the dashed informational variant. A single API-key field at the top is gated by an explicit "Remember this key in my browser" checkbox — untick to clear, tick to persist to localStorage under fu-playground-key. The key is sent only as Authorization: Bearer on the actual API call — never logged, never in analytics, never in error reports. Displayed key in curl/JS/Python tabs is redacted to fu_live_***LAST8 unless Show full key is toggled (same last-8 convention as the apiKeyTail Sentry tag). Response panel surfaces status, latency, pretty-printed JSON, and inline upgrade prompts on 401/403/429. Deep-linkable cards (/playground#get-api-cbm) auto-expand on landing. /api-docs gets a prominent Open Playground CTA at the top plus per-endpoint "Try it →" links on all 18 endpoint headers. Wired into sitemap, nav Platform dropdown, footer Developers column, and the homepage For Developers CTA cluster. Tests: scripts/test-playground-defaults.mjs (defaults parity) and scripts/test-playground-key-persistence.mjs (localStorage tick/untick/reload).
Companion to the recursion fix (#51). New reconciliation backstop at lib/auth/reconcile.ts wired into /api/auth/me and /api/auth/whoami: when KV reads plan: free and the user has a non-null stripeCustomerId, the endpoint asks Stripe whether the customer actually has an active subscription and self-heals if so. Cooldown-gated 1h per email so identity-page refreshes do not hammer Stripe. Emits a tier_mismatch_reconciled Sentry warning on every heal — that is the alert signal that a primary webhook path failed silently again. /api/auth/whoami only invokes the backstop on the plan === free slow path so Pro keys keep the one-KV-read fast path. New manual override at scripts/admin/upgrade-user.mjs — Stripe-gated one-shot for KV writes, refuses to upgrade an unpaid account, reads back to verify both user:<email> and key:<apiKey> records. Triage runbook at docs/runbooks/customer-tier-sync.md. Phase-1 diagnosis archived at docs/audit/customer-pro-tier-sync-2026-05-21.md. FAULT 16 encoded in docs/FAULT-HISTORY-AND-PREVENTION.md.
Two observability gaps closed in response to the Saytz incident. (1) Sentry only had has_api_key:true as a boolean — no way to filter errors to a specific paying customer. The audit pipeline at lib/observability/audit.ts now sets an apiKeyTail Sentry tag with the last 8 chars of the request key (Authorization Bearer or x-api-key). Last 8 only, no fu_live_ prefix — so the message redactor at lib/sentry-redact.ts (which does not walk event.tags) leaves the value alone. The tag is OMITTED entirely for anonymous requests so Sentry does not index empty values. (2) No abandonment detection — Saytz had 4 calls on day 1 and zero since, and we only learned via the customer email. New Vercel cron at /api/cron/customer-health (daily 09:00 UTC) scans KV user:*, sums the last 7 daily-bucket counters for each Pro customer with daysSinceSignup >= 3, and emails a digest to contact@freightutils.com when any total is below 10. Manual mute via a silenced_until ISO field on the KV record. Logic is dependency-injected so the unit test exercises the full branching without hitting prod KV or sending real emails (31 assertions across 7 scenarios). Sentry alert rules to be applied via the dashboard — spec in the PR body. Runbook: docs/runbooks/customer-health-cron.md.
safeUpdateUserPlan in app/api/stripe/webhook/route.ts called itself recursively instead of the imported updateUserPlan. Every checkout.session.completed / customer.subscription.deleted event silently stack-overflowed; the surrounding try/catch ate the RangeError and the webhook still ACK'd 200 so Stripe never retried — 100% silent failure for ~24h on the Sytze upgrade. Replaced with a wrapper in lib/auth/safe-update-user-plan.ts that delegates to updateUserPlan, captures any exception to Sentry with source=stripe-webhook + stripe_event_type + email + plan tags, and re-throws so the route returns 500 and Stripe applies its standard retry backoff. Silent-eat is exactly what made the failure invisible. Added stripeSubscriptionId as an optional field on the User type in lib/auth/kv.ts and extended updateUserPlan to persist it (plus a plan: null sync-only path for transient Stripe statuses like past_due where downgrading is incorrect). Webhook now wires customer.subscription.created and customer.subscription.updated to sync subscription.id on every transition; checkout.session.completed extracts session.subscription so the field is populated at upgrade time. Unit tests pin the no-recursion regression and the User-type round-trip.
A paying customer whose KV record had been hand-edited in the Upstash dashboard (manual recovery from a separate webhook issue) triggered 500s on /account and /api/auth/me. The dashboard edit replaced the JSON value instead of merging, dropping the apiKey, email, and createdAt fields. Both call sites did user.apiKey.slice(-4) and threw TypeError: Cannot read properties of undefined. The production release SHA was unchanged through the incident — the failure was purely data-shape, not a deploy regression. Anonymous traffic and every other authenticated user were unaffected. Fix: defensive guards on both surfaces (apiKey and email accessed via nullish-fallback), the masked-key chip on /account now renders "not available — contact support" instead of crashing, both surfaces emit Sentry.captureMessage("malformed_user_record", {level:"warning"}) when a guard fires, and a new "needs attention" banner on /account tells the customer their subscription is intact while support restores the record. FAULT 17 encoded. Diagnosis at docs/incidents/2026-05-21-prod-homepage-500.md.
Free-tier signups (the top of the API funnel) have been silent on prod — Stripe-side notifications cover paid signups, but free-tier API key issuance fired no internal alert. New helper lib/email/admin-signup.ts sends a transactional email via the existing Resend integration with subject "[FreightUtils] New free signup: {email}" or "[FreightUtils] New Pro signup: {email}" to contact@freightutils.com (override via ADMIN_NOTIFICATION_EMAIL env, e.g. mcristoiu@gmail.com for direct-to-personal). Body carries: email, tier, UTC + UK timestamps, country (x-vercel-ip-country for free flows; Stripe address.country for Pro), referer (free flows only — Stripe webhook is server-to-server), source (legacy-form / magic-link-verify / stripe-checkout), and the optional use_case for legacy-form signups. No PII beyond email; no payment details, no Stripe IDs, no API key values. Three call sites wired: app/api/keys/register/route.ts (after KV insert), app/api/auth/verify/route.ts (with a pre-createUser getUser check so notifications only fire for genuine first-time signups, not returning sign-ins), and app/api/stripe/webhook/route.ts on checkout.session.completed. Non-blocking — every site uses `void notify(...).catch(...)` so a Resend or KV outage cannot delay or fail the signup ack. Kill-switch ADMIN_NOTIFICATIONS_ENABLED=false silences without a deploy. Rate-limit cap 30/h via KV bucket admin-signup-notifs:${YYYY-MM-DDTHH}; above cap console.warn and skip. Contract-verified by scripts/admin-signup-notify-test.mjs which checks the helper exports, the kill-switch boolean logic (unset → on; "true" → on; "false" → off; empty → on), every call site's void+.catch wrapping, and the rate-limit constants. Audit: docs/audit/admin-signup-notifications-2026-05-20.md.
siteStats.ts toolCount + mcpToolCount both moved 18 → 19, propagating to homepage / /about / /api-docs. FREIGHTUTILS_STRATEGY.md mcp.directory bullet updated to live v2.1.1 / 19 tools. CLAUDE.md "Distribution cascade" + STATE.md surfaces table amended to encode Chrome-verified Glama cascade: Glama auto-publishes its build artefact on successful scrape-derived build — the "Build succeeded for FreightUtils MCP Server" maintainer email is informational, no "Create a release" click required. Maintenance grade C → B confirmed post-publish. Added a "verify Glama behaviour, do not predict it" one-liner — this was the third cascade drift in three days. Historical "11" / "18" mentions in past changelog entries + audit doc forensic record left intact. External follow-ups (freightutils-mcp README + server.json + package.json descriptions, Glama dashboard description) listed with exact target text in the PR body. Audits: docs/audit/tool-count-standardise-2026-05-17.md (Phase 1 diagnosis); docs/audit/distribution-cascade-2026-05-16.md (2026-05-17 afternoon amendment).
Sprint 1 made the MCP Registry cascade work, but Chrome verification on Glama showed "Latest release: v1.0.0", Tool Definition cache stuck at 11 tools, Maintenance grade C. Diagnosed that Glama scrapes the GitHub Releases API specifically — freightutils-mcp had zero Releases ever cut. Extended the publish workflow with a final step that creates a Release on every tag push (notes pulled from freightutils-mcp CHANGELOG.md between the version heading and the next one; marked --latest=true when the version matches MCP Registry isLatest). Plus an idempotent publish guard so re-runs on already-published versions are no-ops. Backfilled four historical Releases via workflow_dispatch: v2.1.1 (Latest), v2.1.0, v2.0.0, v1.1.0. Glama re-scrape lag is days; Chrome re-verify in 3–5 days. Walkthrough: future v2.1.2 will npm publish → git push --follow-tags → publish to MCP Registry → cut GitHub Release in one workflow run. Companion PR SoapyRED/freightutils-mcp#3 MERGED. Audit amendment: docs/audit/distribution-cascade-2026-05-16.md (2026-05-17 section).
/account, /contact, /docs/{deprecation,versioning}, /dpa, /for-it, /guides, /guides/[slug], /refund-policy, /roadmap, /signin, /status all rendered <title>X — FreightUtils | FreightUtils on prod (the per-page title hardcoded "— FreightUtils" and the root layout template appended "| FreightUtils" too). Removed the per-page brand from each; the root template adds it once. Recovers ~9 SERP characters per affected page. ADR detail titles regained the "ADR 2025" edition signal — they now read "UN 1203 Petrol — ADR 2025 Class 3 PG II" instead of "UN 1203 Petrol — ADR Class 3 PG II", across ~2,347 ADR detail pages. Prevention: scripts/lint-seo-titles.mjs now walks every app/**/page.tsx and asserts no rendered title contains "FreightUtils" more than once (Rule A) + every ADR fixture title includes "ADR 2025" (Rule B); CI-gated via a new prebuild hook so Vercel runs the lint before next build. Search-performance baseline + re-crawl windows in STATE.md "Search performance baseline" section. Diagnosis + repair: docs/audit/title-template-2026-05-16.md. Encoded as FAULT 15.
The Publish to MCP Registry workflow in SoapyRED/freightutils-mcp now fires on tag push (v*) in addition to GitHub Release + manual workflow_dispatch. The version-resolve step reads from input → tag → package.json so catch-up runs work without an explicit input. MCP Registry caught up in-sprint to freightutils-mcp@2.1.1 (isLatest=true confirmed via the public Registry API). Smithery and Glama are scrape-based — they re-fetch independently with days-of-lag, no event link from the Registry; the prior internal docs claim of a Registry → Smithery/Glama cascade was wrong and is corrected. The freightutils-distribution skill plus a new CLAUDE.md "Distribution cascade" section now anchor the actual sync model. Full diagnosis + version-bump runbook at docs/audit/distribution-cascade-2026-05-16.md. awesome-mcp-servers PR #5358 updated to the 19-tool surface; Punkpeye-merge pending.
All 17 records in lib/data/vehicles-ref.json carry per-record provenance (≥2 sources each: EU Directive 96/53/EC, UK gov.uk vehicle weights & licence categories, 49 CFR Part 393 for US trailers, plus manufacturer specs from Schmitz Cargobull / Krone / Faymonville / DAF / Mercedes-Benz / VW / Ford / Wabash National) plus audited_at / verified / decision_rationale — surfaced on /api/vehicles. The customs duty calculator gained a structured DUTY_METHODOLOGY constant documenting the CIF → duty → VAT formula and Trade Tariff measure-type resolution (103 third-country, 142 preferential, 305 VAT, 695 anti-dumping, 277 restrictions), with 8 HMRC source URLs — exposed via GET /api/duty?methodology=true. A 10-code regression fixture (lib/data/duty-sample-fixture.json: 0901, 2204, 6110, 8471, 8703, 3004, 7113, 4011, 9504, 0207) anchors the live Trade Tariff API integration for future smoke-test extension. Sandbox HTTPS allowlist did not include the cited domains this audit pass — every record + the methodology flagged verified: false until live re-fetch. Gap report at docs/audit/vehicles-customs-completeness-2026-05-16.md; verification runbook at docs/audit/vehicles-customs-verification-2026-05-16.md.
New page at /hs/code/00000000 returns a 200 with a clearly-marked placeholder entry ("RESERVED FOR TESTING — NOT A REAL COMMODITY CODE", duty 0.00%); noindex,nofollow keeps honest crawlers out of the index. A hidden bait link (position:absolute; left:-9999px, tabindex=-1, aria-hidden, rel="nofollow") is seeded on /hs so DOM-parsing scrapers follow it. Hits are logged in middleware with the marker `[ScrapeGuard] HONEYPOT path=/hs/code/00000000 ip=… ua=…`, fired before the HS rate-limit check so a trap-trigger is captured even when the IP is over its 10/5min budget. No auto-block this sprint — evidence collection first. Defensive filter on app/sitemap.ts keeps the path out of sitemap.xml.
50 sample HS codes (20 top-traffic, 15 DG-adjacent in chapters 27/28/29/36/85, 10 high-value in 84/87/71/62/22, 5 coverage) now carry per-record sources / audited_at / verified / decision_rationale via a new sample-provenance file merged at getCodeDetails — surfaced on /api/hs?code=…. The other 6,890 records of the 6,940-code HS 2022 dataset are untouched. Full-dataset gap report: 100% per-field populated, 97/99 chapters present (77 + 98 absent by HS-2022 design), 0 orphan parent references. Sandbox HTTPS allowlist did not reach trade-tariff.service.gov.uk this pass; every sample record flagged verified: false until live re-fetch.
All 10 sea-container records carry ISO 6346 codes (22G1, 42G1, 45G1, L5G1, 22R1, 45R1, 22U1, 42U1, 22P1, 42P1) plus per-record sources / audited_at / verified / decision_rationale — surfaced on /api/containers. 30 LHR-weighted UN/LOCODE anchors (GBLHR, USJFK, SGSIN, HKHKG, AEDXB and others) now carry per-record provenance via a new merged anchor-provenance file — surfaced on /api/unlocode?code=…. Sandbox HTTPS allowlist did not reach primary sources this pass, so no field values mutated; every record flagged verified: false until live re-fetch.
Fixed double-branded titles (root template now appends "| FreightUtils" instead of "| FreightUtils.com"; /changelog uses title.absolute). Dark-mode toggle syncs across tabs via the storage event. /api-docs rate-limit copy normalised to "25 requests per day" across all three mentions. NewsletterCapture added to /hs/code/[code] and /hs/heading/[heading] templates so HS detail pages match the ADR-detail email-capture pattern.
Middleware now emits the User-Agent and full client IP on every 429 — never on the success/cache-hit path. Supports evidence-based firewall rule additions. UA is sanitised (control chars stripped to prevent log injection, internal quotes replaced with apostrophes, truncated to 200 chars, falls back to `ua=empty` for null/whitespace). Log line format converted to space-separated `key=value` pairs for grep/awk parsing. IP resolution unchanged.
Scraper-target page routes now served from Vercel edge cache with explicit Cache-Control (s-maxage=86400, SWR=7d). Warm-cache hits skip page render — cold-serve response unchanged. ISR was already enabled on both routes; this surfaces the cache strategy in next.config.ts so it is tunable and visible via curl -I. The middleware-layer ScrapeGuard rate limiter continues to 429 the active 216.* scraper as designed.
Redis errors in the ScrapeGuard middleware now log at most once per minute per error class instead of per request, preventing the Sentry-event flood we saw during Upstash quota exhaustion on 2026-05-14 (~1K events in 30 min). Fail-open behaviour preserved — Redis failures still allow the request through.
25 anchor UN numbers (55 variant rows) now carry per-record `provenance.sources` URLs, `decision_rationale`, and `audited_at` timestamps. `/api/adr` response surfaces all three fields. Full-dataset gap audit at docs/audit/adr-completeness-2026-05-13.md.
Fixes a long-standing data fault where IATA prefix CV was misattributed; prefix index repaired so cargo-only filters now resolve cleanly. 35-airline anchor set gained per-record provenance. Cargolux, Cathay Cargo, Emirates SkyCargo, and others all verified.
7 verified, 9 variant-decision-needed. Magnitude corrections: AKH height 163→114 cm (LD3-45 is narrowbody-only); PLA dims/MGW rebased from PAJ-class to actual P1P (318×153×163, MGW 3175); ALP re-labelled LD7→LD11; RAP rewritten from passive blanket to active LD-9 thermal container (Cargolux/DSV/ACL/Wikipedia all confirm IATA registration); PMC-Q7 tall variant (300 cm) split into a separate entry. New provenance schema with `audited_at`, `chrome_verified_at`, `soap_signoff_at` timestamps. Predecessor PR #24 (PAG/PGA swap fix) consolidated here.
New /signin page and `/api/auth/public-signin` endpoint. Reveals nothing to anonymous callers about which email addresses are registered; `return_to` query parameter whitelisted to FreightUtils paths only. Nav now flips between "Sign in" and "Account" based on session-cookie identity.
Logged-in customers can view subscription status, billing history, and trigger Stripe customer-portal flows. Anonymous visitors to /account redirect to /signin. Session-cookie identity (`/api/identity/whoami`) + customer-portal endpoints land in the same PR.
Full subscription cancellation flow: customer-initiated cancel through Stripe portal, immediate admin email, customer confirmation email with UK Consumer Contracts Regulations 2013 Reg 37 waiver wording (no 14-day right to cancel after services are partly performed at customer request).
April 2026
New /contact page (mailto + form), /dpa page (Data Processing Agreement listing sub-processors: Vercel, Cloudflare, Stripe, Loops, Sentry). robots.txt and redirects audited and tightened in the same commit.
Fixes B029: an empty `X-API-Key` header was previously bucketed as anonymous instead of unauthenticated. n8n/Zapier integrations that test with an empty key during credential setup now correctly see 401 and surface the error to the user.
/api/mcp tool input schemas migrated from camelCase to snake_case to mirror the freightutils-mcp npm package v2.0.0. JSON contract stays consistent across the website MCP surface and the standalone npm/Zapier/n8n integrations.
Added `lib/seo/page-metadata.ts` builders for the four templated page families. Titles now ≤60 chars with the primary search keyword in the first 50; descriptions ≤155 chars with "free" + ("no login" / "updated YYYY"). Permanent `lint:seo-titles` build-time check enforces the pattern. Fixes the US CTR gap (0.07% vs UK 0.94%) flagged by GSC.
Fixes prefix misattributions on ZP and GS, backfills missing AWB prefix data for AV (Avianca) and UK (Air Bridge Cargo Airlines). 5 newly-active cargo airlines flagged in the dataset.
Every `app/api/**/route.ts` now wrapped with `withAuditRest` (or `withAuditMcp` for the MCP transport route). Emits a single structured `[fu-audit]` line per request with path, method, status, tier, latency. Privacy contract: NO request bodies, NO API key values, NO IPs, NO emails ever logged. Build-time `lint:audit` check fails if any new route forgets to wrap.
New endpoint that validates an API key against the registry and returns tier (free/pro) plus the key prefix (never the full key). Replaces /api/health as the credential-test target for n8n and Zapier integrations — fixes B028, where /health returned 200 to any caller and silently green-ticked invalid keys.
Fixes B003: /api/health reported 18 tools while /api/tools listed 17. Both endpoints now derive from a single registry; the missing entry (ADR LQ/EQ Check, shipped 2026-04-09) was added back. Adding a new tool propagates to both endpoints automatically.
Six endpoints migrated from camelCase to snake_case response fields (/api/unlocode, /api/uld, /api/containers, /api/vehicles, /api/consignment, /api/duty). Site-wide consistency with the other thirteen endpoints. Clean break — no dual-output. Full field rename table in CHANGELOG.md.
Five new platform pages plus a `/changelog.xml` RSS feed. Roadmap is GitHub-seeded so public-facing items stay in sync with the internal sprint plan. Versioning and deprecation policy documents codify API stability commitments.
Check Limited Quantity and Excepted Quantity eligibility for mixed dangerous goods consignments. Multi-item input with per-item breakdown, green/red/amber verdicts. POST /api/adr/lq-check.
15 unit load device types — AKE (LD3), PMC, PLA, and more. Dimensions, weights, volume, aircraft compatibility, and deck positions. Free REST API at /api/uld.
17 road freight vehicle and trailer types — curtainsiders, rigids, vans, US 53ft. Dimensions, payload limits, pallet capacity. Free REST API at /api/vehicles.
POST /api/shipment/summary chains CBM, LDM, chargeable weight, ADR compliance, and UK duty estimation into one response. Accepts road/air/sea/multimodal modes.
Consignment calculator and chargeable weight calculator now support sea freight mode with W/M (Weight or Measure) at 1 CBM = 1 revenue tonne.
Canonical FreightUtils Shipment schema defined — the foundational data model for all composite endpoints.
Estimate UK import duty and VAT for any commodity code using live GOV.UK Trade Tariff data. Supports Incoterm-adjusted CIF, preferential rate flagging, and cross-tool HS code workflow.
116,000+ transport locations from UNECE UN/LOCODE 2024-2. Searchable by name, code, country, and function type (port, airport, rail, road, ICD, border).
Calculate total CBM, weight, LDM, and chargeable weight across mixed items. Available as web tool, REST API, and MCP tool.
FreightUtils MCP Server v1.0.1 now on registry.modelcontextprotocol.io. Also listed on Smithery.ai, mcp.so, and 5+ directories.
Connect AI agents to all FreightUtils tools at freightutils.com/api/mcp/mcp. Streamable HTTP transport via mcp-handler.
Courtesy rate limit changed from 1,000/hour to 100 requests per day per IP across all API endpoints. Real enforcement via Vercel KV middleware.
Calculate ADR exemption thresholds for mixed hazardous loads. Supports multi-substance mixed-load calculation.
2,939 dangerous goods entries from UNECE ADR 2025 edition (licensed from Labeline.com).
6,940 codes from WCO HS 2022 nomenclature with full section/chapter/heading hierarchy.
6,352 entries with IATA/ICAO codes and AWB prefixes. Cargo-only default view with 390 cargo airlines.
March 2026
11 free freight tools with open REST APIs. LDM, CBM, chargeable weight, pallet fitting, container capacity, unit converter, ADR, airlines, INCOTERMS, HS codes.
Full OpenAPI spec covering all endpoints. Compatible with Swagger and Postman import.
All 11 Incoterms with seller/buyer responsibilities, risk transfer, cost transfer, and insurance details.
Dates are approximate. Data sourced from UNECE, WCO, ICC, IATA. For corrections or data issues contact contact@freightutils.com.