Validate Claim
validate_claimCheck a claim graph before relying on or presenting its conclusions. Refetch cited provisions and evidence under your scope, compare quotes, inspect dependencies, and report the result with explicit checked/not_checked lists. Evidence checks and judgment are separate fields on every node: evidence_checks (source_resolution, quotation_match, served_language, dependencies) records what the checker established; agent_assessment.declared_typing records the DET/INT typing the agent declared, attributed to the agent; evidence_ceiling is the highest typing the checked evidence permits; claim_support is always not_checked, because whether evidence supports a claim is reasoning and this check evaluates none. A matched quotation and a DET evidence_ceiling do not mean the conclusion is supported. On interpretations and conclusions, evidence_checks.stated_quantities reports whether each plain deadline, duration, percentage or euro amount in the statement appears in the cited text (all_in_cited_text, not_in_cited_text, not_checked, not_applicable; English only; stated_quantities_detail lists the missing and cited values); not_in_cited_text is a fact about the text, not a finding that the conclusion is false (a value may be derived), and claim support stays not_checked. strength is the legacy name for evidence_ceiling. A resolved evidence node (non-provision) carries a resolution verdict and caps every dependent conclusion at UNVERIFIED. Optionally check quotations in the prose answer. This does not verify legal in-force status, facts, satisfiability, entailment, or whether logic expresses the law. Served-to-this-caller verification (G4) is not checked: served rows do not yet carry request-id binding. No overall pass verdict is returned. Refetches run with concurrency up to 6 under a shared 25-second deadline; the session pool defaults to 2 sessions per downstream URL. Expiry returns bounded=true and unreached_clauses (count); unfinished clauses are unavailable, UNVERIFIED, and in not_checked with reason refetch_budget_exhausted. The summary.display_markdown field contains the presentation text, with findings and disclosures in summary.attention_items.
Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| graph | Yes | The claim graph as an object (a JSON-encoded string is also accepted), schema ansvar.claim-graph.v2 (or v1). Minimal working example: {"schema": "ansvar.claim-graph.v2", "question": "Must a controller notify a personal data breach?", "nodes": [{"id": "C1", "kind": "clause", "canonical_ref": "GDPR:art_33", "lookup": {"source_id": "eu-regulations", "canonical_ref": "GDPR:art_33", "jurisdiction": "EU"}, "quote": "In the case of a personal data breach, the controller shall without undue delay and, where feasible, not later than 72 hours after having become aware of it, notify the personal data breach to the supervisory authority competent in accordance with Article 55...", "quote_language": "en", "binding": "mandate"}, {"id": "K1", "kind": "conclusion", "statement": "The controller must notify the supervisory authority.", "typing": "INT", "depends_on": ["C1"]}]}. Required keys: schema, question (text) and nodes (array). Optional: variables (object mapping names to {kind: design|fact, meaning: text}; default {}) and gaps (array of {statement, reason}; default []). Every node needs id and kind. A clause needs canonical_ref, quote, quote_language and binding (mandate|prohibition|permission|designation|definition); jurisdiction is required unless lookup is supplied; pinpoint is optional. Copy each row's citation.lookup args into lookup. In v2, each clause lookup requires source_id and canonical_ref equal to the node reference. A v2 evidence node needs source_id (the routed citation.source_id), reference (the exact served address), and jurisdiction when the source is registered under several codes; quotation is optional. Missing address fields refuse before refetch with the node field path. Regulatory updates use source_id=reg-intel; CVEs use cve-intel. Pan-European preparatory works remain unsupported and explicitly refuse resolution. A fact needs statement and provenance="customer"; an assumption needs statement and provenance="assumed". An interpretation or conclusion needs statement, typing (DET|INT), and depends_on (node ids). Optional node constraint is CNF (arrays of literal strings; ! negates a variable). Conclusions may carry entails {assume: literal array, expect: SAT|UNSAT}, retained without evaluation. Optional regulatory_basis_unresolved (boolean) is preserved. Limits: 100 nodes, 60 citation refetches, 131072 UTF-8 graph bytes, 25 seconds for refetching. A large graph citing one corpus takes tens of seconds; if the refetch budget runs out the response is marked bounded, names how many clauses it did not reach, and leaves them unverified. | |
| answer | No | Optional prose answer whose quotations should be compared with refetched evidence. Maximum 65536 UTF-8 bytes (MAX_CLAIM_ANSWER_BYTES). |