GET /label/lookup

Turns an ingredient designation as it appears on a label or a datasheet - a common name, an INCI name, a chemical denomination, a food-additive name - into the exact identifiers the EU labelling routes require: a CAS or EC number for /label/chem/classify and /label/chem/svhc, an INCI name for /label/cosmetic/ingredient, an E number for /label/food/additive. The answer comes from the referentials already ingested, with no network call and no secret at request time.

This route resolves an identity. It renders no regulatory verdict - those are the four routes it opens. See the live /catalog for the authoritative listing and price.

A none does not mean “unregulated”

Read this before anything else on the page, because the reader of a labelling route arrives with a compliance question. A none says one thing only: this text is absent from the ingested referentials, at their date. It is never a statement that the substance is unregulated, unclassified, or free to use.

The clearest proof is in the examples below. dimethyl ketone comes back none - and it is a common name for acetone, which carries a harmonised classification in Annex VI (q=acetone returns it). The referentials publish “acetone”, “propan-2-one” and “propanone”, not “dimethyl ketone”, and this route matches only what they publish. A none is a reason to find the designation the regulation uses, never a reason to conclude anything about the substance.

Only the designations the referentials publish

Matching uses only the names, synonyms and identifiers the referentials themselves publish - the ;-separated denominations of an Annex VI entry, the synonyms of a food additive, the INCI names of the cosmetics annexes and of CosIng. There is no synonym dictionary of our own, and that is a decision, not a gap: inventing one would mean selling our own referential as if it were official. Every candidate this route serves can be traced to a line a regulator or the Commission published.

Every answer carries that rule in the fixed sentence of data.limits, served verbatim:

A resolution names what the referential publishes for the text you sent, as of the date carried by the provenance; it does not certify that the entity resolved is the one you had in mind. A common name may designate a group of substances rather than a single one, and this resolution never chooses for you: every candidate found is listed. Matching is exact on the shared normalisation of the designations published by the referentials themselves — no fuzzy matching and no synonym dictionary of our own — so ‘none’ means the text is absent from the ingested referentials at that date, never that the substance is unregulated.

q is a designation, not an identifier: a CAS number or an E number sent as q (67-64-1, E330) comes back none, because the referentials publish those as identifiers, not as names. If the agent already holds one, it should call the target route directly.

The verdict, and why ambiguous is the point here

candidates is always present - empty on none, a one-element list on exact. All three verdicts are a 200, billed: a dated none is what stops an agent from inventing a CAS number.

ambiguous is common here, and it is the feature rather than the defect: one designation usually carries several natures of identifier - a CAS, an EC number, an INCI name, an E number - and they are all returned. One call to ?q=sodium nitrite opens the chemical, the cosmetic and the food route at once. A common name may also designate a group of substances (petroleum gas matches 182 Annex VI entries); the service never picks one.

Five natures, each naming what it opens

Each candidate’s id is the identifier the downstream route accepts as is, and each candidate lists in endpoints the routes that identifier opens - those of the referentials that actually returned it, never a theoretical list:

A name candidate appears when a referential publishes an entry without a usable identifier: a placeholder cell (-) or a composite cell listing several numbers (13463-67-7/1317-70-0/1317-80-2, the form the Cosmetics Regulation publishes for titanium dioxide). Those cells are never served as an id - the downstream route would reject them, and splitting them would be rewriting the source. They are counted in data.source_identifiers_not_usable, and the entry stays reachable through the denomination its referential indexes. A candidate no route would resolve is never served.

Five referentials, and one of them is not binding

CosIng is information_only, and every response says so on its line in data.referentials. It resolves an identity - it is the path from an INCI list on a cosmetic label to an identifier, and many everyday ingredients (Aqua, Glycerin) are only there, because the annexes list regulated substances only. But it carries no legal force: only the four binding publications do. A candidate whose sources is only cosing_index is an identity to look up, never a verdict - the verdict is what /label/cosmetic/ingredient returns for it.

The health and nutrition claims register behind /label/claims/check is never queried here: it holds wordings, not substance identifiers.

Parameters

GET /label/lookup?q=sodium%20nitrite&domain=food

• Normalisation is the referentials’ own. Case and punctuation are folded (punctuation becomes a word separation); stereochemical and positional prefixes and multipliers are kept, because dropping them would make distinct substances converge.
• domain reduces, never reshapes: it narrows the referentials queried (every referential in scope is still listed in data.referentials).
• product_type narrows cosmetic annex entries only - which is why it requires domain=cosmetic: a filter that silently had no effect would return a wider list than asked without saying so. It follows the same rule as /label/cosmetic/ingredient: an entry with no product-type condition - an Annex II prohibition, notably - is never dropped. Hiding a prohibited substance because a product type was given would be the worst possible defect here.
• No pagination. The list is capped at 25; total counts every candidate before the cap and truncated says whether it was cut. Narrow with domain instead of paging - a second page would be a second paid call for the same question. A truncated list is never exact.

200 response - UnifiedResponse

{
  "data": { ... },
  "provenance": {
    "source": "ECHA Annex VI (CLP), harmonised classification and labelling; ECHA Candidate List of SVHC (REACH Art. 59(10)); Regulation (EC) No 1223/2009 on cosmetic products — consolidated text (EUR-Lex), Annexes II to VI; CosIng (European Commission) via public EU Search API; Regulation (EC) No 1333/2008 on food additives — consolidated text (EUR-Lex), Annex II (Union list by food category) and Annex III",
    "fetched_at": "2026-10-15T15:41:31.717766173Z",
    "freshness": { "kind": "snapshot", "as_of": "2026-05-01T00:00:00Z" }
  }
}

• provenance.source: every referential queried.
• freshness.as_of: the oldest of their dates - the per-referential date and regulatory version live in data.referentials.

Fields of data

Each candidate:

Example - exact

GET /label/lookup?q=Advantame:

{
  "data": {
    "query": "Advantame",
    "verdict": "exact",
    "candidates": [
      {
        "id": "E969",
        "id_kind": "e_number",
        "label": "Advantame",
        "sources": ["food_additives"],
        "endpoints": ["/label/food/additive"]
      }
    ],
    "total": 1,
    "truncated": false,
    "referentials": [
      { "referential": "annex_vi", "authority": "binding", "searched": true, "source": "ECHA Annex VI (CLP), harmonised classification and labelling", "as_of": "2026-05-01", "version": "Annex VI to CLP — harmonised entries in application (latest adaptation: ATP22)" },
      { "referential": "candidate_list", "authority": "binding", "searched": true, "source": "ECHA Candidate List of SVHC (REACH Art. 59(10))", "as_of": "2026-10-12" },
      { "referential": "cosmetics", "authority": "binding", "searched": true, "source": "Regulation (EC) No 1223/2009 on cosmetic products — consolidated text (EUR-Lex), Annexes II to VI", "as_of": "2026-05-01", "version": "02009R1223 — EN — 01.05.2026" },
      { "referential": "cosing_index", "authority": "information_only", "searched": true, "source": "CosIng (European Commission) via public EU Search API", "as_of": "2026-05-01" },
      { "referential": "food_additives", "authority": "binding", "searched": true, "source": "Regulation (EC) No 1333/2008 on food additives — consolidated text (EUR-Lex), Annex II (Union list by food category) and Annex III", "as_of": "2026-08-18", "version": "02008R1333 — EN — 18.08.2026" }
    ],
    "limits": "A resolution names what the referential publishes for the text you sent, as of the date carried by the provenance; it does not certify that the entity resolved is the one you had in mind. A common name may designate a group of substances rather than a single one, and this resolution never chooses for you: every candidate found is listed. Matching is exact on the shared normalisation of the designations published by the referentials themselves — no fuzzy matching and no synonym dictionary of our own — so 'none' means the text is absent from the ingested referentials at that date, never that the substance is unregulated."
  },
  "provenance": {
    "source": "ECHA Annex VI (CLP), …; Regulation (EC) No 1333/2008 on food additives — …",
    "fetched_at": "2026-10-15T15:41:31.717766173Z",
    "freshness": { "kind": "snapshot", "as_of": "2026-05-01T00:00:00Z" }
  }
}

A sweetener known only to the food-additive Union list: one identifier, one route. E969 goes straight into /label/food/additive?e_number=E969.

Example - ambiguous, three routes in one call

GET /label/lookup?q=sodium%20nitrite (referentials, limits and provenance as above):

{
  "data": {
    "query": "sodium nitrite",
    "verdict": "ambiguous",
    "candidates": [
      {
        "id": "7632-00-0",
        "id_kind": "cas",
        "label": "sodium nitrite",
        "sources": ["annex_vi", "cosmetics", "cosing_index"],
        "endpoints": ["/label/chem/classify", "/label/cosmetic/ingredient"]
      },
      {
        "id": "231-555-9",
        "id_kind": "ec",
        "label": "sodium nitrite",
        "sources": ["annex_vi", "cosmetics", "cosing_index"],
        "endpoints": ["/label/chem/classify", "/label/cosmetic/ingredient"]
      },
      {
        "id": "Sodium nitrite",
        "id_kind": "inci",
        "label": "Sodium nitrite",
        "sources": ["cosmetics", "cosing_index"],
        "endpoints": ["/label/cosmetic/ingredient"]
      },
      {
        "id": "E250",
        "id_kind": "e_number",
        "label": "Sodium nitrite",
        "sources": ["food_additives"],
        "endpoints": ["/label/food/additive"]
      }
    ],
    "total": 4,
    "truncated": false,
    "referentials": [ "…" ],
    "limits": "A resolution names what the referential publishes for the text you sent, …"
  },
  "provenance": { "…": "…" }
}

Four identifiers of one substance, each naming what it opens: the CAS and EC numbers reach the CLP harmonised classification and the cosmetics annexes, the INCI name reaches the cosmetics annexes, E250 reaches the food-additive Union list. Which one to follow depends on the product the agent is labelling

• the lookup does not guess it.

Example - none

GET /label/lookup?q=dimethyl%20ketone (referentials as above, all five searched: true):

{
  "data": {
    "query": "dimethyl ketone",
    "verdict": "none",
    "candidates": [],
    "total": 0,
    "truncated": false,
    "referentials": [ "…" ],
    "limits": "A resolution names what the referential publishes for the text you sent, as of the date carried by the provenance; it does not certify that the entity resolved is the one you had in mind. A common name may designate a group of substances rather than a single one, and this resolution never chooses for you: every candidate found is listed. Matching is exact on the shared normalisation of the designations published by the referentials themselves — no fuzzy matching and no synonym dictionary of our own — so 'none' means the text is absent from the ingested referentials at that date, never that the substance is unregulated."
  },
  "provenance": {
    "source": "ECHA Annex VI (CLP), …; Regulation (EC) No 1333/2008 on food additives — …",
    "fetched_at": "2026-10-15T15:41:31.646002901Z",
    "freshness": { "kind": "snapshot", "as_of": "2026-05-01T00:00:00Z" }
  }
}

A 200, billed - and the substance is regulated: “dimethyl ketone” is acetone, an Annex VI entry published as acetone; propan-2-one; propanone. q=propanone resolves it (CAS 67-64-1, EC 200-662-2). This is the none the page warned about: the absence of a designation, never the absence of a regulation.

Errors

Only requests the service cannot answer leave the 200 range - none of them are billed. There is no 404 for a designation the referentials do not publish: that is a none.

{ "code": "INVALID_INPUT", "error": "'product_type' narrows cosmetic annex entries only: pass it together with 'domain=cosmetic'" }

{ "code": "INVALID_INPUT", "error": "unknown domain: 'pharma' (expected one of 'chem', 'cosmetic', 'food')" }

Attribution

• Annex VI of the CLP Regulation and the Candidate List of Substances of Very High Concern (REACH Article 59(10)), published by ECHA (legal notice - commercial reuse permitted).
• Regulation (EC) No 1223/2009 on cosmetic products and Regulation (EC) No 1333/2008 on food additives, consolidated texts published on EUR-Lex (European Union).
• The CosIng index of the European Commission, used for identity resolution only.

See also

• GET /label/chem/classify - the harmonised CLP classification behind a CAS or EC number.
• GET /label/chem/svhc - whether a substance is on the REACH Candidate List.
• GET /label/cosmetic/ingredient - the cosmetics annexes’ verdict for an INCI name, CAS or EC number.
• GET /label/food/additive - the Union-list authorisation behind an E number.
• GET /company/lookup, GET /vehicle/lookup and GET /medication/lookup - the same lookup contract for companies, vehicles and medications.
• For agents - discovery surfaces, the live /catalog and how settlement works.