Verify Citations
verify_citationsVerify submitted legal citations against what this gateway actually serves, so a report can record which of its citations are established evidence and which are not. For each subject you supply a reference (and the source_id the row's citation.lookup hint carried, plus the quotation your text relies on), the gateway re-retrieves that provision under YOUR scope through the source-pinned exact lookup, checks the PRODUCER-declared identity of the row it got back against the reference you asked for, and only then compares your quotation with the retrieved body. A displayed reference is never used to decide which row answers a subject. Verdicts per subject: quote_checked (reference resolved and the quotation is in the body), reference_resolved (reference resolved, no quotation was submitted), quote_mismatch, no_quotation_submitted (a quote-bearing subject whose quotation was empty — a refusal, NOT a mismatch), identity_mismatch (the row's producer identity names a different provision), identity_absent, reference_unparseable, not_served, withheld and unavailable. Coverage is advertised, never assumed: producer identity exists TODAY ONLY on the source-pinned exact lookup paths, so a subject with no source_id, or one on a corpus they do not cover, returns identity_absent rather than passing. Entity subjects are the second such path: when source_id names a controls, framework, standards or threat-technique catalog, reference is that row's entity id (a control, clause or technique id such as AC-4(5), A.5.1 or T1059) and the gateway re-retrieves it from that catalog by id — the lane follows the source, not the shape of the string. Court decisions and preparatory works are the third: when source_id names a decision or preparatory-works corpus, reference is that row's single opaque address (an ECLI, a neutral citation, the corpus's own case id, a proposition number) and the gateway re-retrieves it FROM THAT CORPUS — never from another decision corpus of the same jurisdiction — and additionally requires the producer's own court, case number and decision date before a decision verifies. Both lanes compare references byte for byte: these id spaces carry no equivalent spellings. For those two lanes source_id is the ROUTED CORPUS ID the row came from (for example dutch-court-decisions or swedish-preparatory-works), because a decision hint carries the reference and the jurisdiction, not the corpus. It is not the citation's mcp_server, which is that corpus's display name and resolves to no routed id; the response's coverage block lists the corpus ids each of these lanes covers today. Regulatory updates are the fourth: when source_id names the regulatory-update service, reference is the record id that service emitted and the gateway re-fetches that record from it, byte for byte. Because a regulatory update is a live feed rather than a built corpus, such a verdict also carries a freshness block — the record's own publication and observation dates, the feed it came from and that service's own reported data age — and a verdict this gateway cannot date refuses (identity_absent) instead of establishing. A quotation on this lane is compared with the publisher's own title text, which is the only verbatim text the record carries: the service serves no summary and no provision body. CVE records are the fifth: when source_id names the CVE service, reference is the CVE id (CVE-2021-44228) and the gateway re-fetches that record from it, byte for byte, comparing a quotation with the publisher's own description. Such a verdict carries a freshness block too — the record's publication and modification dates, and the declared sync state, age and interval of the feed it was served from — and because that service publishes its own freshness contract, a feed it reports as no longer current, or one older than the sync interval it declares for itself, answers identity_absent with reason cve_pin_feed_stale instead of establishing anything — and so does a miss from such a feed, because an absence read from a stale mirror is not an absence. A KEV listing or an EPSS score is NOT verified on that lane: they are other feeds' assertions about the CVE rather than publisher text of it, and a quotation of one reads quote_mismatch. Their two feeds' ages are stated on the verdict but do not refuse it — the one deliberate exception to the stale-feed rule, scoped to feeds the checked text does not come from. The sanctions lane pins OFAC list/entry/revision lookups at sanctions-intel, checks quotations against publisher names and aliases, and requires fresh list and entry data. The response's coverage block lists which corpora each lane covers today; a subject on a corpus a lane does not yet cover answers identity_absent. A source_id that names a source alias (a row served through edpb or uk-law carries the alias as its citation.source_id) is resolved to the alias's corpus under the alias's own tier; the verdict keeps your source_id and names that corpus in resolved_source_id. First-party doc:// document pinpoints are not verified here (reference_unparseable). withheld means licensing or tier suppression and never means the law does not exist; not_served means no corpus in your scope serves that reference. No verdict asserts legal in-force status, currency, pinpoint correctness, or that the provision supports your conclusion. Limits: 60 subjects a call, one lookup a subject under a shared 25-second budget with concurrency 6; on exhaustion the response is marked bounded and the unreached subjects are unavailable, never verified and never reported as absent law. Lookups run under your own tier and entitlements and count against your normal lookup quota. With attest=true, premium and above receive signed evidence attestations for permitting subjects: integrity and provenance, never legal validity; free and solo keep ordinary verdicts with attestation_refused=tier.
Tool call template
Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.
{
"id": 1,
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"arguments": {
"subjects": "<subjects>"
},
"name": "verify_citations"
}
}Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| attest | No | Request signed evidence attestations (premium and above); defaults to false. | |
| subjects | Yes | Array of subject objects, each checked independently and echoed back in this order. Keys: subject_id (required, your own opaque id, unique within the call, returned verbatim); reference (required, the canonical reference to verify, e.g. the canonical_ref a search row's citation.lookup hint carried); source_id (optional but needed for a verifiable result — the routed corpus id from that same lookup hint, or the row's citation.source_id when that names a source alias such as edpb; without it the lookup is unpinned, carries no producer identity and the verdict is identity_absent); quotation (optional — supply the exact text your report relies on and it is compared with the retrieved body; omit the key entirely for a reference-only claim. Supplying it empty declares a quote-bearing claim with nothing to compare and returns no_quotation_submitted. A quotation shorter than 25 characters (6 for Chinese, Japanese or Korean text), elision marks excluded, is too short to be evidence and returns no_quotation_submitted too, unless it is the provision's whole text; quote a full clause. language (optional — the edition language of a secondary-language row, such as bg or nl, exactly as the lookup that served it named it; the quotation is then compared with that edition. Omit it for the default edition). Maximum 60 subjects a call; a quotation is capped at 16384 UTF-8 bytes. For calendar rows use source_id reg-intel, reference equal to the obligation_id, and identity_shape obligation_id. identity_shape record_id (or omitted) selects publication replay; the two id grammars can overlap. Unknown keys are refused, not ignored. |
Output Schema
| Name | Required | Description | Default |
|---|---|---|---|
| result | Yes |