Skip to main content
Additive changes may ship without notice; anything that can break an integration is listed here first.

2026-11-20

  • Changed: Create Records refuses an archived facility with 409 FACILITY_ARCHIVED and a deleted one with 403 FACILITY_NOT_BELONG_TO_ORGANIZATION, as the file importer already does; nothing is deleted or created. Saving again a file whose records an archived facility already holds still replaces them. A facility of another organization answers as before.
  • Added: a facility column value that names an archived facility of the organization (or of a subsidiary it imports into) says so: its entry of Get Unique Values carries rejection_reason: "facility_archived", and its rows’ error in Validate is the rule type_category_facility_archived (with facility_name) instead of type_category. Such a value is still rejected, as before. A context facility_id that is archived still answers 404 FACILITY_NOT_FOUND.
  • Added: an import template (see Get a template) has context_fields: each value for the whole upload that no column carries (facility_id, transport_direction, matching_method, …), with whether it is required, whether the submit copies it into the rows (submit) and where its values come from (options_source, served by Get Options Source, or fixed options). A context value is accepted if and only if it is one of those. required_context_fields and submit_context_fields stay, derived from it.
  • Changed: a context facility_id of a wastes import is checked against the wastes’ facility options (facility_consumption_facilities): an archived facility answers 404 FACILITY_NOT_FOUND, as one that does not exist. So does a waste water treatment import’s against waste_water_facilities: a facility that is not a waste water plant, or a plant outside the waste water write perimeter, is refused when the session is opened instead of blocking every row. The facility options (and the facility column of wastes and facility consumption files) no longer list archived facilities or those of a disabled subsidiary.
  • Changed: a context value sent with Submit that the template does not take (a target block outside the LCA, an unknown matching_method) is refused before the session is claimed: the session stays as it was.
  • Added: Get Options Source takes project_id, for the project-scoped sra_project_suppliers source: the supplier column of the SRA “Upload emails” template is now a category of the project’s suppliers, checked in the value step.
  • Changed: an import session of the sra_suppliers template needs a project_id, as sra_supplier_emails already did: without one it answers 400 MISSING_PROJECT_ID when it opens, instead of failing at submit. Both templates only import into a Supplier Risk Assessment project: another kind of project answers 400 PROJECT_NOT_SUPPLIER_RISK.
  • Added: Get LCA Dashboard returns each co-product’s allocated impact per category in meta.output_products[].impacts (total and per_unit) and the whole system’s in meta.system_impacts. Both are the same in every view.
  • Added: an import template (see Get a template) has capabilities, always present: custom_columns says whether unmapped file headers may be declared as custom columns on it, exactly what the custom-columns endpoints accept.
  • Added: a template column whose dropdown takes a parameter from its template says so in options_params (see Get a template): the facility consumption templates send their invoice_type ({"invoice_type": "water"} on facility_consumption_water) on the supplier, supplier name and custom emission factor columns, and the district template on its unit column too. Send them with Get Options Source.
  • Added: Get Import Status returns context_requirements: the requirement of each template column the upload’s page makes optional (a column the page fills when the file maps none).
  • Added: each column of Get Unique Values says whether nothing in it needs a person (settles_on_its_own): a column that creates records whose values are all exact matches, empty cells or new records never left out before.
  • Added: GET /v2/imports/{import_id}/status returns date_ambiguity and shape_warnings for a validated session, as Validate returns them, so a client resuming a session does not validate again to get them. date_ambiguity follows the session’s rows (a chosen format, a corrected cell or a deleted row clears its cells); shape_warnings is null for a session validated before this date.
  • Changed: a truncated column of Get Unique Values (more distinct values than the step returns) always says settles_on_its_own: false, even when every value returned would settle: the values past the cap were not looked at. A column that creates records and has no values says true.
  • Changed: the purchases import templates (purchases, purchases_custom_ef, purchases_custom_ef_multidb) accept negative amounts (returns, credit notes) whatever the session was opened with: a negative amount is a negative_amount_warning, never the min_value error it was without negative_purchases=true. The negative_purchases field of Create an import session, Open an import session and validate is still accepted and no longer needed. Leia imports a spreadsheet with a credit note instead of stopping with validation_errors.
  • Changed: Leia stops a spreadsheet where the importer’s columns step would: on every column whose requirement has must_assign, not only the required ones. A transport_routes file with no column or fixed value for electric or refrigerated now stops with import_state.reason (and auto_import.reason) required_field_unmapped:electric,refrigerated, on step columns, instead of going on without them.

2026-11-19

  • Breaking: POST /v1/invoices and PUT /v1/invoices/{invoice_id} reject a heat invoice in a stationary-combustion unit its fuel is not calculated in (one not among the fuel’s units in List Stationary Fuels, such as anthracite in litres) with 422 INVALID_UNIT. It used to be accepted and end in error, uncalculated. With a custom emission factor the unit is not checked.
  • Added: Get Unique Values says what validation does with each value left without an option: if_unassigned on every value (rejected, kept_as_text, creates_record or ignored) and, on a column whose values can become new records, creates_record_type (intensity_metric). A code the waste catalogue lacks ("15 01 99") is rejected even though the column keeps descriptions as typed.
  • Added: every column of an import template (see Get a template) has a requirement: its level (required, conditional or optional), must_assign (the column needs a file column or a fixed value before the rows are read) and, for a conditional column, a hint code. It comes from what the column’s validation rules do with an empty cell; required is unchanged.
  • Added: a spreadsheet uploaded again while Leia imports spreadsheets for the organization is no longer held for the duplicate review: it is classified, and Leia stops before importing it with import_state.reason (and auto_import.reason) duplicate_file while the file it copies is still there and its rows were not imported (already_submitted when they were), on step read. Opening it is no answer; confirming its mapping is. Its duplicate_verdict.status is the new flagged until then, dismissed after. Organizations without Leia, and any other file, are held as before (suspected).
  • Added: Open an import session answers 409 IMPORT_SESSION_LEIA_WORKING, with nothing read or changed, while Leia is about to import the file and has not read it yet.
  • Changed: a waste’s LER code now decides hazardous over a stated value (see Get Waste): a waste with a LER code is hazardous when the code ends in *, whatever was sent, and hazardous_source is ler. A waste priced with a custom emission factor and no LER code has hazardous null, whatever was sent (the custom factor’s own hazardousness is no longer read). A stated hazardous applies only to a waste with neither, as in the app’s form, which locks the field once a LER code or a custom factor is chosen.
  • Added: the wastes import template takes the quantity’s unit (shown as required, a dropdown): kg or t (a waste in tonnes is stored in tonnes), or m3 only for a row priced with a custom emission factor per m3, which takes only m3 (waste_unit_m3_requires_custom_ef, waste_custom_ef_quantity_not_in_factor_unit). An empty cell or a file without the column is kg, and a quantity header in tonnes is converted to kg as before; a stated unit is read in it, and rejected when the quantity header names another (waste_unit_contradicts_quantity_header). A hazardous the LER code or the custom factor decides is warned (waste_hazardous_ignored), not rejected.
  • Added: the wastes import template takes the waste manager (provider_name) and the carrier (transporter_name), optional.
  • Changed: the wastes import template has a description column apart from ler_code: a row gives the waste with a description, a LER code or both (with both, and a treatment given as an R/D code or not at all, the code prices it and the text is its description; a treatment given as text sends the whole row to be classified from its text, as before); a row with neither is rejected on both cells (waste_ler_or_description). ler_code still accepts text, as before.
  • Added: a custom dataset shared with subsidiaries can be filled for the whole group from one file. Its import template (see Get a template) gets an Organization column, and each row lands in the copy of the organization it names (by id, company name, trade name, VAT or one of its internal codes; a blank cell keeps the row in your own dataset). A row that names an organization you cannot write to (one outside your role’s organizations included), or one the dataset is not shared with, is refused (organization_not_receiving_category), also when the dataset stops being shared while the file is in progress. Files uploaded through the data assistant read the same column. A dataset with its own field keyed organization_id keeps that field as data.
  • Added: GET /v2/elastic-data/native-entities/{entity}/rows takes organization_id: the rows (facilities, vehicles…) of a group organization you can write to, so a holding filling a subsidiary’s record picks that subsidiary’s. Any other organization is a 403, one outside your role’s organizations included, as is a Viewer membership.
  • Changed: your role’s organizations bound what you write for the rest of your group. Creating a record for an organization outside them (POST /v2/elastic-data/datasets/{dataset_id}/records with organization_id) answers 403, the same answer as for one outside the group you can write to; editing or deleting a record of such an organization’s copy of a shared dataset (PATCH / DELETE /v2/elastic-data/datasets/{dataset_id}/records/{record_id}) answers 404, as for a dataset you cannot reach.
  • Changed: in GET /v2/elastic-data/datasets and a project’s datasets (GET /v2/elastic-data/projects/{project_id}/datasets, and the answer of PUT on the same path), the record_count of a dataset shared with subsidiaries counts the records of every copy your role can read, not only your own.
  • Added: an intensity measure of a dataset query can divide by a numeric column of one of your categories with denominator (dataset, metric, start_field, end_field or line_period, per, show_quantity), instead of an intensity_metric_group_id. The category is aligned with the rows by facility, organization and period, so each site and month is divided by its own quantity. A cell is null unless the category’s lines cover its period exactly: lines are not prorated. show_quantity: true adds a column with the quantity each cell was divided by. Such a measure only divides (operation other than divide is a 422). Hierarchy rows, attributes of the facility or the company, custom groups over facilities or companies, dimensions the category cannot align on and a workforce *_financial measure as the numerator are refused with 400 DATASET_QUERY_INVALID.
  • Added: dataset query responses carry denominator_issues: why a category divisor’s lines do not count or leave a cell blank, each with a stable code (gap, overlap, missing_period, outside_window…), the facility, company, period and dates concerned, and a count (see Query datasets). Empty without a category divisor.
  • Changed: water_consumption and fuel_consumption now reach the facility axis in GET /v1/datasets/{key}/joinable, so they can be grouped by facility next to the other site datasets and divided by a category’s quantity per facility. Fuels are part of energy: the energy/fuel_consumption pair is marked overlaps: true (don’t add their measures), as are fuel_consumption with invoices and vehicle_consumptions.
  • Changed: deleting a file that filled a custom dataset shared with subsidiaries for the whole group also removes the rows it created in each subsidiary’s copy, not only those in your own dataset.

2026-11-18

  • Added: import rows carry scoped_options (see Get Import Rows and Patch Rows): per column whose options depend on the row, the values that row may take. In vehicle_consumptions, a row’s fuel is offered among the fuels its vehicle declares, and a fuel text that names one of them (“Diesel” on a b7 vehicle) is adapted to it with an options_scope_remapped warning.
  • New: wastes take hazardous (true / false), for a waste whose LER code does not say it or that has none, e.g. a declaration from outside the EU: in Create Wastes (v2), Create Waste and Update Waste (null removes it), in the wastes import template (an optional yes/no column) and read from waste documents. Get Waste and List Wastes return hazardous and hazardous_source: the stated value (user or document), else the LER code’s (ler: a code ending in * is hazardous), else null.
  • Fixed: holding_weights: "as_of_period" on dataset queries did not re-weight a fund’s investees and returned today’s participations (see Query datasets). It now re-weights every participation recorded as a fund investment.
  • Added: holding_weights on dataset queries (see Query datasets). With consolidation: "financial", "as_of_period" weighs group funds with the investment values on record at end_date instead of today’s, so a past period keeps the participation it had then. Defaults to "current" (unchanged behaviour); as_of_period without financial consolidation, an explicit end_date or a dataset with a period window is a 422.
  • New: the ms_nonconformity_types import template (see Get a template) creates the organization’s nonconformity types: one row per type with label_es, label_en, code_prefix (10 characters at most, stored in capitals) and is_active (Yes/No), all required. A type the organization already has (by its English name) is refused (ms_nonconformity_type_not_duplicate), as is a second row of the file with it (ms_nonconformity_types_unique_in_file), an English name without a letter or a digit (ms_nonconformity_type_name_without_letters) and a name or prefix over its length (ms_nonconformity_type_max_length).
  • Changed: the avoided_emissions import template (see Get a template) takes total_emissions with up to six decimals (it took two): they are stored as sent, never rounded. More decimals are still refused (avoided_emission_total_emissions_scale).
  • Changed: Create Facilities (v2) no longer fills defaults. A record without categories creates a facility with none ([]), where it got heat, electricity, water and recharge; a surface_area without surface_area_unit_id is rejected with SCHEMA_INVALID, where it was stored in square meters. Send the unit with every surface. Facilities created before keep what they have.
  • Changed: a logistic hub’s category is required and must be a category of List HOCs. Create Logistic Hubs (v2) no longer defaults it to transshipment_ambient: a record without it is rejected with SCHEMA_INVALID. Create Logistic Hub requires it too and rejects an unknown one with 422, and so does Update Logistic Hub when it changes it (the hub’s own category sent back unchanged is kept, even an older one). An owned hub created through Create Logistic Hub whose address gives no country takes its facility’s country, where it was left empty.
  • Changed: a waste created without total_km_to_waste_center (Create Wastes (v2), Create Waste, the waste imports and Create Records From File Reading) stores it as null, and the waste lists and details answer null, where they answered 0: the distance is unknown, not zero. A 0 you send is kept as 0. The emissions do not change: without a distance there is no transport, as before. Wastes created before keep their 0. Read the field as number | null.
  • Changed: a waste of 0 is rejected everywhere, as Create Wastes (v2) and Create Waste already did. Create Records From File Reading rejects a waste reading whose material_quantity is missing, 0 or less with 422 QUANTITY_MUST_BE_POSITIVE, and the deprecated POST /api/v1/wastes rejects a quantity of 0 with 422 (it only rejected negative ones).
  • Docs: corrections to the logistics v2 pages, no behaviour change. Create Logistics Requests (v2): the first curl example now sends the required client; a 429 can come from the 10-in-flight cap or from the per-minute limit (600 per API key and per source IP, each with bursts of 900); a tkm you send is kept only when the leg arrives with a distance (not an air distance of 95 km or less) and its calculation fails, scaled to the adjusted distance; a calculated leg always has the tkm of its load and distance. Errors: SCHEMA_INVALID no longer says logistics ignores unknown fields, drops NUL characters or skips the number bounds; it applies all three, as every v2 record does. Create Logistic Hubs (v2) and the logistics overview: a leg’s hub_name is compared without case, accents and outer spaces, with spaces, hyphens and underscores alike, not exactly. List TOCs and the pages that show TOC names: toc values use _ (van_3_5_t_diesel), the response has id, wtw can be null, and the example shows real values. Get Ingest Job: a timed-out logistics create can be resent (legs are upserted). The packages and delete pages list the 429 (and PLAN_FEATURE_NOT_INCLUDED on packages).
  • Changed: the social_data import template (see Get a template) asks an organization whose holding has a fund asking it PAI extra questions for those answers too, as the Social and governance record sheet asks them: one required column per question (pai_<code>; pai_<code>_1 and pai_<code>_2 for a two-number question), stored as the record sheet stores them. Other organizations get the template as before. The amounts take five decimals at most, instead of two (the social_data_max_two_decimals rule is now social_data_max_decimals), and so does POST /social_data.

2026-11-17

  • New (Move Transport Routes): POST /v1/transports/move (by ids, up to 1,000) and POST /v1/transports/move-by-filters (the list’s filters and filter_hash) move transport routes to another organization of the group, keeping their ids, sections, source file and custom-column values, and recalculate them there. PATCH /v1/transports/{id} takes the same organization_id for one route, and its attachment_file_id is no longer required: each field is acted on when sent.
  • Added: the custom_ef_record_id of Create Records From File Reading, at the top level or on each reading, also prices the electricity, combustion, water, recharge, district heating and cooling invoices it creates (it priced wastes only), as it prices an invoice created one by one. A supplier_id sent with it is kept. New code: PROCESS_CUSTOM_EF_RECORD_ON_LINES, for a record sent for a process reading. Removed code: CUSTOM_EF_RECORD_NOT_SUPPORTED (a record for a reading other than wastes), as those readings now take it.
  • Changed (emissions): a custom_ef_record_id prices every year an invoice’s period touches, in every channel (the app, POST /v2/invoices/* and Create Records From File Reading): a period over three calendar years took the record for its first and last years only, and its middle year fell back to the default factor.
  • Changed: Create Electricity Invoices (v2) takes supplier_id together with custom_ef_record_id (it answered SCHEMA_INVALID): the custom record prices the market-based part and the supplier is kept on the invoice. Replace an Invoice no longer answers CANNOT_PROVIDE_BOTH_SUPPLIER_ID_AND_CUSTOM_EMISSION_FACTOR_ID for an electricity invoice in Spain that carries both.
  • Changed: the deprecated POST /api/v1/wastes answers 422 CUSTOM_EMISSION_FACTOR_NOT_FOUND_OR_DISABLED when its legacy custom_emission_factor_id names a group the waste’s organization may not use (one of another organization outside its holding, or a disabled one): it only checked that the group existed. The groups of the organization and of its parent organizations keep working.
  • Changed: a file uploaded without the Data Importer (the CSV of a category’s bulk upload) whose custom_ef_record_id names a custom emission factor the row’s organization cannot use, or is not an id, ends that row in error instead of calculating it with the standard factor: transport routes (row error custom_ef_record_invalid), vehicles and employee commuting. The records of the organization’s holding shared with it keep working. A malformed id no longer stops a transport file whose routes were already stored.
  • Changed: the purchases, purchases_custom_ef and purchases_custom_ef_multidb import templates (see Get a template) require expense_type: it decides whether a purchase is reported as a purchased good or service (opex) or as a capital good (capex), and an empty cell was read as opex. An empty recycled is no longer read as 0: it is left empty (it changes no emissions). The purchase endpoints (/v1/purchases, POST /v2/purchases) do not change.
  • Changed: the transport_routes import template (see Get a template) no longer fills in what changes a shipment’s emissions. electric and refrigerated are required wherever the transport offers a choice (transport_route_flag_required): an empty cell read as no; a transport that takes one value for it (a truck is never electric, a car never refrigerated) may leave it empty. unit is required with quantity_kg (required_with): an empty cell read as kilograms. The transport endpoints do not change.
  • Breaking (Rate Limits): the v2 logistics endpoints (POST /v2/logistics/requests, POST /v2/logistics/requests/delete, GET /v2/logistics/packages) count against the account-wide limit and the per-IP ceiling (600 requests / minute, bursts of 900) like every other v2 endpoint: over it they answer 429, and their responses carry the rate limit headers. They were excluded, so only POST /v2/logistics/requests had a limit: 10 concurrent requests, which it keeps. The v1 logistics endpoints stay excluded.
  • Added: an electricity item of Update File Reading takes self_consumption (boolean, false when absent): the invoice Create Records From File Reading creates from it is electricity the facility generated and consumed itself, as self_consumption on an invoice created one by one. Not together with a custom emission factor: Create Records From File Reading answers SELF_CONSUMPTION_WITH_CUSTOM_EMISSION_FACTOR. In Spain the item needs no supplier.
  • Changed: Create Records From File Reading answers 422 WASTE_LER_OR_DESCRIPTION_REQUIRED for a wastes item with neither a ler_code nor a waste_type, as a waste created one by one does: it was saved waiting for a factor match that could not happen.
  • Changed (emissions): logistics distances follow GLEC’s distance adjustment in every channel (the app, the v1 logistics endpoints, Create Logistics Requests (v2) and file uploads):
    • A distance Dcycle calculates from the addresses is the shortest feasible distance and is used as is. It was multiplied by 0.95 (road) and 0.85 (maritime): those legs now count about 5.3 % (road) and 17.6 % (maritime) more tonne-kilometres and emissions.
    • A distance_km you send is read as the actual distance travelled (driven, sailed or flown) and adjusted to the distance GLEC calculates with: divided by 1.05 (road) and 1.15 (maritime), 95 km less for air (a flight covers its great circle distance plus 95 km), as sent for rail. It was used as sent, except in file uploads, which multiplied it by 0.95 and 0.85.
    • New field client_distance_km on the legs you read (GET /v1/logistics/requests and GET /v1/logistics/requests/{id}): the figure you sent, which distance_km is always derived from (null when Dcycle calculated the distance, and on legs stored before this change). Never send distance_km back as a new distance: it would be adjusted again.
    • New per-leg error AIR_DISTANCE_NOT_FLOWN: an air distance_km of 95 km or less cannot be a flown distance. In Create Logistics Requests (v2) that leg fails in the job (the rest of the batch is calculated) and keeps no distance nor tonne-kilometres; in the v1 bulk that record fails; in the v1 single create and edits it answers 422; in file uploads that row errors (ERR_PROCESSING_AIR_DISTANCE_NOT_FLOWN, in the distance_km column); the legacy POST /api/v1/logistics/shipment calculator answers 422 with its usual {"message"}, starting with the code.
    • Editing a leg (PATCH /v1/logistics/requests/{id} and the app’s record sheet): a new distance_km replaces client_distance_km and is adjusted; the stored one sent back (either field) changes nothing. A change of transport mode derives the distance again from client_distance_km; a leg without one (calculated by Dcycle, or stored before this change) keeps its distance through the PATCH, and the record sheet routes it again when it has both addresses.
    • Create Logistics Requests (v2) calculates tonne-kilometres from the load and the distance: a tkm sent there is not used, as before. In the v1 endpoints a tkm you send is scaled like its distance.
    • Legs already stored keep their distance until they are sent again, or edited with another distance or mode.
  • Changed (Delete Logistics Requests (v2)): a leg sent again or edited after the delete job was created is kept instead of deleted, and its item ends failed with the new code UPDATED_AFTER_DELETE_REQUESTED. Wait for the delete job to finish before you send legs again (see Deleting and sending again). Calculating a leg no longer changes its updated_at: it says when the leg was last sent or edited.
  • Fixed (Create Logistics Requests (v2)): a leg deleted while its chunk was being calculated was reported succeeded; its item now ends failed with NOT_FOUND, as one deleted before the calculation does.
  • Added (Get Logistics Packages (v2)): read packages by the package_key you sent with their legs (?package_key=…&package_key=…, 1 to 100), instead of first looking up their Dcycle ids in List Packages. Send id or package_key, not both. Each package comes once, in the order of the keys; a key with no package in your organization is left out of items. Requests by id answer as before.
  • Added (Create Water Invoices (v2), Create District Heating Invoices (v2), Create District Cooling Invoices (v2)): an optional supplier_name, the supplier as written on the bill. It is matched against the suppliers of that type the facility’s organization can see, ignoring case, accents and punctuation, and created for the organization when it has none of that name. On district invoices it is an alternative to supplier_id (not both). The facility_consumption_water and facility_consumption_heat_cooling import templates (see Get a template) take the same: their supplier column is now supplier_name (it was supplier_id on the district template, which is still accepted as a mapping target), a supplier name from the dropdown (facility_consumption_supplier_names options) or any name, found or created for the facility’s organization and of the row’s type. A supplier id in that column stands for that supplier’s name. The water template also takes gal rows end to end (they passed validation and were then dropped).
  • Changed: a logistic hub’s name is compared without case, accents and spaces at either end, and with spaces and hyphens and underscores alike, everywhere a hub is named: hub_name and destination in Create Logistics Requests (v2), hub_id in the v1 shipments endpoints, and the hub name of Create Logistic Hubs (v2) and the import templates. HUB-MAD-001 now finds the hub hub mad 001, and a new hub whose name differs from an existing one only in that way is refused (HUB_NAME_ALREADY_EXISTS) or, where shipments create their hub, the shipment is linked to the existing one. An organization’s hub names are unique in this sense; no existing hubs were affected.
  • Changed: a business travel’s travel_number has no limit beyond 1 or more: the cap of 100 trips a day of its period and of 10,000 a record is gone from Create Business Travels (v2), /v1/business-travels (create and update) and the business_travels import template, where the business_travel_number_within_date_range rule is removed. Nothing that was accepted before is refused now.
  • Changed: the business_travels import template (see Get a template) requires passenger_number for aircraft, bus, ferry, metro, train and trolleybus (required_for_values): their emissions count every passenger, and an empty cell was read as one passenger. For a car or a motorbike it stays optional, as it does not change their emissions. The business travel endpoints do not change.
  • Changed: the hotel_stays import template (see Get a template) requires rooms(optional): the rooms multiply a stay’s emissions, and an empty cell was read as one room. The column keeps its key. A new optional custom_ef_record_id column takes one of your custom emission factor records, configured for hotel stays and per room-night, to calculate the stay with instead of its country’s factor; any other record is refused (hotel_stay_custom_ef_multidb_record_invalid). The CSV upload of /v1/hotel-stays (deprecated) reads the same optional custom_ef_record_id column: a stay naming a record it cannot use ends in error, never with its country’s factor. The other hotel stay endpoints (/v1/hotel-stays and POST /v2/hotel-stays) do not change.

2026-11-16

  • Added (Get Unique Values): the organization, facility, country and currency columns remember a confirmed value across templates. A value your organization confirmed in one template is offered as suggested (resolution_origin: "history_restored") in the other templates’ columns of the same kind, when they all agree, with the new field history_template_id naming the template it came from.
  • Breaking (Create Logistics Requests (v2)): a record follows the contract of every other v2 bulk, so a mistake is a 422 instead of a leg calculated with a value you did not intend. A record is refused (SCHEMA_INVALID) when:
    • shipment_date is missing or after tomorrow in UTC (it defaulted to the upload date, and a resend without it moved the leg to that day);
    • movement_id or client is missing or blank (client defaulted to default_client);
    • load_unit is not kg, t, ton, pallet, pallets, teu or feu (any other value was read as kg);
    • a number is NaN, infinite or 10^12 or more, or distance_km is over 500,000;
    • it has a field the record does not define (it was ignored, so a misspelled field was lost silently);
    • origin or destination has a NUL character (it was dropped).
    A shipment_date outside 1970 to next year, refused before as ACTIVITY_DATE_OUT_OF_RANGE, is now refused as SCHEMA_INVALID with the type activity_date_out_of_range. A toc must now be a TOC exactly, once case, spaces and dots are normalised: a text that only contains one, or a toc that matches nothing when you also send category, is TOC_NOT_FOUND (before, the first TOC contained in the text, or the category’s generic average, was used without an error). The business key fields (movement_id, movement_stretch, movement_stage, client, package_key) are trimmed, as the file import trims them.
  • Changed: the logistics_recharges import template (see Get a template) has a required unit column: the unit of the quantity, which must be the fuel’s (litre_(l), or kilogram_(kg) for compressed and liquefied natural gas, bio-LNG and HVO). A recharge’s quantity is stored in its fuel’s unit with no conversion, so a different unit is now refused (logistics_recharge_unit_matches_fuel) instead of being read as the fuel’s. The recharge endpoints do not change.
  • Changed: the logistics_requests import template (see Get a template) no longer fills in what changes a shipment’s emissions: cleaning and subcontractor are required (they defaulted to No and Yes), and load_unit and load_factor are required when the row has a load (they defaulted to kg and 1). A row with no load is still valid: it is calculated with its TOC’s typical load, and validation now reports it as a logistics_load_default_warning warning. subcontractor_type only takes fis or nis, and only on a subcontracted row; destination_country is required when there is no distance_km; origin_hub_name stands for the origin place and country. The logistics endpoints (/v1/logistics/requests, the v2 bulk) do not change.
  • Changed (rolled out per organization, announced to each before): subcontracted logistics legs dated from 2026 are calculated when they are read, with the factor of their own TOC: tonne-km times its well-to-wheel factor, or for an electric TOC the grid factor of the leg’s origin_country times its energy efficiency, plus 86.6 kg CO₂e for a leg with cleaning. Until now they used the intensity value of the organization, which blends in its own fleet and is 0 for an electric TOC. It reaches the package totals of Get Logistics Packages (v2) and List Packages (total_co2e and each leg’s co2e), the logistics report and its export, where an electric subcontracted leg reads grid_electricity as its emission basis. Own-fleet legs and legs dated before 2026 do not change.

2026-11-13

  • Changed: the automatic import of a spreadsheet (auto_import on Get a File (v2)) no longer stops with unresolved_value for a column with more distinct values than Get unique values returns (1,000, truncated), such as a fleet’s license plates: validation reads every cell, and a value the step did not return either matches an option exactly or fails its row. A column that keeps a value matching no option as typed (a waste’s description and treatment) still stops there.
  • Added: GET /v2/imports/{import_id}/status returns raw_file_id, the file the session reads (the spreadsheet that was uploaded), next to file_id, the file its submit generates. A client holding only a session id can now reach that upload. Null when the file was deleted.
  • Changed: Replace an invoice refuses a facility_percentages that lists the same facility twice, as the v2 create already did: half an invoice and the other half go to one facility as one entry of 1, not two of 0.5. Invoices already stored that way are untouched until you replace one.
  • Changed: Create a business travel no longer stores api for every record: with an API key it stays api, and a person signed in to the app stores manual, as the v2 route already did. A source sent in the body is still kept as it is. Records created before keep their value.
  • Changed: a record created with an API key through the v2 routes now reads api as its source, where it read manual before: invoices (and the fuels, refrigerants, water, district and process routes), wastes (their v1 create too), waste-water-treatment measurements, vehicle consumptions and GdO redemptions. The deprecated v1 creates of invoices, measurements and consumptions keep storing manual. manual from now on means the app’s own form, typed in by a person. An integration that filters its records by source=manual must take api too; nothing else changes, and records created before keep the value they were stored with. A redemption sent with source: "pdf" stays pdf.
  • Added: Suggest mapping answers the reason tab_names when a session reads tabs as one table and their names fill a field no column of the file gives: names that are months (ENERO 2026, FEBRERO 2026…) fill the template’s single date, each row dated the first day of its month; names that are one field’s options (one tab per facility, group company, fuel or treatment) fill that field. The automatic import of a spreadsheet uses it too, so such a workbook no longer stops for the missing field or loses it.
  • Changed: Create Business Travels (v2) takes any passenger_number from 1 up: the cap of 100 is gone, so a group trip (a team building, a charter) is one record. The business travels import template drops the cap too.
  • Changed: creating an objective (POST /v2/management-systems/projects/{project_id}/objectives and POST /v2/management-systems/organizations/{organization_id}/objectives) requires status: one of open, in_progress, completed or closed. It defaulted to open; a request without it is now a 422. Any other value is a 422 on create and on update (PATCH), where it was stored as sent. The bulk upload of objectives also asks for a status on every row: a row without one is refused instead of imported as open.
  • Changed: a market of the general company data (type: "market" on POST and PATCH /v1/sot_general_company) lists its countries in name as ISO 3166-1 alpha-2 codes, comma-separated. They are stored upper case, each once, blank parts dropped ("es, fr,ES" is stored as "ES,FR"). A market with no country, or with a part that is not a country, is 422 MARKET_COUNTRIES_INVALID, as the bulk upload and the review of a PDF already refused it. An update checks it when it sets the name or the type of a record that is a market after the edit.
  • Changed: the reading of a policy, strategy, plan or action PDF answers category: null when the document does not make its subject clear, instead of other: the reviewer picks it, as in the form and the bulk upload.

2026-11-12

  • Changed: the automatic import of a spreadsheet (auto_import on Get a File (v2)) no longer stops for rows with nothing measured yet, such as next month’s rows written ahead with no quantity, when they are its only rows with errors and fewer than the rest: it imports the valid rows and ends completed_with_errors with reason: "rows_left_out", those rows staying as errors in import_state.rows.error. A workbook’s tabs with no value where its main tab has numbers are not read with it.
  • Changed: a spreadsheet whose import sends only its valid rows from now on (a person’s submit with ignore_errors, or the automatic import above) is completed_with_errors, not completed: import_state is done with the new reason rows_left_out. Its public status stays completed. Files imported before keep their state.

2026-11-11

  • Added (BETA): POST /v2/files/{file_id}/import-sessions takes sheet_names: several tabs of a workbook (a month or a site each) read as one table, in one session, their columns lined up with the first tab’s and their names added as a column when they say a period, a site, a company or a category. The session answers source_parts, the tabs it read with the rows each gave and the columns renamed or left out. A tab whose columns cannot be lined up is 422 SHEETS_COLUMNS_DIFFER.
  • Added (BETA): the sheet step of a workbook (sheet_suggestion on the import session responses) answers union: the tabs it suggests reading as one table with its main sheet (sheets), what each other tab is next to it (roles: part, alternative, copy or other), the column their names become (tab_name_column) and whether the plan is decided. Null when the step planned no other tab.
  • Docs: Get a File (v2) no longer lists too_large among the reasons of auto_import: the automatic import does not answer it. The MCP’s get_file returns import_state for a spreadsheet, and its guidance follows it.
  • Fixed: Get a File (v2) and the files listing: while Leia imports a spreadsheet, import_state answers status: "running" with actor: "leia" from the moment its import is decided until she stops, on every step. It answered waiting_person with no reason for most files, as if a person had to act. When her work ended without leaving why, it answers waiting_person with reason: "auto_import_failed:JobEndedWithoutStop" instead of no reason.
  • Added: LCA Scenarios: create a scenario from an LCA, log edits under named design changes, compare it with its baseline split per design change, and save or discard it. Available to organizations with scenario modelling enabled (403 SCENARIO_MODELLING_DISABLED otherwise).
  • Added: List Purchase Suppliers returns each supplier’s custom_values and custom_refs (your organization’s custom columns), and filter_by takes a custom column’s key. Deleting a supplier (POST /v2/purchases/suppliers/delete) deletes its custom column values too.
  • Changed: Get a File (v2), the file.finished webhook and My files: while Leia works a spreadsheet, it answers status: "processing" (step: "processing"), not needs_review. A spreadsheet with an import session takes its status, step and actions from its import_state: needs_review with awaiting_mapping and map_columns while it waits for a person, completed (completed or completed_with_errors) once its rows are saved, also when they were saved in the submit itself, and failed with step: "error" when they will not be (rows_not_saved, or the session failed), instead of processing for good. A spreadsheet with a session open no longer answers not_processed.
  • Changed (BETA): Get a File (v2) reads auto_import from the file’s import_state, so the two agree. A run that failed for good is failed with reason: "auto_import_failed:<code>" (it was null). A stop is held as soon as the session has it, also while a new run is queued on it or after a newer command left the run behind (both answered running or null). service_refused:<code> is held. A run that stopped because a person had the file’s session open answers null instead of held with session_in_progress: import_state.actor is then user.
  • Fixed (BETA): POST and PATCH /v2/purchases/suppliers accept every us_sic_v87 code GET /v1/purchases/suppliers/codes lists. Most of them were rejected with REFERENCE_NOT_FOUND on code.

2026-11-10

  • Changed: List Invoices returns confidence as one object for every invoice read from a PDF: the scores of that invoice’s own record of the document. An invoice of a PDF with several records (a certificate of several extinguishers, a statement of several months) used to return a list with every record’s scores.
  • Changed: Get a File (v2): a classified spreadsheet nobody started importing (the automatic import does not work it) answers import_state.reason: "manual_mode" instead of null, on the classify step with status waiting_person.
  • Changed: List Import Sessions leaves out created sessions unless you filter by status=created. Dcycle opens one when it starts importing a spreadsheet on its own, before reading the file; POST /v2/files/{file_id}/import-sessions reads it into that same session and answers 201 with its import_id.
  • Changed: Import sessions no longer expire. A session that was not submitted stays resumable, and its parsed and validated rows are kept until the session, its file (deleted for good, not when it is only moved to the trash), its project or the organization is deleted. New sessions answer expires_at: null; List Import Sessions leaves out sessions with status expired unless include_expired=true, not sessions whose expires_at has passed: older ones are listed again.
  • Added (BETA): PATCH /v2/purchases/suppliers takes attachment_file_id, a file of My files attached to the supplier as its supporting document (null removes it, omitted keeps it). code and framework become optional, together: a record without them only changes the attachment and recalculates nothing. List Purchase Suppliers returns each supplier’s attachment.
  • Added (BETA): POST /v2/purchases/suppliers takes attachment_file_id too: the new supplier is created with that document attached.
  • Added: Get a File (v2) and the file.finished webhook return import_state for a spreadsheet: the step it is on from its upload to its rows written (upload, classify, sheet, read, columns, values, validate, submit, write), whether it runs, waits for a person or ended, why, who works it, and its row counts. Null for any other file. auto_import does not change.
  • Changed: POST /v2/imports/{import_id}/submit answers 410 IMPORT_SESSION_ARTIFACTS_EXPIRED when the session’s temporary data (kept 7 days) expired before the submit read it, for example on the retry of a submit interrupted before sending anything. The session ends expired instead of staying submitting and failing at every retry; create a new session from the file. If the interrupted submit had already sent the file for processing, it answers 409 IMPORT_SUBMIT_ARTIFACT_EXPIRED_AFTER_UPLOAD and the session stays as it is.
  • Added: List Purchase Suppliers returns source and source_file_id (where the supplier comes from, and its file) and code_assigned_by and code_confidence (who gave it its code, and how sure that is, from 0 to 1). Suppliers created before this change take the source of their first purchase.
  • Breaking (BETA): POST /v2/invoices/fuels requires stationary_fuel_id on every record, also with custom_ef_record_id: a record without it is rejected with SCHEMA_INVALID. The custom factor calculates the emissions, but the invoice’s energy (kWh) is calculated with the fuel, so an invoice without one never had energy.
  • Breaking: POST /v1/invoices and PUT /v1/invoices/{invoice_id} reject a heat invoice without stationary_fuel_id with 422 STATIONARY_FUEL_REQUIRED, also when it sends custom_ef_record_id. One with the legacy custom_emission_factor_id is still accepted without a fuel.
  • Docs: openapi-v2.json said of attachment_file_id, in the records of Create Facilities (v2), Create Logistic Hubs (v2), Create Vehicles (v2), Create Hotel Stays (v2), Create Freight Transports (v2), Create Wastes (v2), Create Waste Water Treatments (v2) and every POST /v2/invoices/{category} (e.g. Create Electricity Invoices (v2)), that null removes the attachment and omitting it keeps it, which is how an edit treats it. These routes only create records: null and leaving the field out are the same, and the record is created without an attachment. Nothing else changes.
  • Added: a commuting response can keep the document that backs it, a file of your files (e.g. the public transport pass): attachment_file_id in the records of Create Commuting Periods (v2) and in the body of Replace Commuting Response (v2) (null removes it, omitted keeps it). A file you cannot attach rejects the record with REFERENCE_NOT_FOUND (error_params.field: attachment_file_id). Get Commuting Response (v2) answers it as attachment (file_id, file_name, file_url), and the file of the bulk upload a response was read from as file_id and file_name. Requests without the field are unchanged.
  • Docs: Get Commuting Response (v2) said response_medium is manual for a response read from a file: such a response is file_upload.

2026-11-08

  • Docs: Delete Logistics Requests (v2) said the packages’ totals are up to date when the job finishes. If recalculating a package’s totals fails, the legs are deleted anyway and that package keeps its previous totals until a later calculation touches it. The page also lists the other causes of DELETE_FAILED, JOB_TIMED_OUT, the comments deleted with the legs and that the order of the ids counts for the Idempotency-Key. Create Logistic Hubs (v2) said a replay answers the same hubs: it answers them as they are now, without the ones deleted since, so match items by source_row_index or client_row_id. It also lists every key and organization error.

2026-11-07

  • Added: PATCH /v2/imports/{import_id}/rows takes match and values instead of corrections: it sets the values ({"column_key": value}) on every row that Get Import Rows lists for the errors_only and filters in match, the same selection as a match in Delete Rows, with the same coercion and checks as one-by-one corrections. A request corrects at most 10,000 cells; a larger match answers 400 IMPORT_CORRECTION_MATCH_TOO_LARGE and changes nothing. A match answers the number of rows corrected (updated_count, also present for corrections) and no updated_rows. Requests with corrections don’t change.
  • Added: GET /v2/imports/{import_id}/status returns context, value_mappings, custom_columns and validation_current, so a client resuming a session gets back what was decided in it. POST /v2/imports/{import_id}/validate accepts an optional context (only the template’s context fields: 400 IMPORT_CONTEXT_FIELD_NOT_ALLOWED otherwise) stored on the session before the rows validate.
  • Changed: POST /v2/imports/{import_id}/validate without value_mappings keeps the value mappings of the previous validation instead of clearing them; send value_mappings: {} to clear them. A client that validated again with only the column mapping lost the values it had mapped (its rows came back with unresolved_value errors).
  • Changed: PATCH /v2/imports/{import_id}/rows corrects only a validated session, as deleting and adding rows already did: any other status answers 400 IMPORT_CORRECTIONS_INVALID_STATUS (a submitted one still answers IMPORT_ALREADY_SUBMITTED). Two cases used to accept corrections and no longer do: a session sent back to mapping after a validation, which was corrected against its previous mapping (validate it again), and one being submitted (wait for the submit to finish). A session never validated answered 400 IMPORT_VALIDATED_ARTIFACT_MISSING until now.
  • Removed: GET /api/v1/custom_emission_groups, the legacy paginated list of custom emission groups. It called a function that no longer existed, so any request that passed authentication answered 500; no request succeeded in the last 30 days. Use List Custom Emission Groups. The other legacy /api/v1/custom_emission_groups routes do not change.

2026-11-06

  • Added: the own workforce records can keep the document that backs them, a file of your files: attachment_file_id in the records of Create Workforce Employees (v2) (the employee’s, e.g. the employment contract: null removes it, omitted keeps it, and the last record of an employee that sends it wins), Create Workforce Trainings (v2) and Create Workforce Absences (v2). A file you cannot attach rejects the record with REFERENCE_NOT_FOUND (error_params.field: attachment_file_id); the answers carry attachment (file_id, file_name, file_url). Requests without the field are unchanged.
  • Added: Create Facilities (v2) takes the categories district_heat_and_steam and, on a waste water treatment plant (a record with waste_water), waste_water_treatment, as the Dcycle app’s facility pages do; they were rejected with SCHEMA_INVALID. waste_water_treatment on any other facility still is.
  • Changed: the full OpenAPI document of the API described every timestamp as format: date-time (RFC 3339, offset required), but the older endpoints send most of them without an offset (2026-11-06T10:30:00), so a client or a validator generated from it rejected real responses. Those fields are now format: x-naive-date-time, with a pattern that also accepts an offset; date-time stays where the offset is always there. It also declares the 422 of the v2 standard resources as the problem details they answer (application/problem+json), not as {"detail": [...]}. No response changes, and openapi-v2.json already described both: it does not change. A client you generate again from the full document types those fields as strings instead of dates.
  • Changed: a supplier’s country is stored as its ISO 3166-1 alpha-2 code when a purchase or an uploaded purchases file creates the supplier. POST /v1/purchases, PATCH /v1/purchases/{id} and uploaded files take the code in any case (cn) or the country’s name (China, EEUU) and store CN; a value that names no country answers 422, and in an uploaded file that row is an error and is not saved, while the others are. /v2 already required the code; with custom_ef_record_id an unknown one is now COUNTRY_NOT_SUPPORTED too. Until now the supplier was stored with what was sent, and the purchases list of the organization failed. A supplier stored that way before lists its country as its code, or null.
  • Changed: a purchase matched by its description whose description does not identify the item on its own, such as an internal ERP or lot code (“Green G.CR.ERW”), is read again with the supplier’s name and your organization’s name and sector before falling back to the supplier’s factor. Until now it took the supplier’s registered sector, which is not always what it sold (a coffee-trading platform registered as computer services). Purchases already calculated keep their factor.
  • Fixed: POST /v2/logistics/requests, POST /v1/logistics/requests/bulk and POST /v1/logistics/requests give a TOC new to your organization an intensity value for the year of the leg’s shipment_date. Until now they took year, or the current year when omitted, so a leg of another year had no value for its own year: kgco2e was null in List Logistics Requests and the leg counted 0 in the logistics report. Legs already stored are not changed.
  • Changed (BETA): the logistics intensity values (by TOC and year, hub and project) are recalculated every night for an organization with own-fleet energy (recharges, or supercharger invoices, assigned to a TOC), instead of every 15 minutes after the legs written or deleted by POST /v2/logistics/requests and POST /v2/logistics/requests/delete (paused since 2026-11-04). For other organizations a TOC and year that appear for the first time get the TOC’s well-to-wheel factor. Package totals (Get Logistics Packages (v2)) do not use these values and are up to date when the job finishes.
  • Breaking (BETA): POST /v2/logistics/requests requires subcontractor on every record: true if a subcontractor moved the leg, false if your own fleet did. A record without it, or with null, is rejected with SCHEMA_INVALID. Until now a missing value was taken as subcontracted and a null as own fleet, so a leg could be calculated with the wrong method without any error. POST /v1/logistics/requests/bulk does not change.
  • Added (BETA): POST /v2/employees takes on_existing=reject|reuse. The default, reject, answers 422 EMPLOYEE_ALREADY_EXISTS as before. With reuse, a record whose email (case aside) an enabled employee of the organization already has creates nothing, and its item answers that employee with created: false (without name and email if they answered an anonymous survey); a replay answers the same. Every item carries created. Job items gain the stage value reuse: the record matched an employee who already existed and wrote nothing.
  • Changed (BETA): the auto_import of GET /v2/files/{file_id} says imported only once every row is saved; until then it stays running. A new state, completed_with_errors, says some rows failed, failed now also covers saving that failed (or never started: reason rows_not_saved), and rows_imported and rows_failed count them. The file.finished event already waited for the rows, so its status does not change.
  • Added (BETA): the file.finished webhook event. A spreadsheet uploaded with an API key (POST /v2/files, or the presigned flow and PATCH /v1/files/batch-update) sends it each time it stops: imported (completed), waiting for a person (needs_review, with review_url and the automatic import’s reason) or failed. Its data is the file as GET /v2/files/{file_id} answers it, so there is no need to poll. Subscribe an endpoint to it with event_types. The envelope’s api_version stays 2026-10-01. A file a person uploads in the app or through the Dcycle MCP sends none. The presigned flow, completed with an API key, now records an ingest job with entity_type files, born completed, as POST /v2/files already did.
  • Removed: POST /v2/imports/sessions/ensure-from-raw, earlier than the 30 April 2027 announced with its deprecation this morning: nothing called it any more. Open the import session of an uploaded file with POST /v2/files/{file_id}/import-sessions: it takes the same fields with the file id in the path, answers 201 or 200 (reused), and a workbook with several tabs and no sheet_name is 422 SHEET_REQUIRED with the tabs in error_params.sheets instead of an answer without import_id.

2026-11-05

  • Added: POST /v2/files answers 503 UPLOADS_BUSY with Retry-After when too many uploads are in progress, before reading the body; retry with the same Idempotency-Key. GET /v2/files/{file_id}/sheets reads a workbook once and answers the same tabs after that.
  • Breaking: POST /v2/purchases and POST /v1/purchases ask for the same data as the Dcycle app’s form: the supplier always (supplier_id, or supplier_name / business_name with its country), also with a custom emission factor, and a description when there is no amount spent (quantity) and no custom emission factor. Until now a record with custom_ef_record_id could go without a supplier, and one with only a physical amount and no description was accepted and then failed without emissions; both now answer SCHEMA_INVALID (v1: 422). Purchases uploads ask the same: the supplier on every row, a custom-factor upload included, and the purchased product only on a row without an amount spent and without a custom factor.
  • Changed: in a purchases upload, Recycled is a percentage from 0 to 100 (25 is 25 %), as in the Dcycle app’s form; it was a fraction from 0 to 1. A value between 0 and 1 is accepted with a warning that it reads as under 1 % (a spreadsheet cell formatted as a percentage holds the fraction). An upload already opened before the change keeps the fraction. POST /v1/purchases keeps taking recycled as a fraction.
  • Changed: a purchase sent without matching_method to POST /v2/purchases or POST /v1/purchases, or in a bulk upload that picks no method, is matched by its description when it has one, and by the supplier’s activity otherwise; until now it was always matched by the supplier. The same purchase now gets the same emission factor however it arrives, as in the Dcycle app. Send matching_method: "supplier" to keep the supplier’s factor. An edit (Update Purchase) without matching_method keeps the purchase’s method, unless the description is added or emptied: then it is deduced again.
  • Added: Get Purchase Version History reports the changes of non_currency_quantity, non_currency_unit_id and recycled, which it never reported, and follows each supplier_id change (still the ids) with a supplier change carrying the suppliers’ names (of the purchase’s organization or group; any other supplier by its id).
  • Added (BETA): GET /v2/logistics/packages reads up to 100 packages by id (?id=…&id=…) with their current distance and emissions totals, in the order of the ids; an id with no package in your organization is listed in not_found. One request per 100 packages instead of one GET /v1/logistics/packages/{package_id} per package.
  • Added (BETA): the Import API’s routes on a file. POST /v2/files uploads one file of up to 25 MB in one call and starts its classification, with a required Idempotency-Key. Its type comes from its extension and bytes, not from what the request declares, and bytes that are not what the extension says are refused. GET /v2/files/{file_id} answers its state: status (processing, needs_review, completed, failed), step, reason, actions, review_url, classification, template and automatic import. GET /v2/files/{file_id}/sheets answers the tabs of a spreadsheet and the sheet selector’s answer. POST /v2/files/{file_id}/import-sessions opens the import session of a file: 201 created, 200 reused, 422 SHEET_REQUIRED with the tabs. New problem codes: FILE_TOO_LARGE, UPLOAD_TIMEOUT, FILE_CONTENT_NOT_ALLOWED, FILE_EMPTY, FILE_LIMIT_REACHED, SHEET_REQUIRED, FILE_NOT_IMPORTABLE, IMPORT_FILE_TYPE_NOT_SUPPORTED, IMPORT_FILE_TOO_LARGE, IMPORT_TEMPLATE_NOT_IN_PLAN, IMPORT_HEAVY_OPERATION_BUSY. Every other route under /v2/files is unchanged, errors included.
  • Added: GET /v1/ingestions/tabular/{id} answers auto_import (state, reason, step, import_id), null when the automatic import did not run for the file. Nothing else in its answer changes.
  • Deprecated: POST /v1/files/upload, in favour of POST /v2/files. It keeps working as it did until its removal on 30 April 2027, and its answers carry Deprecation, Link and Sunset.
  • Changed: a request body over 100 MB answers 413 REQUEST_BODY_TOO_LARGE on every route, as soon as it passes the limit. The largest file you can upload does not change. A JSON body, such as a record with an attachment inlined as base64, had no such limit before.
  • Added: facilities, logistic hubs, vehicles and waste water treatments can keep the document that backs them, a file of your files: attachment_file_id in the records of Create Facilities (v2), Create Logistic Hubs (v2), Create Vehicles (v2) and Create Waste Water Treatments (v2) (REFERENCE_NOT_FOUND for a file you cannot attach), and in Update Facility, Update Logistic Hub, Update Vehicle and Update Waste Water Treatment (null removes it; sent alone, nothing is recalculated). Their reads answer it in attachment (file_id, file_name, file_url), and so does List Suppliers (v2) for custom suppliers. Deleting or trashing the file unlinks the record, which stays. A waste water treatment read from a plant report answers the reading’s confidence. Existing records and requests are unchanged; file_id of Update Waste Water Treatment is deprecated in favour of attachment_file_id.
  • Added: a shipment of goods and a GLEC recharge can keep the document that backs it (e.g. the carrier’s invoice, the delivery note or the fuel ticket), a file of your files: attachment_file_id in the records of Create Transports (v2) (REFERENCE_NOT_FOUND for a file you cannot attach), in Update Transport Route (null removes it) and in Create Recharge. The new Update Transport Route Attachment changes only the attachment, without recalculating the route. Get Transport Route and Create Recharge answer it in attachment (file_id, file_name, file_url). Deleting or trashing the file unlinks the record, which stays. Get Hotel Stay answers confidence for a stay read from a hotel invoice. Existing records and requests are unchanged.
  • Added: a hotel stay can keep the document that backs it (e.g. the hotel’s invoice or the booking confirmation), a file of your files: attachment_file_id in the records of Create Hotel Stays (v2) (REFERENCE_NOT_FOUND for a file you cannot attach) and in Update Hotel Stay (null removes it; sent alone, the stay is not recalculated). Get Hotel Stay answers it in attachment (file_id, file_name, file_url). Deleting or trashing the file unlinks the stay, which stays. Existing records and requests are unchanged.
  • Changed (BETA): a PATCH /v2/purchases/suppliers job calculates the suppliers’ purchases itself and finishes once they are calculated; until now it finished once they were queued, and they could stay in_progress for some minutes after ingest_job.finished. A supplier whose purchases end in error now fails its item, with EXCHANGE_RATE_NOT_FOUND or CALCULATION_FAILED and how many failed in error_params.purchases_failed; until now its item succeeded. Chunks hold up to 5,000 purchases between their suppliers, and a supplier with more gets a chunk of its own. Disabled purchases are no longer reset.

2026-11-04

  • Changed: an employee none of whose commuting periods could be calculated, and with no emissions from any other period, answers status error instead of uploaded (Get Employee, List Employees (V2), which filters on it with status[]=error). Correct a period (a motorbike commute needs its fuel, for example) and the employee is uploaded again once it calculates. An employee with some period calculated stays uploaded. Commuting sent through POST /v2/employees/commuting keeps reporting such a period in its ingest job (CALCULATION_FAILED) and does not set error.
  • Fixed: the D&B lookup Dcycle runs after POST /v2/purchases or a purchases upload keeps a code you give that supplier meanwhile, with PATCH /v2/purchases/suppliers or in the app. Until now the lookup could write its own classification over it when it finished, or leave the supplier in_review without a code.
  • Fixed: PATCH /v2/purchases/suppliers recalculates the purchases of the suppliers it classifies, however many they have. Until now the purchases of a supplier with more than about 26,000 in one organization stayed in_progress, without emissions, and its item ended CALCULATION_FAILED. Changing a supplier’s code in the app failed the same way, and so did the recalculation the supplier import offers once its suppliers added up to more than about 26,000 purchases of one organization; both are fixed too.
  • Fixed (BETA): POST /v2/use-of-sold-products calculates each use with the calculation preferences (GWP version and emission factor sources) of its product’s organization when one request carries uses of several organizations of a group. Until now such a request could calculate them all with the preferences of one of those organizations.
  • Added: List Suppliers (v2) answers how each custom supplier was created, in source (manual or bulk_file; null for official suppliers and for those created before it was recorded), and the file a bulk_file one came from, in source_file_id. Requests are unchanged.
  • Deprecated: POST /api/v1/wastes, DELETE /api/v1/wastes/{waste_id}, DELETE /api/v1/purchases/{purchase_id}, DELETE /api/v1/vehicles/{vehicle_id} and DELETE /api/v1/vehicle_consumptions/{vehicle_consumption_id} keep working until April 30, 2027 and answer Deprecation, Link (rel="successor-version") and Sunset: Fri, 30 Apr 2027 00:00:00 GMT. Move to: send wastes with POST /v2/wastes and delete with POST /v2/wastes/delete, POST /v2/purchases/delete, POST /v2/vehicles/delete and POST /v2/vehicle-consumptions/delete.
  • Changed: List Waste Emission Factors is no longer marked deprecated: the route stays, read-only, with no retirement date.
  • Added: an invoice and a waste can keep the document that backs them (e.g. the bill or the waste manifest), a file of your files, apart from the file they were read from: attachment_file_id in the records of POST /v2/invoices/{category} (e.g. Create Electricity Invoices (v2)) and Create Wastes (v2) (REFERENCE_NOT_FOUND for a file you cannot attach), and on Create Invoice, PUT /v1/invoices/{invoice_id}, Update Invoice, Create Waste and Update Waste (404 for a file you cannot attach; on an edit null removes it and omitting it keeps it). Every facility’s part of a split record takes it. Get Invoice and Get Waste answer it in attachment (file_id, file_name, file_url): for an older invoice or waste created by hand, the document it was given in file_url. Deleting or trashing the file unlinks the record, which stays. Existing requests are unchanged; Get Waste Version History also reports attachment_file_name.
  • Added: a vehicle consumption can keep the document that backs it (e.g. the fuel ticket or the fuel card statement), a file of your files: attachment_file_id in the records of Create Vehicle Consumptions (v2) (REFERENCE_NOT_FOUND for a file you cannot attach). Get Consumption answers it in attachment (file_id, file_name, file_url). Deleting or trashing the file unlinks the consumption, which stays. Existing records and requests are unchanged.
  • Added: POST /v2/use-of-sold-products takes an optional attachment_file_id per record: a file of My files attached to the use as its supporting document. A record that replaces a use keeps its document unless it sends the field (null removes it). An attachment_file_id that cannot be attached is REFERENCE_NOT_FOUND (error_params.field is attachment_file_id).
  • Changed: in POST /v2/use-of-sold-products a record that replaces a use without changing what its emissions are calculated from (only its attachment, or nothing) is not calculated again, unless the use’s last calculation did not end cleanly. Its job item is succeeded at once; a job with only such records is born completed and still sends ingest_job.finished for an API key.
  • Changed (BETA): the background recalculation of the logistics intensity values after POST /v2/logistics/requests and POST /v2/logistics/requests/delete, added on 2026-11-03, is paused while it is made to fit the largest organizations. The values catch up later, in the background.
  • Changed: the ingest jobs pages say what a JOB_TIMED_OUT record means in an update job (the change is stored, its emissions were not calculated again: send it again with a new key) and that a JOB_TIMED_OUT record of a vehicle or facility delete goes to support, because a new delete request rejects its id with NOT_FOUND; that resending PATCH /v2/purchases/suppliers with a new key sets the code again and recalculates the purchases; and that the copy of a request is deleted 90 days after it was accepted, sooner when the organization is deleted (it said “at once”).
  • Changed: historical emissions (get, list, create, update) answer source_file_id: the spreadsheet a bulk_file entry was imported from. file_id, file_url and file_name are only the evidence attached to the entry: an imported entry no longer answers its spreadsheet there (existing ones were moved to source_file_id), and attaching a file to it keeps its source. Requests are unchanged.
  • Added: a page for PUT /v1/invoices/{invoice_id}, the replace of an invoice and its facility split: the fields that keep their value when left out, when a row keeps its id, the move to a facility of an organization below the invoice’s, the row the response carries and its errors, the three 409 of the move included. The Invoices with a PDF guide links to it instead of the PATCH page.
  • Changed: the Security, Support and API introduction pages and the organizations, business travels, unknown vehicles, vehicles, vehicle fuels and logistics overviews point to Rate Limits: outside logistics, which has limits of its own, a limit per API key across every route and a ceiling per IP address. They gave older figures, or said those routes had no limit.

2026-11-03

  • Fixed (BETA): the logistics intensity values (by TOC, hub and project) follow the legs written or deleted by POST /v2/logistics/requests and POST /v2/logistics/requests/delete: they are recalculated in the background, usually within 15 minutes after the organization’s last logistics ingest job ends. Until now nothing recalculated them after a v2 write.
  • New: Get Consumption and List All Vehicle Consumptions return each consumption’s source, as invoices and wastes do: manual (a form or the API), pdf (read from a ticket) or bulk_file (a spreadsheet import). Older consumptions read from a ticket or imported from a spreadsheet have theirs; the rest of the older ones answer null, which means manual.
  • Changed: Get Transport Route Version History report the changes of quantity_transported (it was named quantity and never reported), supplier, spend_amount and currency_unit_id, and of each leg of the route by position, as sections.{n}.{field}, emission factor included.
  • Added: Update Transport Route takes transport_direction, which moves a shipment between sent (downstream) and received (upstream) goods.
  • Added: Get Transport Route and List Transport Routes answer how each route was created, in source (manual, pdf, bulk_file or api; null for routes created before it was recorded). For a route read from a document, Get also answers that reading’s confidence. Requests are unchanged.
  • Changed (BETA): in POST /v2/employees/commuting and Replace Commuting Response (v2) a record with situation: inactive or terminated may send modes, and keeps them: the situation only labels the period, whose modes are calculated as an active one’s. Until now such a record was refused with modes, so marking a response terminated replaced its commute with a period with no day and lost its emissions. A record that is not active and sends no modes is calculated at 0 kg instead of left without results.
  • Breaking: in the commuting CSV upload end_date is required, and the in_itinere rows of an employee with the same dates are one commuting response (one period per transport), written whole or refused (an in_labore row stays a period of its own). A row whose commuting_type is blank is in_itinere; one that is neither in_itinere nor in_labore fails with its response. A response that shares days with a stored one is refused instead of updating it, so uploading the same file again refuses it. Until now an uploaded row with the dates of a period stored by hand or by an earlier upload updated it.
  • Changed: the commuting survey is not sent to whoever already answered its period, from any channel, however the recipients are selected: Resend Survey (emails included, and exclude_responded_period now leaves out an employee whose response overlaps the period, not only one with a response of its exact dates) and Import Employees from CSV. The X-Already-Responded response header says how many were left out.
  • Breaking (BETA): POST /v2/employees/commuting takes one record per employee and period with its modes, each day of the week in one mode: in_person (a transport_type with its vehicle details, days by name such as monday, distance_km or origin and destination, daily_trips and its own custom_ef_record_id) and at most one telecommuting (days or total_hours, hours_per_day and its own custom_ef_record_id). A hybrid employee sends both kinds in one record. transport_type: telecommuting, weekly_travels, weekly_telecommuting, total_km, total_telecommuting_hours, hours_worked_per_day and telecommute_custom_ef_record_id leave the record. Each record is stored as one response of the employee, as the app, the survey and the Data Importer store theirs: a commuting period per transport, with the days at home on the first. The job item’s entity_id is that response instead of the period, and a failed item names its period_id. A record whose dates overlap another response of its employee, whatever the weekdays, is refused with COMMUTING_RESPONSE_OVERLAP, or COMMUTING_RECORDS_OVERLAP when the other is an earlier record of the request; they replace COMMUTING_PERIOD_ALREADY_EXISTS and DUPLICATE_COMMUTING_PERIOD, which only caught the same dates. ROUTE_LIMIT_EXCEEDED and a custom factor error name their mode (modes.1.distance_km, modes.1.custom_ef_record_id).
  • Breaking (BETA): Delete Commuting Periods (v2) rejects a field the body does not declare with 422 REQUEST_VALIDATION_FAILED naming it, so a body with ids and a misspelt or extra field is refused: send only ids or response_ids. Until now it was ignored.
  • New (BETA): Delete Commuting Periods (v2) takes response_ids, the entity_ids of a create job, and deletes every period of each response (one with no period left is NOT_FOUND); ids still deletes single periods.
  • New (BETA): Get Commuting Response (v2) (GET /v2/employees/commuting/responses/{response_id}) reads a response with its modes, the period that holds each and its custom answers, and Replace Commuting Response (v2) (PUT) writes it again whole: its periods are rewritten in place and a job (operation: update) recalculates only the periods whose emissions change. response_id is the entity_id of a create job’s item.
  • New (BETA): a record of POST /v2/employees/commuting takes custom_responses, the answers to your organization’s own survey questions by question key, stored on the response’s first period; an answer its column does not take is refused with CUSTOM_RESPONSE_INVALID.
  • Changed (BETA): an ingest job’s source can be survey (an employee’s answer to the commuting survey, which sends no webhook), and a replaced commuting response’s job has operation update. Each answer to the commuting survey and each save of a commuting response in the app is now a job of entity_type employee_commuting, listed by GET /v2/ingest-jobs next to your integration’s; that list does not filter by source, so tell them apart by each job’s source (survey, app or api_bulk).
  • Changed (BETA): every v2 activity date is tomorrow in UTC at the latest (activity dates); a later one rejects its record with SCHEMA_INVALID and date_in_the_future on its field.
    • Newly refused: a future start or end date of a waste, of an invoice of any kind (POST /v2/invoices/*); a future start_date of a vehicle consumption; a future check-out of a hotel stay, sent or worked out from nights. Send each record once its period has ended.
    • Newly accepted: tomorrow in UTC for purchase_date, a waste water treatment’s end_date, accident_date and birth_date, which stopped at today.
    • Same rule, new error: business travels, hotel check-ins, sold product periods, purchases, waste water treatment periods, accidents and birth dates answer date_in_the_future on the date’s field instead of a record-level message.
  • Breaking (BETA): a country Dcycle has no purchase factors for answers COUNTRY_NOT_SUPPORTED, as in hotel stays, in supplier_country of Create Purchases (v2) and in country of Create Purchase Suppliers (v2) (error_params.field names the field); it answered REFERENCE_NOT_FOUND. A unit_id that is not a unit in Create Vehicle Consumptions (v2) answers UNIT_NOT_SUPPORTED, as in purchases and invoices; it answered REFERENCE_NOT_FOUND. REFERENCE_NOT_FOUND keeps its other causes. If your integration branches on these codes, read the new ones.
  • New (BETA): LICENSE_PLATE_ALREADY_EXISTS in Create Vehicles (v2) gives the vehicle that already has the plate in error_params.vehicle_id, also when the plate was written with other spacing, hyphens or case: send its consumptions with that id.
  • Changed (BETA): Create Facilities (v2) takes country in upper or lower case (es or ES) and stores it uppercase; a lowercase code was a SCHEMA_INVALID.
  • Docs: The same data in each route maps the fields that carry the same data under different names (distance, quantity, supplier_id, the carrier) and where their ids come from.
  • Deprecated: in favour of the v2 route of each resource, with Deprecation, Link and Sunset: Fri, 30 Apr 2027 00:00:00 GMT headers like the other deprecated routes:
  • Docs: the pages of the deprecated routes say which cases they still cover until they are retired: reusing an existing employee or linking it to a survey (POST /v1/employees); in_labore periods and survey answers (POST /v1/employee-historic); recycled, purchases with no supplier, and custom_emission_factor_id (POST /v1/purchases); and moving the source files to the trash (delete_source_files) in the deletes of business travels, transports, invoices and vehicle consumptions. Create Commuting Periods (v2) says every period is stored as in_itinere.
  • Fixed: the deprecated single deletes (DELETE /v1/employees/{id}, /v1/employee-historic/{id}, /v1/business-travels/{id}, /v1/hotel-stays/{id}, /v1/transports/{id}, /v1/own_workforce_absenteeisms_accidents/{id}, /v1/own_workforce_trainings/{id} and /v1/purchases/suppliers/{id}) answer their Deprecation, Link and Sunset headers; their 204 carried none.
  • Fixed: the legacy GET and PATCH /waste_water_treatments/{id} answer Deprecation with the day they were deprecated (October 26) and a Link to the same measurement (/v1/waste-water-treatments/<id>/detail and /v1/waste-water-treatments/<id>); it named the route template.
  • Changed: the Link header of the deprecated POST /v1/invoices has rel="deprecation": it points to the page that maps each kind of invoice to its route, not to a route.
  • Added (BETA): POST /v2/logistics/requests/delete deletes up to 5,000 logistics legs by id as an ingest job (operation: delete), like the other v2 resources: all or nothing (NOT_FOUND, DUPLICATE_ID), Idempotency-Key, dry_run, and the packages’ totals recalculated.
  • Breaking (BETA): in POST /v2/logistics/requests a leg’s hub is referenced, never created.
    • hub_id is now the hub’s Dcycle id (a UUID): a hub name in hub_id is SCHEMA_INVALID. Send the name in hub_name instead: it matches the hub’s name exactly. Send one or the other.
    • A hub that does not exist, is another organization’s, or is archived or in error is REFERENCE_NOT_FOUND; an archived or error hub adds its status to error_params. Until now an unknown or inactive name created a new subcontracted hub with the record’s origin as its address. Create your hubs first with Create Logistic Hubs (v2).
    • POST /v1/logistics/requests/bulk is unchanged.
  • Fixed: a free plan organization that a paying organization accepted as its subsidiary follows the paying organization’s plan on the free plan usage limits (facilities, files, imports, document reading, description matching, survey emails, AI requests). A link request that was never accepted, or a paying organization that is disabled, changes nothing: the free plan limits still apply. A fund lends its plan only to the organizations it holds directly. Such a subsidiary still takes no subsidiaries or linked organizations of its own.
  • Added (BETA): a purchase that the free plan’s quota of description lines left without description matching says so with PURCHASE_DESCRIPTION_QUOTA_EXHAUSTED in its error_messages: a warning when the supplier’s factor priced it, and next to PURCHASE_CALCULATION_NO_TOTAL_IMPACTS_PRODUCED when it has only non_currency_quantity and fails. The quota applies to an organization on the free plan with no paying group above it (see How the emission factor is chosen).
  • Changed: an organization on the free plan can open and delete its webhook endpoints (GET and DELETE /v2/webhook-endpoints/{endpoint_id}), as it lists them; both answered 403 PLAN_FEATURE_NOT_INCLUDED before.
  • Added: Create Transport takes an optional origin_country and destination_country per section (ISO 3166-1 alpha-2): when you know the country of a place, Dcycle geocodes it within that country instead of inferring the country from the address text. Requests without them are unchanged.
  • Changed: when the sea separates a road or rail section’s origin and destination (mainland to Mallorca), the NO_DISTANCE_FOUND error now says so: use maritime for that leg or send kms_manual.

2026-10-30

  • Added: List Vehicles answers how each vehicle was created, in source (manual, bulk_file or api; null for vehicles created before it was recorded) and, for a file import, the uploaded file in source_file_id. Requests are unchanged.
  • Added: a purchase can keep the document that backs it (e.g. the order or the delivery note), a file of your files: attachment_file_id in the records of Create Purchases (v2) (REFERENCE_NOT_FOUND for a file you cannot attach), in Create Purchase and in Update Purchase, where null removes it. Reads answer it in attachment (file_id, file_name, file_url). Deleting or trashing the file unlinks the purchase, which stays. Existing records and requests are unchanged.
  • Fixed (docs): Update Purchase no longer lists file_id, file_name and file_url: the endpoint never accepted them (422). Use attachment_file_id to attach a document.
  • Fixed: the Total and subtotal rows of the remuneration pivot table’s remuneration_avg are the simple mean of their employees’ annual full-time salaries, like its rows: each employee counts as 1, whatever their full-time equivalent in the year. The Totals weighted each row by its FTE, so a Total did not match the same average over all its employees. Rows, and the own_workforce_remuneration dataset, do not change. wage_gap is still FTE-weighted.

2026-10-29

  • Breaking (BETA): POST /v2/sold-products takes what the Dcycle app and the file import take.
    • The name is trimmed and matched exactly. Foo adds to Foo, but foo creates another product (the match ignored case). Send each product’s name as it is stored.
    • A sale’s country is a two-letter code of a country the app offers. RoW is SCHEMA_INVALID: send the quantity sold in each country. A code the app does not offer is COUNTRY_NOT_SUPPORTED (it was REFERENCE_NOT_FOUND).
    • A period cannot end in the future: tomorrow in UTC at the latest.
  • Fixed (BETA): two requests to POST /v2/sold-products of one organization at the same time no longer answer an undocumented 409 or 500: the second waits for the first and answers with the documented codes (for example PERIOD_OVERLAPS).
  • Fixed (BETA): POST /v2/use-of-sold-products recalculates with the new factor when one request changes the fuel or gas of periods of the same product and year; it kept the old one.
  • Docs: Create Use of Sold Products (v2) says that a use deleted before its job reaches it fails with CALCULATION_FAILED.
  • Added: POST /v2/employees/survey-rows/resend takes change_pending_period: the selected pending rows are sent for start_date/end_date (then required) instead of their own period, and their survey moves to it, so a survey sent for the wrong year can be corrected. A survey moves only when its email goes out; when the person already has that survey for the new period, the send lands on it and the moved one is removed. Emails sent before keep their old period until their deadline. Without the field nothing changes.
  • Sunset: every route deprecated in favour of a successor is retired on April 30, 2027. Until then it keeps working, and from today its responses carry Sunset: Fri, 30 Apr 2027 00:00:00 GMT (RFC 8594) next to Deprecation and Link (rel="successor-version"); each route’s page states the date. POST /transport_routes, DELETE /transport_routes/{id} and POST /vehicle_consumptions, deprecated without these headers until now, answer all three. Move before that date:
  • Changed (BETA): the copy Dcycle keeps of every accepted v2 bulk request is deleted 90 days after the request was accepted (see Copy of each request). Jobs, their items and the records they wrote are kept as before; nothing changes in any answer.
  • Added: deleting a custom dataset shared with subsidiaries (DELETE /v2/elastic-data/datasets/{dataset_id}, the MCP delete_dataset tool and dcy elastic dataset delete) deletes it in your organization only. Each subsidiary keeps its copy and its records as an independent dataset, no longer linked to yours (source_dataset_id is null). The delete answers those copies in the new detached_copies list (dataset_id, organization_id, organization_name, record_count), and GET /v2/elastic-data/datasets/{dataset_id}/copies lists them before you delete.
  • New (BETA): PATCH /v2/purchases/suppliers gives up to 1,000 existing purchase suppliers their activity code (code and framework, the rules of POST /v2/purchases/suppliers) and recalculates their purchases in an ingest job with operation update, a new value of that field. It is the way to classify a supplier POST /v2/purchases created in_progress, or one the create rejected with SUPPLIER_ALREADY_EXISTS. Unknown or foreign supplier_ids are REFERENCE_NOT_FOUND; all or nothing, Idempotency-Key required, ?dry_run=true.
  • Deprecated: PATCH /v1/purchases/suppliers/{supplier_id}, in favour of PATCH /v2/purchases/suppliers. It keeps working and answers Deprecation, Link (rel="successor-version") and Sunset headers until it is retired on April 30, 2027.
  • Added: GET /v1/purchases and GET /v1/purchases/{purchase_id} return source, where the purchase comes from: manual, bulk_upload, api (sent with an API key, through POST /v2/purchases or POST /v1/purchases) or ocr. Older purchases are filled in from their file or ingest job where it tells; the rest are null.
  • Changed (BETA): a purchase with matching_method: "description" and a physical amount is matched to an ecoinvent or an Agribalyse factor (Agribalyse covers foods as sold: raw, frozen, cooked); it was ecoinvent only. A physical match the matcher is not sure enough to apply on its own still counts, and now ends with LOW_CONFIDENCE_PURCHASE_FACTOR_MATCH and needs_factor_review: true until it is confirmed in the Dcycle app (see Create Purchases (v2)).
  • Changed (BETA): POST /v2/waste-water-treatments accepts null in kg_r_wwt_line and kg_r_sludge_line and reads it as 0 (no methane recovered), as POST /v1/waste-water-treatments and the Data Importer do. Before, a null rejected the whole batch (422 BULK_RECORDS_REJECTED, SCHEMA_INVALID on the record).
  • Changed (BETA): a purchase whose currency has no exchange rate to EUR for its date fails its job item with EXCHANGE_RATE_NOT_FOUND, error_params.field unit_id, as business travels and freight transports answer it; it was CALCULATION_FAILED. The purchase keeps PURCHASE_EXCHANGE_RATE_NOT_FOUND in its error_messages. A business travel’s EXCHANGE_RATE_NOT_FOUND item names its field too (currency_unit_id); its error_params were null.
  • Changed (BETA): a business travel with a spend follows the freight transport rule: the distance keeps the priority. A travel with a distance_km, or with places its distance is measured from, is calculated from its distance; its spend prices it only when it has no distance, its places cannot be measured (a place on no map, no route for its transport) or no distance factor applies (see What a trip cost). Before, the spend always priced it and no distance was looked up. The same in POST and PATCH /v1/business-travels and the Data Importer.
  • Changed (BETA): a freight transport with a spend whose distance cannot be had (a place on no map, no route for its transport) is priced by its spend; it failed with DISTANCE_CALCULATION_ERROR. Without a spend nothing changes. POST and PUT /v1/transports with multidb: true price it by its spend too, instead of refusing the request.
  • Fixed: a measurement edited with PATCH /v1/waste-water-treatments/{wwt_id} to null in kg_r_wwt_line or kg_r_sludge_line, as the page allows, is recalculated with 0 methane recovered. Its recalculation failed and left it error, with its emissions missing or incomplete; send the PATCH again to recalculate such a measurement.
  • Fixed: POST /v1/transports and PUT /v1/transports/{transport_route_id} answer a currency_unit_id that is not a currency with 422 INVALID_CURRENCY_UNIT, as the business travel routes do; it was 400. Both pages document spend_amount, currency_unit_id and multidb, and what the PUT does with the spend.
  • Fixed (BETA): a freight transport sent without a spend gets one problem per section without a distance, with loc ["sections", "<position>"], once the record’s fields are valid; since 2026-10-26 it got one problem on the record, naming only the first such section.
  • Fixed (BETA): a business travel or freight transport priced by its spend with a custom_ef_record_id that is not a factor per unit of currency ended active with no emissions. POST /v2/business-travels and POST /v2/transports reject it with UNIT_NOT_ALLOWED_FOR_CUSTOM_EMISSION_FACTOR (error_params.field names it), and one whose places cannot be measured fails with the reason of its distance.
  • Fixed (docs): the waste water treatment pages and the GHG Protocol guides ask for the values the calculation reads. kg_sludge is dry sludge, in kg of dry matter, and the kg_bod_* fields carry BOD, or COD when the plant’s lines are dqo_industrial, whose factors are per kg of COD (see Organic load: BOD or COD); Create Facilities (v2) says what a plant’s water_type implies. The legacy waste water treatment delete and bulk upload pages point to their v2 routes.
  • Docs: the page of every v2 route that takes ?dry_run=true says that the Idempotency-Key header is still required (a dry run does not check or store it). Without the header a dry run answers 422 REQUEST_VALIDATION_FAILED.
  • Docs: Amounts of money lists the amount and currency fields of each route and their cap: an amount that prices the record (spend_amount, a purchase’s quantity) stays below 10^12, as any other number; a total_spend, which is only kept, below 10^15. The freight transport spend entry is listed on 2026-10-26, the day it shipped, and What a shipment cost names the do_not_know product as EXIOBASE does.
  • Docs: Create Business Travels (v2) and Create Transports (v2) show a record with a spend in their Python example.

2026-10-28

  • Changed: Delete Facilities (v2) packs facilities with little data into the same chunk (up to 50) instead of one chunk per facility, so a job deleting many empty facilities finishes in seconds and its chunks_total is lower. A facility with a lot of data still gets a chunk of its own, and a facility that cannot be deleted still fails alone.
  • Fixed: a waste keeps 3 decimals of its quantity, 1 kg in tonnes. base_quantity and each facility’s quantity were stored with 2, so a waste in tonnes was rounded to 10 kg: 0.004 t became 0 t and had no emissions. Wastes stored before keep their 2 decimals.
  • Breaking (BETA): POST /v2/wastes rejects a record in tonnes with more than 3 decimals, finer than 1 kg (for example 1.2505 t), with the new record code TOO_MANY_DECIMALS: send that quantity in kilograms (1,250.5 kg). A quantity in kilograms takes any number of decimals and is stored rounded to the gram.
  • Fixed (docs): Create Wastes (v2) and the wastes-in-tonnes entry (now under 2026-10-26) said waste factors are per kilogram and a waste in tonnes is converted to kilograms (×1,000), and Wastes that it is converted to kilograms: a quantity in a unit other than that of the emission factor applied (per tonne in DEFRA, the default database) is converted to it. Wastes and List Units state the unit rule of Create Wastes (v2): kilograms or tonnes, or the unit of the record’s custom factor (only m3 with a per-m3 factor). This changelog lists wastes in tonnes on 2026-10-26 and the wastes source and version history on 2026-10-22, the days they shipped.
  • Fixed: a request with an API key that cannot be used answers the code of its reason, as Errors documents: 401 INVALID_API_KEY (unknown or revoked key, or an x-organization-id that is not a UUID), 400 ORGANIZATION_ID_REQUIRED, 401 API_KEY_ORG_MISMATCH, 401 API_KEY_EXPIRED, 403 API_KEY_SCOPE_DENIED, 401 API_KEY_NO_OWNER and 403 LOGGED_USER_NOT_MEMBER. Since 22 October all of them answered 401 AUTHENTICATION_REQUIRED on the /v1 and /v2 routes, which now stays for a request with no credentials (AUTHENTICATION_REQUIRED). On the /v1 and /v2 routes, webhook endpoints included, a key of another organization answers API_KEY_ORG_MISMATCH whether or not the organization exists. The legacy /api/v1/... routes are unchanged.
  • Docs: Authentication and Security describe keys as they are: a name, an access (read and write, or read only) and an expiry chosen when creating them, the date each key expires in Company settings → API Keys (its creator is also emailed 14 and 3 days before), and what happens to a key when its creator leaves the organization or becomes a Viewer.
  • Changed: Create Webhook Endpoint and Update Webhook Endpoint reject a field the body does not declare with 422 REQUEST_VALIDATION_FAILED naming it, as the other v2 routes do. A misspelt field ({"enabeld": true}) was ignored: 200 with nothing changed.
  • Fixed: a webhook test or retry refused with 429 while two of your organization’s calls were stuck no longer pushes their expiry back, so retrying every Retry-After gets a slot within two minutes.
  • Fixed: an ingest_job.finished webhook whose first attempt was cut short (Dcycle’s worker restarting mid-attempt) is still delivered, about 40 seconds later; before, it could stay pending with no further attempt.
  • Documented: at most two webhook tests or retries run at a time per organization (Rate limits), and disabled endpoints count toward the 20 an organization can have (WEBHOOK_ENDPOINT_LIMIT_REACHED).
  • Added: a bulk ingest job whose chunks_done has not moved for 4 hours is closed: its records still pending end failed with the new item code JOB_TIMED_OUT, and the job ends completed_with_errors and sends its ingest_job.finished webhook. A job no longer stays processing forever, so an integration waiting for the webhook always gets it. See A job that does not finish.
  • Fixed: remuneration_avg and wage_gap of own_workforce_remuneration (and the social dataset and remuneration pivot tables that share them) annualize each remuneration over its own period: amount / its days × days of the year, an open-ended one or one of a year or more being annual as it is. Each year’s pay was divided by the contract’s days across the whole query, so a query spanning two years returned about half of the annual salaries, and a leaver’s partial-year pay far less. Other values that change: a contract with several remunerations in a year (a monthly payroll) now gives the annual salary, not about one period’s pay; a query covering part of a remuneration no longer scales it up; and an open-ended remuneration that starts during the queried year is an annual amount (it was scaled up from its days in the query). A remuneration shorter than a year must hold the pay for its period, not an annual rate.
  • Added: Update Purchase (PATCH /v1/purchases/{purchase_id}) takes an optional organization_id that moves the purchase to the header organization or one of its accepted, enabled descendants, keeping its id and custom-column values, linking it to the target’s supplier of the same name and country, and recalculating its emissions under the target. Any other target answers 403, as does a move without a user session; a custom emission factor the target cannot use answers 422 MULTIDB_CUSTOM_RECORD_NOT_FOUND_OR_DISABLED, a custom emission group of the current organization 422 PURCHASE_MOVE_LEGACY_CUSTOM_FACTOR, and a custom-column value the target cannot take 409 PURCHASE_MOVE_CUSTOM_COLUMN_CONFLICT.
  • Added: template columns of Get Import Template may carry options_filter, the option values a row may take by the value of another of its columns. The fuel_type column of business_travels uses it to list the fuels each transport_type accepts, the same ones the row validation accepts. aircraft and ferry list none: leave their fuel blank (a fuel on an aircraft row is still accepted and ignored). Other columns are unchanged.
  • Changed: every step of an import session (status, rows, mapping, validation, row edits, unique values, submit) answers 404 with code PROJECT_NOT_FOUND when the session’s project_id is no longer a project your organization can import into. Only the submit answered it before; deleting the session still works. Create Session checks the project before reading the file.
  • Changed: POST /v2/dashboard answers 404 with code PROJECT_NOT_FOUND for a project your organization cannot open (it answered 403). A project you open but cannot edit the dashboard of still answers 403.
  • Documented: How Bulk Ingestion Works says, for each create route, what a record sent again with a new Idempotency-Key does: duplicated, rejected (with its code), merged or replaced. It also lists POST /v2/logistic-hubs and the logistic_hubs entity type, and the organization waste water treatment measurements and uses of sold products are written to. The lookups table adds where vehicle_id, vehicle_fuel_id, employee_id, own_workforce_id, period_id and lifespan_unit_id come from. The API is unchanged.
  • Changed: PUT /v1/invoices/{invoice_id} with one facility of the invoice replaced by another (the bulk split too) moves that invoice in place: it keeps its id, its comments, its project assignments and its custom-column values, and its emissions are recalculated under the new facility. It used to replace it with a new invoice (new id, comments lost). When the new facility belongs to another organization, its custom-column values move to that organization’s invoice columns, and the move is refused before anything is written with 409 INVOICE_MOVE_LINKED_DATA (the invoice is assigned to projects of its organization), 409 INVOICE_MOVE_CUSTOM_SUPPLIER (a custom supplier the target cannot use) or 409 INVOICE_MOVE_CUSTOM_COLUMN_CONFLICT (a custom-column value the target cannot take: another type there, or a relation/user column). A facility dropped from a split no longer leaves its calculation data behind.

2026-10-27

  • Added: a business travel can keep the document that backs it (e.g. the ticket or the agency’s invoice), a file of your files: attachment_file_id in the records of Create Business Travels (v2) (REFERENCE_NOT_FOUND for a file you cannot attach), in Create Business Travel and in Update Business Travel, where null removes it. Reads answer it in attachment (file_id, file_name, file_url). Deleting or trashing the file unlinks the travel, which stays. Existing records and requests are unchanged.
  • New: Upload Files (POST /v1/files/presigned-urls) takes an optional context_purchase_matching_method (supplier or description) per file: how the purchases of a purchase invoice are matched to an emission factor, as matching_method on PATCH /v1/files/purchase-invoices/batch-update but without skipping classification. When the file is read as a purchase invoice, its review pre-selects that method; for any other document it is ignored. Any other value answers 422.
  • Documented: the per-IP ceiling on API key requests (600 / minute, burst 900, shared by every key sent from one IP; policy baseline_api_key_ip). The rate limit headers of a successful response now describe the limit closest to running out — the per-IP one when it is — instead of always the key’s.
  • Fixed: a response row of List Survey Rows (and its export) shows the period its commutes cover: start_date is its first commute’s start and end_date its last commute’s end. It showed the period the survey was sent for, so a response whose dates were narrowed by the respondent or edited later (e.g. moved from 2026 to 2025) kept the sent period, and period_date:rd[from,to] left it out of the period it is calculated in. Pending rows are unchanged.
  • Fixed (BETA): re-sending a stored remuneration (same employee, contract and remuneration_start_date), with POST /v2/own-workforce/employees or an own workforce import, stores its new remuneration_amount. The record was accepted and counted as imported while the stored amount was kept, so a corrected salary, or a yearly export that re-sends each period with that year’s pay, was lost without an error. The remuneration’s dates are kept as before.
  • Fixed: an own workforce import row that adds a new remuneration to a contract already stored (same employee and contract_start_date, a new remuneration_start_date) writes it. Since 2026-03-31 it closed the stored remuneration the day before and was counted as imported, but the new remuneration was not written. POST /v2/own-workforce/employees was not affected.
  • Deprecated: the organization_id field of POST /v2/facilities records. A create belongs to the organization in x-organization-id: send the organization that should own the facility there. The field keeps working; no removal date yet.
  • Added: Create Business Travels (v2), Create Business Travel and Update Business Travel take what a trip cost, optional: spend_amount with currency_unit_id. A travel with a spend is calculated from it with the EXIOBASE spend factor of its transport in the organization’s country, whatever its number of trips, passengers, places and distance, so these become optional (see What a trip cost). A job item of one whose rate to EUR cannot be had fails with EXCHANGE_RATE_NOT_FOUND; v1 answers 422 INVALID_CURRENCY_UNIT for a currency that is not one. Existing records and requests are unchanged.
  • Deprecated: the CSV uploads that have a v2 successor: POST /v1/vehicles/bulk/csv and /api/v1/vehicles/bulk/csv in favour of POST /v2/vehicles; POST /v1/vehicle_consumptions/bulk/csv and /api/v1/vehicle_consumptions/bulk/csv in favour of POST /v2/vehicle-consumptions; POST /v1/hotel-stays/bulk/csv in favour of POST /v2/hotel-stays; POST /v1/employee-historic/bulk/csv in favour of POST /v2/employees/commuting; and POST /api/v1/purchases/bulk/csv in favour of POST /v2/purchases. They keep working and answer Deprecation and Link (rel="successor-version") headers; no removal date (Sunset) yet.

2026-10-26

  • Changed: on the free plan, purchases sent with matching_method: "description" (Create Purchases (v2)) are matched by description up to a lifetime quota of 500 distinct description lines per organization, never refunded; a line the organization already holds a match for reuses it without counting. Past the quota they get the supplier’s factor, as with supplier, with no error. Paid plans work as before.
  • New: PATCH /v1/waste-water-treatments/{wwt_id} edits a measurement (the fields sent) and recalculates its emissions; it answers the detail with status: "loading". GET /v1/waste-water-treatments/{wwt_id}/detail also returns the rest of the measurement: kg_bod_per_m3_wwt_line, kg_bod_per_kg_sludge_line, kg_bod_per_m3_wwd_line, kg_n_per_m3_wwt_line, kg_n_per_m3_wwd_line, kg_r_wwt_line and kg_r_sludge_line.
  • Deprecated: the legacy GET /waste_water_treatments/{id} and PATCH /waste_water_treatments/{id}, in favour of the two above. They keep working and answer Deprecation and Link (rel="successor-version") headers; no removal date (Sunset) yet.
  • Changed: an organization on the free plan, with no paying group above it, gets 403 PLAN_FEATURE_NOT_INCLUDED on the v2 routes its plan does not include, a dry run included: purchases and suppliers, wastes, business travels, hotel stays, freight transports, employees and commuting, sold products and their use, logistics, own workforce, waste water treatments, water, district heating, district cooling and process invoices, and the webhook endpoints except their list and deliveries. What the free plan includes lists the routes it keeps.
  • New: Create Transports (v2) takes what a shipment cost, optional: spend_amount with currency_unit_id. With it the weight and the sections’ distances become optional; the distance keeps the priority, and a shipment whose distance cannot price it is priced with the EXIOBASE spend factor of each section’s transport in the organization’s country (see What a shipment cost). A job item of one whose rate to EUR cannot be had fails with EXCHANGE_RATE_NOT_FOUND. Requests without a spend are unchanged.
  • New: wastes can be sent in tonnes: POST /v2/wastes takes unit_id cab66828-c2b1-431b-92af-f9ab37149d3c (metric_tonne_(t)) besides kilograms, and GET /v2/units?type=wastes lists it. A quantity in a unit other than that of the emission factor applied (per tonne in DEFRA, the default database) is converted to it, and the calculation detail in the app shows that step.
  • Deprecated: the own workforce CSV uploads POST /v1/own_workforces/bulk/csv, /v1/own_workforce_trainings/bulk/csv and /v1/own_workforce_absenteeisms_accidents/bulk/csv, in favour of POST /v2/own-workforce/employees, /trainings and /absences. They keep working and answer Deprecation and Link (rel="successor-version") headers; no removal date (Sunset) yet.
  • Added: Update Business Travel (PATCH /v1/business-travels/{business_travel_id}) takes an optional organization_id that moves the travel to the header organization or one of its accepted, enabled descendants, keeping its id and custom-column values and recalculating its emissions under the target. Any other target answers 403, as does a move without a user session; a custom emission factor the target cannot use answers 422 MULTIDB_CUSTOM_RECORD_NOT_FOUND_OR_DISABLED, and a custom-column value it cannot take 409 BUSINESS_TRAVEL_MOVE_CUSTOM_COLUMN_CONFLICT.
  • Changed: an organization can have at most 20 webhook endpoints; creating one more answers 422 WEBHOOK_ENDPOINT_LIMIT_REACHED. Send Test Event and Retry Webhook Delivery run at most two at a time per organization; a third answers 429 with Retry-After. Existing endpoints and scheduled deliveries are not affected.
  • Changed: moving a facility with Update Facility also carries the uploads still waiting for review from its page (into the target’s root folder, renamed name (1) on a clash, as are moved files that sat in a folder of the current organization) and splits the offset of a GdO redemption shared across the group’s CUPS again, so the facility’s share lands in the target with the same total. New 409s, nothing changed: FACILITY_MOVE_FILE_PROCESSING while one of its files is being read, FACILITY_MOVE_SHARED_CUPS_GDO_REDEMPTION when such a redemption covers only one of the two organizations, and FACILITY_MOVE_LINKED_DATA also lists project links of its uploaded files.
  • Fixed (BETA): on every v2 bulk route, a record whose calculation fails unexpectedly (for example, emissions too large to store) fails alone: its job item ends failed with CALCULATION_FAILED (DELETE_FAILED in a delete job), the record is left in error where it has a status (not a logistics leg, which stays active without emissions), and the other records of its chunk are calculated. The whole chunk was retried until it was dropped, and the job stayed processing for good, without its ingest_job.finished webhook. A chunk whose every record fails unexpectedly is retried once, about 15 minutes later, in case the error is temporary; then its records end failed and the job finishes. See A job that does not finish.
  • New: List Logistics Requests and List Logistics Recharges return each record’s source: manual, bulk_file or api. A request or recharge created with an API key, or through Create Requests in Bulk, Create Requests (v2) or Create Recharges in Bulk, is api; one created by a person signed in to the app is manual. Sending again a record that already exists keeps its source. Records created before this date are bulk_file when they come from a file, and null otherwise.
  • New: List Periods of a sold product returns each period’s source: manual, bulk_file or api, and null on periods created before this date. A period created through Create Sold Products (v2) is api with an API key and manual for a person signed in to the app.
  • New: List Periods of a sold product returns each period’s created_at (UTC), and uploaded_by_user: who uploaded the file the period was imported from (null for a period typed in the app or sent with an API key).
  • New: Create Logistics Recharge takes the fuel by type, fuel (e.g. diesel), instead of fuel_id: the variant for the recharge’s country is used, as in a file import. Its response says with updated_existing whether it updated a recharge with the same country, date, quantity and license plate instead of adding one. With fuel, a country without a region (such as AQ) is a 422 no_fuel_region on country. List Available Fuel Types takes country and then lists the variant each fuel type gets there (none for a country without a region; 422 for a code that is not ISO 3166-1 alpha-2).
  • Fixed: a recharge created with a project_id in Create Logistics Recharge recalculates that project’s intensity values, as a file import into the project does; only the organization’s were recalculated. Sending again a recharge already linked to the project answered 409 UNIQUE_CONSTRAINT_VIOLATION; it updates it and keeps one link.
  • Changed: a recharge created by a person signed in to the app follows the rules of the file import (quantity above 0, a TOC that exists, a date not in the future and from 1970), answered field by field, and is stored with the country and the TOC the import stores (the TOC row of that vehicle and type an import uses). With an API key nothing changes.
  • Changed (BETA): in POST /v2/own-workforce/employees a record with the contract_start_date of a stored contract and another contract_end_date ends that contract on that day (a leaver), and its remunerations that run past it. A record without contract_end_reason keeps the stored reason. The stored contract kept its end date, and an omitted reason erased the stored one. The own workforce file import applies the same rules.
  • Breaking (BETA): in POST /v2/own-workforce/employees the end date sent for a stored contract must be before the employee’s next contract starts, and a new contract must start after the end date another record sends for the stored contract before it, or the record answers CONTRACT_OVERLAP; the end date was left out instead. A contract with contract_end_reason needs an end date once the request is applied (the record’s, the stored one, or one another contract gives it): new record code CONTRACT_END_REASON_WITHOUT_END_DATE (error_params.field: contract_end_date). Sending again the reason an open stored contract already has is accepted. Until now the contract was stored open with its reason. The own workforce file import applies the same rules as row errors; a stored open contract is imported without its new reason, and the row says so.

2026-10-25

  • Changed: the commuting survey email always has Dcycle’s localized subject. subject in Resend Commuting Survey, CSV import and the v2 resend is accepted and ignored.
  • New: daily survey email budgets per organization (5,000 recipients a day, 25 on the free plan), per address (3 a day) and per request (5,000 addresses): 422 SURVEY_SEND_TOO_MANY_RECIPIENTS, 429 SURVEY_EMAIL_DAILY_LIMIT_EXCEEDED and 429 SURVEY_EMAIL_RECIPIENT_LIMIT_EXCEEDED, with Retry-After. The v2 resend by filters accepts at most 5,000 recipients per send (was 50,000).

2026-10-24

  • Changed: Process File (POST /v1/files/send-event) accepts only the event_type values PROCESS_DOCUMENT and CLASSIFY_DOCUMENT; any other value answers 422. Internal events were never documented for this route. PROCESS_DOCUMENT works as before.
  • Changed: a webhook attempt, test and retry included, now ends after 20 seconds in total, from connecting to the first bytes of your server’s body. A server that answers within the documented 10 seconds is not affected; one that sends its headers slowly past 20 seconds gets READ_TIMEOUT. A host whose name servers do not answer within 5 seconds gets DNS_ERROR.

2026-10-23

  • Fixed: POST /v2/vehicles compares license plates on their letters in any script, not only Latin ones. Two plates that differ only in Greek, Cyrillic, Arabic, Chinese, Japanese or Korean letters (ΑΒΓ-1234 and ΚΗΧ-1234), or that are written with Arabic-Indic digits, were taken for the same plate and answered DUPLICATE_LICENSE_PLATE or LICENSE_PLATE_ALREADY_EXISTS. Spaces, hyphens, dots, case and accents on Latin letters (Ñ = N) are still ignored, and Arabic-Indic digits now count as 0-9 (٩٩٩٩ = 9999). The vehicle and consumption imports match plates the same way.
  • New: historical emissions (get, list, create, update) return source: manual, bulk_file or estimation, and null on entries created before this date. An entry created through this API is manual.
  • Changed: an x-partner header that is not a plain partner name (only letters, digits, - and _, up to 64 characters) is refused with 400 INVALID_PARTNER_HEADER on every route. A missing header, or your organization’s partner such as dcycle, works as before.
  • Fixed: with an API key, file URLs in responses (for example file_url in historical emissions) are presigned links that expire after one hour, as they already were for users of the app. They came back unsigned, and the private bucket answered 403 to them. Download the file when you read it, or read the record again for a fresh link; do not store the URL.
  • New: a hotel stay can be calculated with a record of your custom emission factor databases, as invoices and business travels: custom_ef_record_id in Create Hotel Stays (v2), Create Hotel Stay and Update Hotel Stay (null on update goes back to the default factor of the country), and in Get Hotel Stay. The record must be for hotel_stays and per room-night.
  • Changed: Update Hotel Stay clears name, hotel_name and address when sent as null; until now a null was ignored for every field.
  • Changed: a stay created by a person signed in to the app, through Create Hotel Stays (v2) or Create Hotel Stay, has source manual; with an API key it stays api.
  • Fixed: an API key writes only while the person who created it may write in the organization in x-organization-id. If that person is a Viewer there (for example after an admin changed their role, or a parent organization’s key used in a subsidiary where its creator is a Viewer), every write answers 403 VIEWER_CANNOT_WRITE, on /v1, /v2 and the legacy /api/v1 routes alike; reads keep working. Writes through such a key used to go through.
  • Fixed: Create LCA from JSON and Create LCA from Excel reject a key whose creator is no longer an enabled member of the organization with 403 LOGGED_USER_NOT_MEMBER, as every other route does.

2026-10-22

  • New: Get Waste and List Wastes return each waste’s source, as invoices do: manual (a form or the API), pdf (read from a document) or bulk_file (a spreadsheet import). Older wastes read from a document or imported from a spreadsheet have theirs; the rest of the older ones answer null, which means manual.
  • New: GET /v1/wastes/{waste_id}/versions returns a waste’s change history, newest first: who changed which fields and when, as for purchases and business travels. transaction_id returns only one change. A parent organization can read it for a waste in a subsidiary’s facility.
  • Fixed (BETA): on every v2 bulk route, a request sent while another with the same Idempotency-Key is still being processed answers 409 IDEMPOTENCY_KEY_IN_PROGRESS at once, as documented, and writes none of its records. It could start writing and fail with an undocumented 409 UNIQUE_CONSTRAINT_VIOLATION on a resource with a unique key, such as Create Workforce Employees (v2). The detail of IDEMPOTENCY_KEY_REUSED now also says “or on another route”.
  • Fixed (BETA): on every v2 bulk route, a text with a lone surrogate (\ud800: valid JSON, but not a character) is SCHEMA_INVALID (string_unicode) on its field, and a refused record’s client_row_id with one is echoed with the surrogate as the text \ud800. It answered 500.
  • Fixed (BETA): on Create Logistics Requests (v2) and Create Logistic Hubs (v2), a NUL character is string_contains_nul in every text field, as on the other routes. Some fields refused it with another type (such as value_error or string_pattern_mismatch) and others could answer 500. In a leg’s origin and destination it is still dropped, as before and as in v1. On Create Commuting Periods (v2) it is string_contains_nul on transport_type, fuel_type, vehicle_size and renewable_energy too, whatever the period: some periods answered value_error on the record, others dropped the value.
  • Fixed (BETA): on every v2 bulk route, a body nested more than about 1,000 levels deep is 422 REQUEST_VALIDATION_FAILED (too_deep). It answered 500.
  • Fixed (BETA): a whole number in plain digits (no . or exponent) past about 10^308 is SCHEMA_INVALID (number_out_of_range) in a whole-number field without a cap of its own. Outside the logistics routes, that is today only a vehicle’s registration_year in Create Vehicles (v2). It answered 500.
  • Breaking: writes to survey templates (create, update, clone, delete, set default, block, unblock) need the Employees permission (emission:employee_mobility::all) on the caller’s role, or on the role of the API key’s creator. Without it they answer 403 PERMISSION_DENIED; until now membership was enough. Reads are unchanged.
  • Breaking: API keys expire. A new key lasts 90 days by default (a year at most), and every existing key expires on 20 January 2027. An expired key answers 401 API_KEY_EXPIRED; the person who created it is emailed 14 and 3 days before, and once it has expired. See Expiry, read-only keys and last use.
  • New: a key can be read-only: any write it sends answers 403 API_KEY_SCOPE_DENIED. Existing keys keep reading and writing.
  • New: keys created from now on are stored only as a hash and start with dck_; the list of keys shows the first characters of each one and when it was last used.
  • Changed (BETA): Create Hotel Stays (v2) takes a stay still going on, as the app’s form: only the check-in cannot be in the future, with a day of margin over UTC. The address is trimmed, and an empty or blank one is no address, as in the app and the file import. nights is no longer capped at 366 (100,000 at most, a technical bound), as in the app (which sends them with the dates), the file import and v1; the check-out it gives must be by the end of next year. A stay still going on, an empty address and more than 366 nights were SCHEMA_INVALID; a blank address was stored as sent.
  • Breaking (BETA): Create Employees (v2) stores and answers emails trimmed and in lower case, and names trimmed and with single spaces, as the file import stores them. A blank name or email counts as not sent: a record with only a blank name is SCHEMA_INVALID (it was stored), and one with a blank email and a name is taken (it was SCHEMA_INVALID).
  • Docs: Create Employees (v2) says that situation changes neither what the commuting periods calculate nor whether the survey is still pending for them. The employees and hotel stays overviews link their v2 pages.
  • Changed: POST /v1/files/readings accepts a recharges reading only in kg (kilogram_(kg)), as Create Refrigerant Invoices (v2) does. Any other unit answers 400 UNIT_NOT_SUPPORTED and creates nothing; before, a unit it could not match created the invoice in review, and litres or kWh one that never calculated.
  • New: GET /v1/logistics/requests/unique-values takes search with field=vehicle_license_plate: only the plates matching it, on letters and digits and ignoring case (1234-ab finds 1234AB6), so a dropdown can reach the plates past the first 1,000. Above 250,000 shipments it matches the start of the plate only.
  • Changed: List LER Codes and List R/D Codes describe each code with its official English text: the European List of Waste (Decision 2014/955/EU) and Annexes I and II of Directive 2008/98/EC. The previous descriptions were machine translations. Codes and ids do not change; a search on the old wording may no longer match.
  • New: Get Waste Water Treatment and List Waste Water Treatments return each treatment’s source, as wastes and invoices do: manual (a form or the API) or bulk_file (a spreadsheet import), and file_id, the raw file of the import that created it. Older imported treatments have theirs; the rest of the older ones answer null, which means manual.
  • New: GET /v1/waste-water-treatments/{waste_water_treatment_id}/versions returns a treatment’s change history, newest first: who changed which fields and when, as for wastes. transaction_id returns only one change. A parent organization can read it for a treatment in a subsidiary’s plant.
  • Fixed: List Facilities returns the wwt_line, sludge_line, wwd_line and water_type of each waste water plant, as documented; they always answered null. A plant takes waste water treatments only with its three lines set.
  • Docs: facilities document type as the site’s ownership (owned or rented), not office or warehouse. Create Facilities (v2) accepts every ISO 3166-1 country, plus Kosovo (XK): send GB, not UK.
  • Docs: Update Waste says what an edited waste needs: a LER code, a description or an emission factor the waste already has, else 422 WASTE_LER_CODE_REQUIRED. It, Update Invoice and Delete Invoice document the 404 for an id outside your perimeter, not a 403. The waste pages share one list of statuses, and Get Waste explains WASTE_DEFAULT_TREATMENT_ASSUMED and how to calculate the real treatment.
  • Docs: List Units gives type=electricity and type=process as the units of their invoices, List Suppliers (v2) documents every field it returns, and Create Invoice with PDF Attachment says which routes take the PDF.
  • Docs: the invoice, facility and waste pages link the v2 creates and deletes instead of the deprecated v1 ones, broken links are fixed, and pages that shared a title in the navigation name their resource.
  • Fixed (docs): this changelog lists the logistics route change on 2026-10-18 and the vehicle move on 2026-10-16, the days they shipped; the public spec’s single ApiKey scheme is labelled Breaking (SDK); and 2026-09-29 lists the Breaking change of POST /v2/wastes (custom_ef_record_id instead of custom_emission_factor_id).
  • New: Update Facility moves a facility, with all its data, to another organization when the body has organization_id: the header organization or one of its subsidiaries. The facility, its invoices, wastes and waste water treatments keep their ids and custom-column values; their emissions are re-attributed to the target without being recalculated, so every figure reads the same there. A move is refused (409/422, nothing changed) while something of the current organization still points at the facility, a supply contract also serves another facility or is held by the current organization, an uploaded file also produced another facility’s records, or a custom emission factor its records use is not available to the target.
  • Breaking: an API key acts only as the member who created it (Company settings → API Keys shows who created each key). x-user-id used to let a request with a key act as any member of the organization; it is now optional everywhere, including the legacy /api/v1/... routes, and ignored. Requests without it, or with the creator’s id, are unchanged. A request that named another member now acts as the key’s creator, with the creator’s role: to act as that member, use a key they created. A key with no creator answers 401 API_KEY_NO_OWNER.
  • Fixed: Get Import Rows with errors_only=true applies filters and sort_by, as documented. It returned every error row, unfiltered and in row order.
  • New: Delete Import Rows takes match (errors_only, filters): it deletes every row Get Import Rows lists with those parameters, on every page, instead of sending their row_indices.
  • Changed (BETA): a business travel sent with distance_km may give its origin or its destination alone, as POST /v1/business-travels and the Data Importer allow (the origin’s country picks the factor’s), and takes renewable_energy: do_not_know, as v1 does. Both were SCHEMA_INVALID.
  • Fixed: a business travel whose origin has two characters but is not a country code (UK, NY) is located on a map to pick its factor’s country, in v1, v2 and the Data Importer. It was read as a country code, found no factor and failed to calculate.
  • Docs: Create Business Travels (v2) says that passenger_number counts only with a factor per passenger-km (not with one per vehicle-km, as most car factors from 2024), what to do for each failure code, and to delete a failed travel before sending it again.
  • Breaking (BETA): a freight transport is SCHEMA_INVALID with a transport_date in the future (date_in_the_future), as in the app and the Data Importer; until now it was calculated. A road section needs travel_method, which is truck, car or bicycle: motorbike and electric_kick_scooter are not among the transport combinations and were always refused as TRANSPORT_COMBINATION_INVALID, and a road section without a vehicle was too. origin and destination are 1 to 256 characters once trimmed (they were up to 500, and a longer one failed in the job): a blank or longer one is SCHEMA_INVALID.
  • Changed (BETA): the job item of a freight transport that failed carries error_params naming the first section that failed and why ({"field": "sections[i]", "message": ...}). A place on no map is DISTANCE_CALCULATION_ERROR; an error on our side, such as the geocoder refusing us, is CALCULATION_FAILED (it was DISTANCE_CALCULATION_ERROR), and the section’s error_messages in Get Transport Route say impact_calculation_error.
  • Changed: when the geocoder refuses a request (an expired key, its own error), the routes that locate an address answer 503 GEOGRAPHIC_POINT_GEOCODER_REFUSED instead of 404 GEOGRAPHIC_POINT_GEOCODE_NOT_FOUND: the address is not at fault, and it can be sent again later.
  • Docs: Create Transports (v2) says to delete a failed shipment before sending it again (it stays stored, in error), and List Job Items says which routes give error_params.
  • Breaking (BETA): the supplier country of POST /v2/purchases/suppliers (country) and of POST /v2/purchases (supplier_country) must be one Dcycle has purchase factors for: another, such as XX or the withdrawn AN, is REFERENCE_NOT_FOUND naming the field. Until now any two letters were stored, and every spend-based purchase of that supplier ended CALCULATION_FAILED after the 202. A purchase sent by supplier_id whose supplier was stored with such a country, or without one (error_params.country: null), is now refused with the new record code SUPPLIER_COUNTRY_NOT_SUPPORTED. A purchase with custom_ef_record_id is not checked. Send the supplier’s current ISO 3166-1 alpha-2 code (XK for Kosovo). A blank supplier name (business_name, supplier_name) is SCHEMA_INVALID (string_blank).
  • Fixed (BETA): Create Purchase Suppliers (v2) said a supplier’s status is active or in_progress; it can also be in_review, when it could not be classified.
  • Breaking (BETA): on Create Workforce Trainings (v2) and Create Workforce Absences (v2), a description or type of blanks only is SCHEMA_INVALID (string_too_short): texts are trimmed, as the file import and the app trim them. Absence hours that round to 0 are hours_round_to_zero; 1e-05 was stored as 0. On Create Workforce Employees (v2), external_employee_id and employment_category are trimmed the same way, and a blank one is SCHEMA_INVALID.
  • Changed (BETA): Create Workforce Trainings and Absences (v2) take dates and hours as the file import does. Dates have no upper bound, so an apprenticeship or a long leave may be sent with its planned end: training dates from 1970, as the import; absence and accident dates from 1900, as the employees’ contracts (accident dates stay not in the future). Hours with more than 1 decimal are rounded half up (0.25 is 0.3) instead of refused, and so is a remuneration_amount of Create Workforce Employees (v2) with more than 2 (0.125 is 0.13). An absence of 1,250,000 days or more needs its absence_hours (absence_hours_required). These dates, hours and amounts were SCHEMA_INVALID.

2026-10-21

  • Changed: in Create Purchases (v2), the suppliers classified by their activity (SIC/CNAE code, from D&B) are those that make up 80 % of a chunk’s spend, ranked once per supplier by the amount of its purchases in EUR: an amount in another currency is converted to EUR, and a purchase with only non_currency_quantity counts 0. A supplier classified from D&B keeps that activity; the generic one never replaces it.
  • Changed: training hours accept zero in the Excel importer, manual form and POST /v2/own-workforce/trainings. Negative hours are rejected.
  • Changed (security): a key whose creator no longer belongs to the organization in x-organization-id, or that is sent without that header, answers 401.
  • New: POST /v2/employees/commuting takes total_telecommuting_hours on a telecommuting period: the hours worked from home in the whole period, instead of weekly_telecommuting (never both), for when only that total is known. The emissions of the days at home are then calculated on those hours. POST /v1/employee-historic and PATCH /v1/employee-historic/{employee_historic_id} take it too, and the commuting periods return it (List, Get, and the periods of GET /v1/employees); null when the period gives weekdays. It takes 0.01 to 100,000 hours, and the employees download shows it in its own column.
  • Deprecated: the v1 single deletes that have a v2 bulk delete: DELETE /v1/business-travels/{id}, /v1/hotel-stays/{id}, /v1/employees/{id}, /v1/employee-historic/{id}, /v1/own_workforces/{id}, /v1/own_workforce_trainings/{id}, /v1/own_workforce_absenteeisms_accidents/{id}, /v1/transports/{id}, /v1/vehicles/{id} and /v1/vehicle_consumptions/{id}, in favour of the POST /v2/.../delete of each resource (one id is a batch of one). They keep working and answer Deprecation and Link (rel="successor-version") headers; no removal date (Sunset) yet.
  • New: GET /v1/logistics/requests/unique-values supports field=vehicle_license_plate: the license plates your active shipments carry (at most 1,000, in plate order), with counts, to filter List Logistics Requests by filter_by=vehicle_license_plate:in[...]. project_id narrows them to the shipments linked to a project.
  • Changed: GET /v1/stationary/fuels returns each fuel’s units: the units its emission factors convert from, to send as unit_id in POST /v2/invoices/fuels. An invoice in another unit is stored but not calculated, and its job item ends failed with CALCULATION_FAILED; see which unit to send.
  • Deprecated: POST /v1/invoices/bulk-delete, in favour of POST /v2/invoices/delete, and POST /v1/purchases/bulk-delete, in favour of POST /v2/purchases/delete. Both keep working and answer Deprecation and Link (rel="successor-version") headers; no removal date (Sunset) yet. The bulk-delete-by-filters routes have no v2 successor and are not deprecated.
  • Fixed (BETA): the v2 bulk creates (all but logistics requests and logistic hubs) answer 422 SCHEMA_INVALID on the field, instead of 500, for a text with a NUL character (string_contains_nul) and for a number that is not finite (finite_number) or not below 1,000,000,000,000 (10^12) in absolute value, 10^15 for an amount of money (total_spend) (number_out_of_range). A business travel with an infinite distance was accepted (202) and its job never finished.
  • Breaking (BETA): a distance is capped at 500,000 km (business travels distance_km, commuting periods total_km, wastes total_km_to_waste_center) and a refrigerant recharge at 1,000,000,000 kg: above, less_than_equal. Until now such a record was stored.
  • Breaking (BETA): a commuting period requires what the app’s form and the Data Importer require, without which it cannot be calculated. SCHEMA_INVALID now answers a car without vehicle_size or fuel_type, a motorbike without fuel_type, total_km: 0 without origin and destination, a place to measure with no letter or digit, daily_trips 0 on a commute or above 3, a telecommuting period with neither weekly_telecommuting nor total_telecommuting_hours, or with daily_trips above 0, a commute while inactive or terminated, a start_date in the future (date_in_the_future) and an end_date after Dec 31 of the current year (date_after_this_year). Until now such a period was stored and ended CALCULATION_FAILED, with 0 kg, or calculated as if the employee were active. A request leaves at most 500 distinct routes to measure (a route repeated in the request counts once): new record code ROUTE_LIMIT_EXCEEDED.
  • Changed (BETA): on a commuting period:
    • fuel_type left out on a bus, metro, train or trolleybus is do_not_know (a trolleybus without one had no factor);
    • carpool left out or null is false;
    • total_km: 0 with origin and destination is measured, as when left out;
    • with total_km above 0, origin or destination may be sent alone;
    • the places are trimmed, and a blank one counts as not sent;
    • an inactive or terminated period takes no transport (its commute fields may be sent empty), is stored as the app stores it, with no emissions, and its job item succeeds;
    • a telecommuting period is stored with 0 km and 0 trips, as in the app.
  • Fixed (BETA): the commuting page, and List Employees, said daily_trips counts one-way trips. It counts round trips, each twice total_km, as the calculation always did.
  • New (BETA): POST /v2/waste-water-treatments writes up to 5,000 measurements of waste water treatment plants in one request, all or nothing, with v1’s fields (no file_url). Their CH₄ and N₂O emissions are calculated by an ingest job; a measurement whose calculation fails ends error and its item CALCULATION_FAILED. Periods of a plant may overlap, as in v1. New record codes FACILITY_NOT_WASTE_WATER and WASTE_WATER_LINES_NOT_CONFIGURED (a plant without its three lines, which v1 accepts and fails to calculate); a period cannot end in the future. POST /v2/waste-water-treatments/delete deletes up to 5,000 by id with their emissions. Job entity_type waste_water_treatments.
  • New (BETA): POST /v2/facilities takes an optional waste_water block (wwt_line_id, sludge_line_id, wwd_line_id, methane_burned) that makes the facility a waste water treatment plant, and every facility it answers carries waste_water (null for the rest). A line not in the catalog is REFERENCE_NOT_FOUND (error_params.field waste_water.<slot>); one not of its slot’s line_type, or three lines of different water_type, is the new record code WASTE_WATER_LINE_INVALID.
  • New: GET /v1/waste-water-treatments/lines, the catalog of treatment lines (id, line_type, water_type, line_code, line_original_name) a plant takes its lines from. It answers with x-api-key and x-organization-id; see Lookups outside the document.
  • Deprecated: POST /v1/waste-water-treatments, in favour of POST /v2/waste-water-treatments, and POST /v1/facilities/waste_water, in favour of POST /v2/facilities with waste_water. Both keep working and answer Deprecation and Link (rel="successor-version") headers; no removal date (Sunset) yet.
  • New (BETA): POST /v2/use-of-sold-products sets how up to 5,000 periods of sold products are used (period_id, a periods[].id of POST /v2/sold-products; lifespan and lifespan_unit_id; at least one of the blocks electricity, combustion, water and fugitive, the last in kg), all or nothing, and calculates their emissions as an ingest job (202). One use per period: a record for a period that has one replaces it and calculates it again. New record codes DUPLICATE_PERIOD (two records for one period) and USE_OF_PRODUCT_INCONSISTENT (periods of a product sharing a calendar year must burn the same fuel and leak the same gas, or none, counting the stored uses and the request). POST /v2/use-of-sold-products/delete deletes up to 5,000 uses by id with their emissions; the periods and their sales stay. Job entity_type use_of_product.
  • Deprecated: POST /v1/sold-products/{sold_product_id}/use-of-product and PATCH /v1/sold-products/{sold_product_id}/use-of-product/{use_of_product_data_id}, in favour of POST /v2/use-of-sold-products. Both keep working and answer Deprecation and Link (rel="successor-version") headers; no removal date (Sunset) yet.
  • New (BETA): POST /v2/sold-products creates up to 1,000 sold products with their sales (1 to 20 periods, each with a unit_id and the quantity sold per country, ISO alpha-2 or RoW) in one request, all or nothing, and answers them with the periods written (201). Nothing is calculated. Products belong to the x-organization-id organization, where a name is one product (case aside): a record whose name exists adds its periods to it. Unlike the file import, nothing is merged silently: new record codes PERIOD_OVERLAPS (a period overlapping one the product has, the same one included), DUPLICATE_PRODUCT and DUPLICATE_SALE. POST /v2/sold-products/delete deletes up to 5,000 by id with their periods, sales, use data and emissions, as an ingest job. Job entity_type sold_products.
  • Deprecated: POST /v1/sold-products/bulk-delete and DELETE /v1/sold-products/{sold_product_id}, in favour of POST /v2/sold-products/delete. Both keep working and answer Deprecation and Link (rel="successor-version") headers; no removal date (Sunset) yet.
  • Deprecated: the v1 writes that a v2 bulk route now replaces: POST /v1/business-travels and POST /v1/business-travels/bulk-delete, in favour of POST /v2/business-travels and POST /v2/business-travels/delete; POST /v1/transports and POST /v1/transports/bulk-delete, in favour of POST /v2/transports and POST /v2/transports/delete; POST /v1/hotel-stays and POST /v1/hotel-stays/bulk-delete, in favour of POST /v2/hotel-stays and POST /v2/hotel-stays/delete; POST /v1/employees, in favour of POST /v2/employees, and POST /v1/employee-historic, in favour of POST /v2/employees/commuting; POST /v1/own_workforces/bulk-delete, POST /v1/own_workforce_trainings/bulk-delete and POST /v1/own_workforce_absenteeisms_accidents/bulk-delete, in favour of POST /v2/own-workforce/employees/delete, POST /v2/own-workforce/trainings/delete and POST /v2/own-workforce/absences/delete; POST /v1/vehicles and POST /v1/vehicles/bulk-delete, in favour of POST /v2/vehicles and POST /v2/vehicles/delete; and POST /v1/purchases/suppliers/anonymous and DELETE /v1/purchases/suppliers/{supplier_id}, in favour of POST /v2/purchases/suppliers and POST /v2/purchases/suppliers/delete. All keep working and answer Deprecation and Link (rel="successor-version") headers; no removal date (Sunset) yet.
  • Breaking (BETA): POST /v2/invoices/electricity takes toc, the type of the own electric vehicles a supercharger charges. It is required when a facility of the invoice belongs to an active supercharger hub (new record code TOC_REQUIRED) and rejected when none does (TOC_NOT_ALLOWED); it must name an electric road TOC of List TOCs (category road, wtw 0), looked up for the facility’s region, then GLO, then EU, or the record is REFERENCE_NOT_FOUND. Only the supercharger’s share of the invoice keeps it, and it feeds that TOC’s GLEC intensity value, as the invoices of a supercharger sent from the app do. Until now the route did not take toc, so a supercharger’s invoice sent through it did not count for its vehicles. The v1 invoice routes and invoice file uploads take the European electric TOC as the last resort too: a supercharger outside the regions with electric road TOCs keeps the TOC it is sent with (until now it was dropped), and an edit keeps the one it has.
  • New (BETA): POST /v2/hotel-stays creates up to 5,000 hotel stays in one request, all or nothing. Their emissions are calculated by an ingest job, which places the stays sent with an address on the map first. POST /v2/hotel-stays/delete deletes up to 5,000 by id with their emissions. Unlike POST /v1/hotel-stays, a stay has at least one night, cannot be in the future, and its country must have a hotel emission factor (List Available Countries): new record code COUNTRY_NOT_SUPPORTED. Job entity_type hotel_stays.
  • Docs: Create Wastes (v2) says how a waste sent with a LER code and no R/D code is priced: with DEFRA’s landfill factor for its material (its combustion factor where DEFRA has no landfill one), or OCCC’s average for its fraction; not with “a factor defined for that LER with no treatment”, which not every LER has. It also says that a unit other than kilograms needs your own factor in that unit, that a collection with a single date sends it as both dates, that a period over several years is split by days, which database edition prices each year, and how the transport is priced. List Wastes filters ler_code[] and rd_code[] by the catalog’s exact code (150101, R01; the examples showed 15 01 01 and R1, which match nothing) and lists review, and Get Waste says that a factor matched with low confidence leaves a waste uploaded, not review. How bulk ingestion works explains how to test an integration on your own organization, which replaces the advice to use a test organization. The Scope 3 Category 5 and ISO 14064 Category 4.2 waste guides send their data through Create Wastes (v2) and no longer show v1 code.

2026-10-20

  • New (BETA): POST /v2/purchases/suppliers creates up to 1,000 purchase suppliers in one request, all or nothing, and answers them (201). A supplier sent with its activity code and framework (cnae_2009 or us_sic_v87, from GET /v1/purchases/suppliers/codes) is classified at once (active); one sent without is classified as POST /v2/purchases classifies a new supplier (in_progress until then). A later purchase with the same supplier_name and supplier_country uses it. New record codes SUPPLIER_ALREADY_EXISTS (an existing supplier is never reclassified) and DUPLICATE_SUPPLIER. POST /v2/purchases/suppliers/delete deletes up to 5,000 by id as an ingest job, refusing a supplier that still has purchases (SUPPLIER_HAS_PURCHASES). Job entity_type purchase_suppliers.
  • New (BETA): POST /v2/employees/commuting creates up to 5,000 commuting periods of your employees in one request, all or nothing, with the transport rules of POST /v1/employee-historic. The distances of periods sent without total_km (from origin and destination) and the emissions are calculated by an ingest job, which fails a period whose distance cannot be had with DISTANCE_CALCULATION_ERROR per item. POST /v2/employees/commuting/delete deletes up to 5,000 by id with their emissions. A period belongs to its employee (employee_id), who has one per start_date and end_date: new record codes COMMUTING_PERIOD_ALREADY_EXISTS and DUPLICATE_COMMUTING_PERIOD. Job entity_type employee_commuting.
  • New (BETA): POST /v2/employees creates up to 1,000 employees of the commuting survey in one request, all or nothing, and answers them (201), with status: loading until their commute is known; POST /v2/employees/delete deletes up to 5,000 by id with their commuting periods, emissions and survey answers, as an ingest job. Employees belong to the x-organization-id organization, where an email is unique (case aside): new record codes EMPLOYEE_ALREADY_EXISTS and DUPLICATE_EMAIL. Creating an employee sends no email. The other /v2/employees routes keep their current error shape.
  • New (BETA): POST /v2/logistic-hubs creates up to 1,000 logistic hubs in one request, all or nothing, and answers them (201). country and address are required; the address is not geocoded (logistics requests that reference the hub geocode it when they are calculated). An owned hub takes the facility_id of its facility, and category is a category of List HOCs. New record codes HUB_NAME_ALREADY_EXISTS, DUPLICATE_HUB_NAME, FACILITY_REQUIRED_FOR_OWNED_HUB, FACILITY_NOT_ALLOWED_FOR_SUBCONTRACTED_HUB and FACILITY_ALREADY_LINKED_TO_HUB.
  • Changed (BETA): in POST /v2/wastes and the invoice routes, facility_percentages may add up to less than 1 again: only the shares sent are counted and the rest of the record is not, as for an organization that takes its part of a shared office’s invoice. [{facility_id, percentage: 0.5}] keeps half of the record. Over 1 is still FACILITY_PERCENTAGES_INVALID. This reverts the “exactly 1” rule of 2026-10-19.
  • Changed: PATCH /v1/wastes/{waste_id} takes a split of facility_percentages under 1 too: one facility at 0.5 keeps half of the waste. Since 2026-10-19 it answered FACILITY_PERCENTAGES_INVALID.
  • New (BETA): POST /v2/transports creates up to 5,000 freight shipments in one request, all or nothing, each with 1 to 20 sections. The distances of sections sent without one and the emissions are calculated by an ingest job, which fails a shipment whose section distance cannot be had with DISTANCE_CALCULATION_ERROR per item. POST /v2/transports/delete deletes up to 5,000 by id with their sections and emissions. The record checks what POST /v1/transports checks: the weight unit, each section’s transport against the combinations (new record code TRANSPORT_COMBINATION_INVALID) and its custom factor. Unlike v1, transport_direction is required and a section’s distance is distance_km.
  • Breaking (BETA): in POST /v2/vehicle-consumptions a vehicle priced by its own emission factor takes only that factor’s unit, as the app’s form and the file import; any other is the new record code UNIT_NOT_ALLOWED_FOR_CUSTOM_EMISSION_FACTOR, with the accepted unit in error_params.supported. Until now a vehicle with a legacy custom group took any unit and its factor was applied to whatever was sent (1,200 km against a per-litre group came out as 1,200 litres), and one with a record of your custom databases took the fuel’s units, so a unit the record does not price (km against a per-litre record) was accepted and failed in the calculation. Euros are no longer turned into litres for these vehicles, not even for a factor per litre: send litres. See A vehicle with its own emission factor.
  • New (BETA): POST /v2/business-travels creates up to 5,000 business travels in one request, all or nothing. The distances of travels sent without one and their emissions are calculated by an ingest job, which reports a route the transport cannot make (TRANSPORT_MODE_INVALID_FOR_ROUTE) or a place it cannot find per item. POST /v2/business-travels/delete deletes up to 5,000 by id with their emissions. The record checks what the file import checks: the fuel against the transport (new record code FUEL_NOT_ALLOWED_FOR_TRANSPORT), 1 to 100 passengers, no dates in the future.
  • Changed (BETA): what an Idempotency-Key identifies on the v2 bulk routes.
    • It identifies the route and the body. Reusing a key on another route answers 422 IDEMPOTENCY_KEY_REUSED, as with another body. Until now, the same key and body sent to another route (e.g. district heating instead of district cooling) answered the first route’s result and wrote nothing.
    • It expires after 24 hours. After that, a request with the key is a new one. Until now a key never expired, so a batch sent again after its records were deleted answered the old result and wrote nothing.
    • A whole number is the same written 1200 or 1200.0, so such a retry is no longer IDEMPOTENCY_KEY_REUSED.
  • Fixed (BETA): the public OpenAPI document marks Idempotency-Key required on POST /v2/vehicles, as the API already required it for a batch. It was the only v2 route publishing it as optional.
  • Fixed (BETA): POST /v2/own-workforce/employees takes contract and remuneration dates as the file import does: from 1900, with no upper limit. A contract or remuneration dated before 1970 or after next year answered 422 SCHEMA_INVALID.
  • Breaking (BETA): in POST /v2/own-workforce/employees a remuneration that starts before its contract answers 422 SCHEMA_INVALID on remuneration_start_date (type: remuneration_before_contract), as the file import refuses it. Until now it was stored.
  • Fixed (BETA): POST /v2/own-workforce/employees answers 422 SCHEMA_INVALID for a remuneration_amount of 100,000,000 or more, or with more than 2 decimals, instead of 500: the amount is stored with 2 decimals and must be under 100,000,000.
  • New (BETA): POST /v2/own-workforce/absences creates up to 1,000 absences and accidents at work of your employees in one request, all or nothing, and answers them (201); POST /v2/own-workforce/absences/delete deletes up to 5,000 by id, as an ingest job. New record codes ABSENCE_OVERLAP and ACCIDENT_DUPLICATE.
  • New: POST /v2/wastes records and PATCH /v1/wastes/{waste_id} take file_id: a file of your files, uploaded beforehand, attached as the supporting document instead of a base64 data URI in file_url. The waste stores the file’s URL and name. An unknown file, one outside your perimeter or not uploaded answers REFERENCE_NOT_FOUND (field: "file_id") in POST /v2/wastes and 404 in the PATCH; file_id with a data URI file_url is rejected. Data URIs keep working as before.
  • New: a custom emission factor for wastes may be per cubic metre as well as per kilogram: GET /v2/units?type=custom_emission_factors_wastes lists both. A waste in POST /v2/wastes stays in kilograms unless it is priced with a custom factor (custom_ef_record_id), which also accepts the factor’s unit. A per-m3 factor takes a waste in m3 only: one in kilograms with it answers UNIT_NOT_SUPPORTED, since kilograms cannot be converted to m3.
  • Fixed (BETA): POST /v2/vehicles accepts every fuel of List Vehicle Fuels that has emission factors for the vehicle class. Seven of them (marine_gas_oil, biomethane, biopropane, biogas, biomethanol, jet_kerosene, jet_kerosene_renewable) answered REFERENCE_NOT_FOUND, as if they did not exist, although they have.
  • Fixed: the Deprecation header of POST /v1/facilities and DELETE /v1/facilities/{facility_id} dates their deprecation to 2026-10-09, the day POST /v2/facilities shipped (it said 2026-10-07).
  • Docs: How bulk ingestion works puts the flow of every v2 bulk route on one page: which routes answer at once (201) and which hand the work to an ingest job (202), all or nothing, dry_run (it neither checks nor stores the Idempotency-Key), what the key identifies for 24 hours, the perimeter of organizations a request can write to, how to follow a job, and every entity_type. The v2 pages link it and say the same: a synchronous create sends no ingest_job.finished webhook; a job’s webhook goes to the organization the batch was sent as; a 422 lists up to 100 invalid records; the shared id of a waste split across facilities is source_invoice_id. The invoice pages say which emission factor each category uses and what co2e adds up; the purchase page, how suppliers are classified and matched by name and how returns net; the vehicle consumption page, which fuels and dates a euro amount covers; the own workforce page, how to change a stored contract. List Vehicle Fuels names the units field units, as the response does (it said fuel_units, and its examples failed on it). Changes that shipped without an entry are now listed under their own dates: 2026-10-14 (two ingest job fixes), 2026-10-15 (purchase_type) and 2026-10-19 (low-confidence matches).

2026-10-19

  • Changed: POST /v1/files/readings no longer dates an item without a start or end date with today, and refuses a year outside 1970 to next year or an end before the start (422 READING_ITEMS_INVALID, one entry per bad date); PUT /v1/files/readings/{file_reading_id} applies the same rule to invoice and waste items. Confirming a reviewed ESG document refuses a date out of range (DOCUMENT_DATE_OUT_OF_RANGE; valid_until may run up to 2100) and a valid_until before the document date (VALID_UNTIL_BEFORE_DOCUMENT_DATE), whether or not the date was edited.
  • New (BETA): POST /v2/own-workforce/employees loads up to 1,000 rows of the own workforce template (employee, contract, remuneration) in one request, all or nothing, applied as the file import applies them, and answers the employees (201); POST /v2/own-workforce/employees/delete deletes up to 5,000 employees by id with their contracts, remunerations, trainings and absences, as an ingest job. New record codes CONTRACT_OVERLAP and REMUNERATION_OVERLAP.
  • New (BETA): POST /v2/own-workforce/trainings creates up to 1,000 trainings of your employees in one request, all or nothing, and answers them (201); POST /v2/own-workforce/trainings/delete deletes up to 5,000 by id, as an ingest job. A training belongs to its employee (own_workforce_id).
  • Deprecated: POST /v1/vehicle_consumptions and POST /v2/vehicle_consumptions (one consumption per request, calculated in the request), in favour of POST /v2/vehicle-consumptions. Both keep working and answer Deprecation and Link (rel="successor-version") headers; no removal date (Sunset) yet.
  • Breaking (BETA): in POST /v2/wastes and the invoice routes, facility_percentages must add up to exactly 1 (within 0.0001), so a record is always attributed whole; a split under 1 answers FACILITY_PERCENTAGES_INVALID (it kept only that share). In wastes facility_percentages is now required and facility_id, deprecated on 2026-10-19, is no longer accepted. In wastes, the invoice routes and POST /v2/facilities, a field that a record, or one of its facility_percentages entries, does not declare is rejected (SCHEMA_INVALID, extra_forbidden) instead of ignored, as purchases and vehicles already do: a waste’s entry takes only facility_id and percentage (organization_id is rejected).
  • Fixed: PATCH /v1/wastes/{waste_id} takes facility_percentages and applies the edit to every part of a split waste: the split is rewritten (a facility kept keeps its part, a new one gets a part, one left out is deleted with its emissions) and every part takes the new quantity, dates and codes. Until now the field was ignored and only the part in the path changed, so the app’s edit form left a split waste with its old split and its other parts with the old quantity. The split must add up to exactly 1 (FACILITY_PERCENTAGES_INVALID, FACILITIES_MUST_BE_UNIQUE, REFERENCE_NOT_FOUND); a waste with a part outside your perimeter keeps its split (resend it as stored to edit your part). Without facility_percentages nothing changes.
  • Docs: OpenAPI & SDKs no longer says the document includes every path: it describes the v2 standard resources only. The page now lists the routes that give the ids a record needs (organizations, facilities, units and currencies, LER and R/D codes, suppliers, fuels, refrigerant gases, custom emission factors, vehicle catalogs, TOCs, hubs), which are not in it and answer x-api-key and x-organization-id alone.
  • Changed: GET /v1/stationary/fuels and GET /v1/recharge/gases authenticate like GET /v1/vehicle-fuels. x-user-id is ignored: with an API key the request acts as the key’s owner, who must be a member of the x-organization-id organization. With an API key, a missing or malformed x-organization-id answers 422 (it was 400 or 401) and an organization that does not exist 404 ORGANIZATION_NOT_FOUND (it was 401). A user signed in to the app must send x-organization-id and be a member of that organization (403 LOGGED_USER_NOT_MEMBER; superadmins excepted), and supplier accounts get 403. The lists do not change.
  • New: Custom KPI Presigned URL takes an optional content_hash (SHA-256 of the file) and ?upload_checksum=sha256, which returns upload_headers: the headers to PUT the file with. With them, storage refuses bytes other than the hashed ones. Without the parameter nothing changes.
  • New (BETA): POST /v2/vehicle-consumptions creates up to 5,000 vehicle consumptions in one request, all or nothing, and calculates their emissions as an ingest job; POST /v2/vehicle-consumptions/delete deletes up to 5,000 by id with their emissions. A consumption belongs to its vehicle’s organization and can carry what it cost (total_spend, currency_unit_id). New record codes VEHICLE_ARCHIVED, FUEL_NOT_IN_VEHICLE, UNIT_NOT_ALLOWED_FOR_FUEL and NO_FUEL_PRICE_FOR_SPEND.
  • Fixed (BETA): the public OpenAPI document no longer marks pages as required in the paginated v2 lists (List Ingest Jobs, List Job Items and the webhook endpoint and delivery lists): the API never sent it, so a client generated from the document rejected correct answers. The answers do not change. The document now also declares the headers these routes send: Location on every 202, Idempotent-Replayed on the success of every route that takes an Idempotency-Key, and the 409 IDEMPOTENCY_KEY_IN_PROGRESS with its Retry-After.
  • New (BETA): a replayed request (same Idempotency-Key and body) on the asynchronous v2 routes (POST /v2/wastes, /v2/invoices/*, /v2/purchases, their deletes, POST /v2/facilities/delete, POST /v2/vehicles/delete and the logistics bulk) answers Idempotent-Replayed: true, as the synchronous ones already did.
  • New (BETA): POST /v2/vehicles creates up to 1,000 vehicles in one request, all or nothing, and answers them (201); POST /v2/vehicles/delete deletes up to 1,000 by id with their consumptions, as an ingest job. Vehicles belong to the x-organization-id organization. New record codes LICENSE_PLATE_ALREADY_EXISTS, DUPLICATE_LICENSE_PLATE and FUEL_NOT_ALLOWED_FOR_VEHICLE_TYPE.
  • Docs: Create Vehicle and Delete Vehicle no longer promise what the routes do not do: creating a vehicle calculates nothing (its co2e is 0 until it has consumptions), and deleting one deletes its consumptions.
  • Breaking (BETA): POST /v2/wastes takes facilities as facility_percentages, like the invoice routes, and honours every entry: a single entry now goes to that entry’s facility with its share. Until now a single entry was ignored and the whole waste went to facility_id. facility_id is deprecated: it is read only when facility_percentages is absent (the whole waste to that facility) and ignored, unchecked, when a split is sent. A custom_ef_record_id must be visible from the organization of the first facility of the split (it was the organization of facility_id). A disabled facility is rejected like any facility outside your perimeter.
  • Breaking (BETA): one code per rule across wastes and the invoice routes. A facility that does not exist or is outside your perimeter answers REFERENCE_NOT_FOUND (error_params.field: facility_id), like every other id; wastes answered FACILITY_NOT_BELONG_TO_ORGANIZATION and invoices FACILITY_NOT_FOUND. In the invoice routes a facility listed twice answers FACILITIES_MUST_BE_UNIQUE and shares over 1 FACILITY_PERCENTAGES_INVALID, as in wastes, instead of SCHEMA_INVALID. POST /v1/wastes keeps its codes.
  • Docs: Errors explains the two vocabularies of validation problems (REQUEST_VALIDATION_FAILED and a record’s SCHEMA_INVALID).
  • New (BETA): credit notes (abonos) on electricity and fuel invoices. POST /v2/invoices/electricity, POST /v2/invoices/fuels, Create Invoice and PUT /v1/invoices/{invoice_id} accept a negative quantity: it cancels consumption and subtracts its kWh and emissions. A facility’s consumption of one energy (same fuel and unit, all its invoices added up) never goes below zero: a credit note that would take it there is rejected with the new code NET_CONSUMPTION_NEGATIVE, and so is deleting or lowering invoices that the credit notes still need (POST /v2/invoices/delete and the v1 deletes).
  • Breaking (BETA): on the other POST /v2/invoices/* routes (water, refrigerants, process, district heating and cooling) a negative quantity is rejected with the record code NEGATIVE_QUANTITY_NOT_ALLOWED instead of SCHEMA_INVALID.
  • Changed: total_spend may be negative (a refund) on every POST /v2/invoices/* route. Until now it was rejected with SCHEMA_INVALID.
  • Changed: Create Invoice and PUT /v1/invoices/{invoice_id} answer a negative base_quantity on a water, recharge, process or district_heat_and_steam invoice, or on an electricity invoice billed in a currency, with 422 NEGATIVE_QUANTITY_NOT_ALLOWED instead of a validation error. The bulk split refuses a split that takes a facility’s consumption below zero with 422 NET_CONSUMPTION_NEGATIVE, and the bulk update lists the invoices whose unit change would do it in failed_ids.
  • Docs: the Quickstart now uses the v2 API end to end: create a facility, check a waste with ?dry_run=true, send it, follow its ingest job (or receive ingest_job.finished) and clean up. Authentication recommends x-api-key on v1 and v2, says not to send x-user-id, and explains what a key can access. Core Concepts has a new section, “Sending Data Through the API”. The API introduction lists Purchases, Units and Vehicles among the modules with v2 endpoints, and /v2/vehicles in the v2 standard. The guides that still send data through v1 say so at the top and link the v2 routes. The API key screen is called Company settings → API Keys everywhere, as in the app, and the Get API Key button of the navigation bar opens the page that explains how to get one.
  • Docs: the MCP quickstart’s OpenAI example sends the API key as Bearer dcy_<key>, the prefix the MCP server requires (without it the connection was refused), and asks for approval before a tool writes or deletes (read-only tools still run on their own). The CLI pages, examples and automation guides install and use the real CLI, dcy (from Dcycle’s Homebrew tap), instead of pip install dcycle-cli and dc, and no longer pass --yes to dcy logistics upload (it has no such flag). The MCP pages no longer tell you to install dcycle-mcp with pip or uvx: that package is not published by Dcycle; use the hosted server.
  • Changed: a purchase or a waste whose emission factor is matched from its description with low confidence counts in the totals while it waits for review. The purchase ends active (until now in_review) with LOW_CONFIDENCE_PURCHASE_FACTOR_MATCH in error_messages, and the waste uploaded (until now review) with LOW_CONFIDENCE_WASTE_FACTOR_MATCH; both answer needs_factor_review: true until the factor is confirmed in the app (Create Purchases (v2), Create Wastes (v2)). A waste left without a factor for some year stays in review. Records stored under the old rule were moved, with their warning.

2026-10-18

  • Breaking (BETA): the logistics bulk create moves to POST /v2/logistics/requests, one route for one or many legs, as POST /v2/wastes and POST /v2/facilities already are. Same body (records, 1 to 5,000), headers, 202 with the ingest job, errors and behaviour; only the path changes. POST /v2/logistics/requests/bulk no longer exists. The previous POST /v2/logistics/requests, which created one leg synchronously, is replaced: to create one leg, send one record.

2026-10-16

  • New: Update Vehicle (PATCH /v1/vehicles/{vehicle_id}, and PATCH /v2/vehicles/{vehicle_id}) takes an optional organization_id that moves the vehicle to the header organization or one of its accepted, enabled descendants, keeping its id, consumptions and custom-column values and recalculating its emissions under the target. Any other target answers 403, as does a move without a user session.
  • Deprecated: POST /v1/purchases, in favour of POST /v2/purchases, and DELETE /v1/purchases/{purchase_id}, in favour of POST /v2/purchases/delete. Both keep working and answer Deprecation and Link (rel="successor-version") headers; no removal date (Sunset) yet.
  • Changed: Create Logistics Requests (v2) rejects a record without distance_km whose origin or destination is missing or blank (422 BULK_RECORDS_REJECTED with the record code ADDRESS_REQUIRED). Until now it was accepted with 202 and always failed to calculate with NO_DISTANCE.
  • Docs: Create Logistics Requests (v2) is marked BETA, like the other v2 standard resources, and documents how it reads the record fields (a toc that matches no TOC falls back to the generic average of category, an unknown load_unit is read as kilograms, an existing hub_id replaces origin, year and tkm are not used to calculate), that legs without movement_id of the same client, movement_stretch and movement_stage are one leg, the actual cause of each per-leg error code, and the 401, 400, 404, 403 and 429 answers. Its example sent a toc that does not exist (van_diesel); the TOC is van_3_5_t_diesel. The route’s breaking change is now listed on 2026-10-14, the day it shipped, instead of 2026-10-13.
  • Changed: Update Commuting Period keeps the total_km you send for a commute with origin and destination. Until now every update with both addresses replaced the km with the distance calculated from them; it is now calculated only when the route changes or no km are sent.
  • New: Create Commuting Period calculates total_km from origin and destination when no km are sent (or 0), as the update and the CSV upload do. Until now it stored the commute with 0 km.

2026-10-15

  • Docs: List Purchase Suppliers (GET /v1/purchases/suppliers) documents the lookup that gives the supplier_id of Create Purchases (v2). The filter_by operators of List Purchases are corrected: they are two letters (ne, ge, le, lk for “contains”…), not neq, gte, lte or contains, which were ignored.
  • New: List Waste Water Treatments and its detail return file_name, the original name of the attached document (null when it was not sent with one).
  • New: POST /v2/wastes, Create Waste and Update Waste take an optional file_name, the original name of the attached document, returned as file_name.
  • Fixed: those routes upload a file_url sent as a base64 data URI (data:<mime>;base64,...) and store its URL. Until now they stored the data URI itself, so the attachment could not be downloaded. A data URI that is not valid base64 is now rejected (422, or SCHEMA_INVALID for the record on POST /v2/wastes).
  • New: Create Invoice (POST /v1/invoices) and PUT /v1/invoices/{invoice_id} take an optional file_name: an attachment sent as a base64 file_url keeps that name in your files and in file_name, instead of a generated adjunto-factura-<id>.
  • New (BETA): POST /v2/purchases creates up to 5,000 purchases in one request, all or nothing, as an ingest job that classifies the suppliers sent by name and calculates the emissions; POST /v2/purchases/delete deletes up to 5,000 by id. Purchases belong to the x-organization-id organization. New record code NEGATIVE_AMOUNT_NOT_ALLOWED.
  • Fixed: Get Invoice (GET /v1/invoices/{invoice_id}) answers co2e with the invoice’s calculated emissions, the same figure as List Invoices. Until now it read a column the current calculation does not fill, so invoices calculated by it (every /v2/invoices one) showed 0.
  • Breaking (BETA): every POST /v2/invoices/* create route takes the facilities only as facility_percentages, now required: an invoice of one facility is [{"facility_id": "...", "percentage": 1}]. facility_id is no longer a field of the record (it is ignored), so a record that sends only facility_id is rejected with SCHEMA_INVALID.
  • Deprecated: POST /v1/invoices, in favour of the POST /v2/invoices/* route of each kind of invoice, and DELETE /v1/invoices/{invoice_id}, in favour of POST /v2/invoices/delete. Both keep working and answer Deprecation and Link (rel="successor-version") headers; no removal date (Sunset) yet.
  • Breaking (SDK): the public OpenAPI document authenticates every operation with one security scheme, ApiKey (x-api-key), and keeps x-organization-id as a required header of each operation. Until now each operation named schemes the document does not define (OAuth2PasswordBearer, APIKeyHeader), which made it invalid OpenAPI, and also listed x-api-key and the legacy x-user-id as parameters; the unused Organization scheme is gone. The API does not change, but a client generated again loses the x_api_key and x_user_id arguments from its methods: set the API key once, in its configuration, and keep passing the organization.
  • New: organization_name on each project of List Projects: the name of the organization that owns it (its company name, or its trade name when the company name is empty), so a project inherited from a parent organization names the parent.
  • Docs: List Suppliers (v2), List Stationary Fuels and List Refrigerant Gases document the lookups that give the ids the invoice routes take.
  • Fixed: Get Facility, List Facilities and Update Facility return supercharger: whether the facility’s logistics hub is an active supercharger (an electric-vehicle charging station), null when the facility is not a hub. These routes left it out, so the app stopped offering the vehicle’s TOC on a supercharger’s electricity invoices.
  • Fixed: List TOCs (GET /v1/logistics/tocs) answers a wtw of 0 (the electric road TOCs) as 0; it answered null, so the app’s supercharger TOC selector listed no TOC.
  • Changed: an invoice’s toc is kept only on the electricity rows of a supercharger hub’s facility (Create Invoice and PUT /v1/invoices/{invoice_id}); on any other row it is dropped, as is a toc that matches no TOC. PUT /v1/invoices/{invoice_id} keeps each row’s toc when the body omits it; send "toc": null to clear it. Until now an omitted toc cleared it. The POST /v2/invoices/* routes don’t take a toc yet: link a supercharger’s invoice to its vehicles with POST /v1/invoices or in the app.
  • New: an electricity invoice PDF uploaded to a supercharger hub’s facility takes the toc of the vehicles it charges: toc on POST /v1/files/invoices/presigned-urls and context_toc on POST /v1/files/presigned-urls (with context_category: "electricity" and context_facility_id). The invoices created from the file carry it. It is ignored for any other category or facility.
  • Fixed: the logistics intensity values, and so GET /v1/logistics/report, follow every change to a supercharger hub’s invoices (create, edit, delete, recalculation): they are recomputed about a minute later. Until now they kept the values of the last logistics upload. Values stored before this release change on the next such change. A project’s report counts the project’s supercharger invoices in its own-fleet TOCs; it read fuel recharges only, so its electric TOCs reported 0.
  • New: each tocs_info row of GET /v1/logistics/report carries no_consumption_data: true for an own-fleet TOC with no consumption in the report year (no fuel recharge, no supercharger invoice), whose iv and emissions of 0 are not a measurement.
  • New: custom KPIs take the value type number_breakdown (Create KPI): an amount per category, e.g. litres of diesel and of petrol. Its categories are its options (2–100, as a select), a submission carries one entry per answered category with option_id and value_numeric, and the KPI’s value is their sum. The newest submission is the KPI’s whole answer: an edit resends every amount of a breakdown it changes.
  • Changed: an option_id on a value of a custom KPI without options answers 422 OPTION_NOT_ALLOWED (Submit Response, Edit Assignment Values); until now it was stored and ignored.
  • Changed: two entries for one custom KPI that is not a number_breakdown in the same submission answer 422 DUPLICATE_KPI_VALUE when they name different option_ids; with the same kpi_id and option_id they stay a 422 validation error.
  • Changed: a custom KPI submission or edit takes up to 1000 values (was 100).
  • Fixed: a purchase whose emission factor is matched from its description to a physical-unit database (e.g. ecoinvent) takes purchase_type: average_data, as announced on 2026-10-09; until now it kept spend_based (v1 and POST /v2/purchases). Purchases calculated before this fix keep spend_based until they are recalculated.

2026-10-14

  • Breaking: POST /v2/logistics/requests/bulk is an ingest job, like Create Wastes (v2). It answers 202 with the job and Location: /v2/ingest-jobs/{id} instead of 200 with results[]/errors[]; requires Idempotency-Key; is all or nothing (one invalid record answers 422 BULK_RECORDS_REJECTED and writes nothing, where until now the valid records were kept), including two records of the batch for the same leg (DUPLICATE_RECORD, until now the last one was kept); accepts ?dry_run=true and client_row_id; ignores options.continue_on_error; limits the identifying fields (movement_id, client, package_key, hub_id, …) to 255 characters; and answers errors as problem details. Each leg that cannot be calculated is a failed job item with its reason (NO_DISTANCE, SEA_CROSSING_REQUIRES_MARITIME_MODE, …). These batches no longer appear in the app’s activity panel. POST /v1/logistics/requests/bulk is unchanged.
  • Changed: ingest job items (GET /v2/ingest-jobs/{id}/items) may carry a category’s own error_code and error_params, not only CALCULATION_FAILED / DELETE_FAILED.
  • Changed: a bulk request that arrives while another one with the same Idempotency-Key is still being processed (every v2 bulk endpoint that takes the header, e.g. POST /v2/wastes and POST /v2/facilities) answers 409 IDEMPOTENCY_KEY_IN_PROGRESS with Retry-After, instead of an undocumented 409 UNIQUE_CONSTRAINT_VIOLATION. Nothing of that request is written; retry it with the same key and body.
  • Changed: sections[].electric and sections[].refrigerated of Create Transport Route and Update Transport Route are optional and default to false (until now omitting them answered 422; an update replaces the sections, so an omitted flag is stored as false). The shipments bulk upload no longer needs the electric and refrigerated columns either: a missing column reads as “no”, like a blank cell.
  • Changed: in GET /lca/portfolio/{lca_id}/dashboard the B6/B7 top_contributors are split by the country the consumption happens in, given by their new country_code: electricity used in Spain and in Portugal are two contributors (until now one, with both summed).
  • Fixed (BETA): a chunk of a v2 ingest job interrupted by a deploy is run again by another worker. Until now its records were reported failed (CALCULATION_FAILED) and stayed so, although nothing was wrong with them. To calculate a record that failed this way in an earlier job, delete it and send it again: sending it again without deleting it stores it twice. Fuel invoices (POST /v2/invoices/fuels) are calculated in chunks of 50.
  • Fixed (BETA): an ingest job’s finished_at is the time its last chunk settled; until now it was when that chunk started, up to about 20 minutes early for fuel invoices.
  • Fixed (BETA): POST /v2/invoices/process answers batches of thousands of records; until now one of 5,000 ran until it timed out and wrote nothing.

2026-10-13

  • Changed: reading items’ supplier_match.method (List File Readings) reports decisions instead of jev for the approximate supplier fill of Spanish electricity invoices; it now fills only from a probability of 0.8 and never picks between two catalogue rows of the same company. Readings already stored keep jev: treat both values as an approximate fill.
  • New: files and their readings list their classifications in classifications (List File Readings, GET /v1/files, GET /v1/files/{file_id}, GET /v2/files): id, document_type, category, confidence_level, custom_label and the reading of each one. Every file has one classification today, so the list has one entry that repeats tags and reading; it is the groundwork for documents that bill several services, and no endpoint creates a second classification yet. When a file has several, tags will be [] and category and reading null: read classifications.
  • Changed: Create Records From File Reading takes every reading of one file, with supplier_id, stationary_fuel_id, refrigerant_fuel_id, custom_emission_factor_id and facilities on each reading. Three new errors: 409 FILE_READINGS_INCOMPLETE (a reading of the file is missing from the request, or one sent no longer exists because the file was reprocessed meanwhile; until now that save could answer 500), 409 FILE_READINGS_NOT_READY and 422 REVIEW_DECISIONS_MUST_BE_PER_READING. 400 MULTIPLE_FILES_NOT_SUPPORTED is now only for readings of two different files, and an empty file_readings is a 422. A request with the one reading of a file works as before.
  • Changed: a file’s document type is stored in its full form. Confirming a category sent as a bare word (electricity) stores invoice::electricity, so tags is ["invoice", "electricity"] where it was ["electricity"]; the last tag, and category, are the same. Types that were already complete are unchanged.
  • Changed: Retry File Processing with a detail.category the file is not classified as replaces the file’s classification before the processing is queued (it used to be relabelled when the processing finished), and answers 409 FILE_ALREADY_PROCESSED when the file already has records. Without detail.category nothing changes.
  • Changed: Update reading keeps country_code, customer_name and customer_vat of electricity, stationary combustion and water items when the request omits them, like issuer_legal_name and issuer_tax_id (until now an edit that left them out removed them from the reading). Send the key explicitly, even as null, to overwrite it.
  • New: POST /v2/invoices/electricity creates up to 5,000 electricity invoices per request with only the fields electricity has (kWh, supplier, CUPS, self-consumption, split across facilities), as an ingest job that calculates their emissions. The first of one create route per invoice category; the rest are listed below, and which route to use maps them.
  • New: POST /v2/invoices/fuels does the same for fuel invoices (stationary combustion, v1’s heat type): a fuel (stationary_fuel_id) or a custom factor, a combustion unit, and the supply point (cups).
  • New: POST /v2/invoices/refrigerants does the same for refrigerant recharges (fugitive emissions, v1’s recharge type): a gas (refrigerant_fuel_id) or a custom factor, in kg.
  • New: POST /v2/invoices/water does the same for water invoices: a water unit (m³, litre, gallon or kg), the supply point (water_supply_point) and the meter (water_meter_number).
  • New: POST /v2/invoices/district-heating and POST /v2/invoices/district-cooling do the same for energy bought from a district network (v1’s district_heat_and_steam type, one route per heat_cooling_type): kWh, MWh, MJ or GJ, and a supplier of the route’s family.
  • New: POST /v2/invoices/process does the same for process emissions (v1’s process type), which are always calculated with your own factor: custom_ef_record_id is required.
  • New: POST /v2/invoices/delete deletes up to 5,000 invoices of any category as an ingest job. An invoice split across facilities goes whole; invoices imported by Datadis are refused (DATADIS_MANAGED).
  • Changed: GET /lca/portfolio/{lca_id}/dashboard also lists top_contributors for the downstream stages, which until now returned an empty array: the destination country (A4), the consumption type (B6/B7) or the treated material (C4). Three stage labels are renamed so the dashboard, the Excel export and the report builder use one name per stage: A1 "Raw material supply" is now "Materials and components", A2 "Transport to manufacturer" is "Transport" and C4 "Waste treatment" is "Disposal".
  • Added: POST /v2/employees/survey-rows/export downloads the selected mobility survey rows as an Excel file, exactly as List Survey Rows shows them, in one sheet with fixed columns.
  • Breaking (BETA): POST /v2/facilities takes the surface unit as surface_area_unit_id, a surface unit of GET /v2/units (square meter, square foot or hectare; anything else is UNIT_NOT_SUPPORTED), instead of the surface_area_unit text. A surface_area without a unit is in square meters. type, logistic_factor and cups_list are no longer part of the request: every facility is created owned, with a logistic factor of 0.8, and its CUPS codes come from its invoices; sending them has no effect. The created facility no longer returns type, logistic_factor, surface_area_unit and the v1-only fields (organization_name, invoices_in_review_count, waste water lines…).
  • New: supplier_match.method on the items of a file reading (Get reading) gains the value org_history, for electricity and water: the supplier was filled from what your organization confirmed for the same issuer when saving earlier reviews. Such a match also carries votes and key (tax or name). Accept org_history wherever you read supplier_match.method.

2026-10-09

  • Breaking: activity dates must fall between 1970 and next year (until now any year was stored, e.g. 0100-01-01 or 2064-03-01, and calculated for that year). A date outside the range answers 422: Create Wastes (v2) rejects the batch with SCHEMA_INVALID; the v1 creates and updates of wastes, purchases, shipments (transport_date), vehicle consumptions, invoices, business travels, hotel stays and employee periods answer a validation error on the date (type: activity_date_out_of_range); logistics requests (shipment_date) and recharges (date) answer ACTIVITY_DATE_OUT_OF_RANGE, and their bulk endpoints report it per record. Create Waste and the invoice create and replace also require end_date on or after start_date, and Update Waste checks it against the stored date when only one of the two is sent (type: end_date_before_start_date). On the invoice replace, the shipment update and Update Waste only a date you change is checked: resending a stored date unchanged (e.g. a 1969 record) is still accepted. On the other updates (purchases, vehicle consumptions, business travels, hotel stays, employee periods, logistics, legacy transport routes) every date you send is checked: saving a record whose stored date is outside the range requires correcting that date. Files uploaded for these categories report such a row as an error. Records you already stored are not changed.
  • Changed: logistics requests created through the API (v1 and v2, single and bulk: Create, Create in bulk) and the legacy POST /api/v1/logistics/shipment calculator, sent with load_unit teu or feu, now count 10,000 kg per TEU and 20,000 kg per FEU (until now 20,000 and 40,000), the GLEC Framework / ISO 14083 default for a loaded container and the value file uploads already used. Requests stored before this date are not recalculated.
  • New: POST /v2/facilities creates up to 1,000 facilities in one request and answers 201 with them, in the order sent. All or nothing, idempotent (Idempotency-Key, with Idempotent-Replayed: true on a retry), dry_run. country is required and address is never geocoded.
  • New: POST /v2/facilities/delete deletes up to 1,000 facilities with their invoices, wastes and waste water treatments, as an ingest job (entity_type: facilities, operation: delete).
  • Deprecated: POST /v1/facilities and DELETE /v1/facilities/{facility_id}, superseded by the two above. They keep working and answer with Deprecation and Link: rel="successor-version" headers. No removal date yet.
  • Changed: /v2/facilities errors are problem details, like the rest of v2.
  • Changed: a purchase’s purchase_type (Create, Update, Get, List) follows the emission factor that prices it, and gains a third value, average_data: spend_based for EXIOBASE, average_data for a physical-unit database (e.g. ecoinvent), supplier_specific for a custom factor. The value you send is replaced by the one its factor implies, except on purchases priced with a custom_emission_factor_id: until now a purchase sent as supplier_specific without a custom factor came back supplier_specific; it now comes back with its factor’s type (spend_based by default). Accept average_data wherever you read or filter by purchase_type.
  • New: GET /v1/logistics/requests/unique-values supports field=vehicle_type: the TOC names (<vehicle>_<type>) your active shipments use, with counts, exactly as the vehicle_type[] filter of List Logistics Requests takes them. The new project_id parameter narrows them to the shipments linked to a project.

2026-10-08

  • Changed: shipments (Create, Update) accept a quantity_transported of up to 999,999,999.99999 (9 integer digits, 5 decimal places; until now 9,999,999.999) and a section kms_manual with up to 6 decimal places (until now 2). Bulk uploads keep 5 weight decimals and 6 km decimals instead of rounding them to 3 and 2.
  • New: GET /v2/imports/{import_id}/status returns sheets, every sheet name of the uploaded workbook, and Create Import Session returns them too when it resumes an existing session for the same file (until now sheets came empty there). Sessions created before this date return [].

2026-10-07

  • Breaking (BETA): unit_id is required on every record of POST /v2/wastes and must be kilograms (61743a63-ff70-459c-9567-5eee8f7dfd5c), the unit the app’s waste form uses. A waste priced with a custom_ef_record_id may also be in that record’s unit. A record without a unit (SCHEMA_INVALID) or with another one (new code UNIT_NOT_SUPPORTED) is rejected with the batch (422 BULK_RECORDS_REJECTED). Until now tonnes, grams and pounds were accepted and calculated: convert them to kilograms. Volume, energy, distance or currency units were stored and then failed as CALCULATION_FAILED.
  • Changed (BETA): ingest_job.finished is sent only for jobs submitted with an API key. A job a person starts while signed in to Dcycle (e.g. the app’s waste form) notifies that person in the app instead, and its source is the new value app; jobs sent with an API key keep api_bulk.
  • New: every webhook delivery records why its last attempt failed (last_error_code) and the start of what your server answered (last_response_body, up to 2 KB): see Debugging a failed delivery.
  • Fixed: timestamps of webhook endpoints and deliveries end with Z (UTC), like the rest of v2; they came without it.
  • Fixed: the app-wide rate limit answers 429 on the v2 resources as problem details (TOO_MANY_REQUESTS), not as {"detail": "Rate limit exceeded"}.
  • Changed: stationary combustion items of file readings carry printed_fuel_name, the fuel product as printed, kept by Update Reading when omitted. When the fuel read contradicts it (e.g. TECNODIESEL E+10, a diesel, read as petrol e10), the item’s stationary_fuel_id is null and the fuel must be chosen before the record calculates.
  • Changed: organizations below a project’s organization see its dashboard in GET /v2/dashboard and GET /v2/dashboard/{id} when they were invited to it, organization by organization, and only with the widgets of the views they may see. Projects that still use the old “share with child organizations” setting keep sharing with every organization below until that setting is saved again in the app.
  • Fixed (docs): the error-handling examples of Create Wastes (v2) read the rejected records from errors. Until now they read detail.records, which moved on 2026-10-05: if you copied them, update your code.
  • Fixed: the OpenAPI document declares every error as problem details (application/problem+json, schema Problem) instead of FastAPI’s default validation body, and its descriptions are written for integrators. Regenerate your client so it parses errors correctly.
  • Docs: Create Wastes (v2) lists where to find every id (facility, unit, LER and R/D codes), and the ingest job pages say what to do when a job stops making progress. The job status failed and the item status skipped are reserved: no job or item ends in them today.

2026-10-06

2026-10-05

  • Changed: the pending mobility survey export leaves out people who were most likely never asked about the period (ex-employees who left before it, and people uploaded afterwards in a list answered for other periods only), and a period longer than a year gets a sheet per year, counted from the period’s start.
  • Deprecated: POST /v1/wastes, superseded by POST /v2/wastes. It keeps working and now answers with Deprecation and Link: </v2/wastes>; rel="successor-version" headers. No removal date yet; it will be announced here with notice.
  • Deprecated: POST /v1/waste/bulk-delete, superseded by POST /v2/wastes/delete (asynchronous, idempotent). Same headers, no removal date yet.
  • New: Errors in v2 follow RFC 9457 problem details (application/problem+json) for /v2/wastes, /v2/ingest-jobs and /v2/webhook-endpoints. Breaking for those BETA endpoints: rejected records moved from detail.records to errors, and validation errors from detail to errors.
  • New: v2 timestamps are UTC with Z (2026-10-05T09:35:07.921Z).
  • New: GET /v2/ingest-jobs lists your jobs; POST /v2/wastes?dry_run=true validates without writing.
  • Fixed: X-RateLimit-Limit now reports the bucket capacity, so X-RateLimit-Remaining never exceeds it; the IETF RateLimit and RateLimit-Policy headers are added (Rate limits).
  • New: OpenAPI document of the v2 API, for SDKs and Postman.
  • New: POST /v2/wastes/delete: asynchronous bulk delete as an ingest job.
  • New: Retry Webhook Delivery, and api_version in every event.

2026-10-01

  • New: Webhooks: signed ingest_job.finished events with retries.
  • Breaking (BETA): Idempotency-Key is required on POST /v2/wastes; the same key with a different body is 422 IDEMPOTENCY_KEY_REUSED.

2026-09-29

  • New: POST /v2/wastes (BETA): up to 5,000 wastes per request, validated as a whole, calculated asynchronously as an ingest job.
  • Breaking (BETA): later the same day, POST /v2/wastes takes a custom emission factor as custom_ef_record_id, a record of your custom emission factor databases. It no longer accepts custom_emission_factor_id: send custom_ef_record_id instead.