{"uid":"cap_VS3k9X8jH3xIikkPslnTh","slug":"salt19-agent-claim-output-verifier-07a6bd18","name":"SALT19 Agent Claim Output Verifier","description":"SALT19 is an independent applied AI systems lab behind EvoMind governed cognition, the MCP-native Agent Utility Grid, the ARCS research community, AeroClear UAS flight intelligence, and practical software for real-world work.","url":"https://api.salt19.com/v1/agent-claim-output-verifier","method":"POST","headers":{},"bodySchema":{"type":"object","properties":{"claim":{"type":"string","maxLength":12000,"minLength":1},"context":{"type":"string","maxLength":8000},"evidence":{"type":"array","items":{"oneOf":[{"type":"string","maxLength":8000,"minLength":1},{"type":"object","required":["statement"],"properties":{"id":{"type":"string","maxLength":256},"source":{"type":"string","maxLength":1000},"relation":{"enum":["SUPPORTS","CONTRADICTS","NEUTRAL"],"type":"string"},"statement":{"type":"string","maxLength":8000,"minLength":1},"observed_at":{"type":"string","maxLength":128}},"additionalProperties":false}]},"maxItems":64},"required_standard":{"type":"string","maxLength":2000}}},"responseSchema":{"type":"json","example":{"verdict":"PARTIALLY_SUPPORTED","confidence":0.5,"contradictions":[],"missing_evidence":["Independent evidence supporting or contradicting: payment reconciliation succeeded"],"supported_elements":["The deployment is healthy"],"unsupported_elements":["payment reconciliation succeeded"]}},"example":null,"exampleRequest":null,"tags":["x402"],"displayCostAmount":"0.45","displayCostAsset":"USDC","priceDynamic":false,"priceHint":null,"priceStatus":"priced","priceSource":"probe","requiresHandshake":false,"reviewCount":0,"rating":{"score":"0.00","successRate":"0.00","reviews":0,"stars":null,"state":"unrated"},"availabilityStatus":"unknown","priceObserved":null,"sessionDeposit":null,"pricing":{"kind":"static","summary":"$0.45/call","primary":{"kind":"static","protocol":"x402","network":"base","amountUsd":"0.45","per":"call","confidence":"exact"},"accepted":[{"kind":"static","protocol":"x402","network":"base","amountUsd":"0.45","per":"call","confidence":"exact"}]},"paymentMethods":[{"uid":"pm_Or9ilqxZ9PLmJp6UWw73u","protocol":"x402","methodType":"crypto","chain":"base","mode":"charge","costAmount":"0.45","costPer":"request","priority":0,"asset":"0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913","unit":"request","depositMicros":null,"planRef":null}],"brandName":null,"brandSlug":null,"brandBaseUrl":null,"brandDocsUrl":null,"whatItDoes":"Verifies whether an agent-generated claim is supported, partially supported, or contradicted by a given set of evidence, returning a structured verdict with confidence score and breakdown.","exampleAgentPrompt":"Check whether the claim 'the deployment is healthy and payment reconciliation succeeded' is supported by this evidence — logs showing healthy pods but no payment confirmation records — using a 'preponderance of evidence' standard, and tell me what's supported, what's contradicted, and what evidence is missing.","exampleUseCases":[{"title":"Auditing autonomous agent task output","prompt":"My agent just claimed it successfully completed a data migration and all records transferred without error — here are the logs it had access to. Can you verify whether that claim is actually supported by the evidence, tell me the confidence level, and flag anything that's missing or contradicted?"},{"title":"Validating AI-generated compliance assertion","prompt":"Verify whether the claim 'all KYC checks passed for this batch of users' is supported by the attached audit trail entries. Use a 'clear and convincing evidence' standard and tell me which elements are backed up, which aren't, and what contradictions exist."},{"title":"Checking agent reasoning in multi-step workflow","prompt":"In my agent pipeline, step 3 concluded that 'the vendor invoice matches the purchase order and is approved for payment.' Here's the supporting context and the raw invoice and PO data as evidence. Can you run this through the claim verifier and give me a verdict with confidence score and any unsupported elements?"}],"resultDescription":"A JSON object containing: a verdict string (e.g. SUPPORTED, PARTIALLY_SUPPORTED, CONTRADICTED, UNSUPPORTED), a confidence score between 0 and 1, a list of contradictions found in the evidence, a list of missing evidence items needed to fully support the claim, a list of supported claim elements, and a list of unsupported claim elements.","failureModes":["Claim text or evidence items exceed maximum character limits, causing validation error","Evidence array exceeds 64 items limit","Missing required 'claim' field returns 400 error","Ambiguous or vague claims may result in low-confidence PARTIALLY_SUPPORTED verdicts with broad missing_evidence lists","Required standard field not interpretable may result in generic evaluation criteria being applied","Payment of 0.5 USDC not completed causes 402 Payment Required response"],"whenToPreferThis":"Choose this endpoint when an AI agent needs to audit or sanction its own outputs or another agent's claims before acting on them — especially in governed, high-stakes, or multi-step agentic workflows where unverified assertions could cause errors downstream. Particularly useful when structured evidence items with source attribution and temporal metadata are available, and when a specific evidentiary standard (e.g. preponderance, clear and convincing) should be applied. Prefer this over general LLM fact-checking when you need a structured verdict with confidence scores and a broken-down analysis of supported vs. unsupported claim elements.","instructions":null,"reviewSummary":null,"reviewSummaryHighlights":null,"reviewSummaryConcerns":null,"reviewSummaryGeneratedAt":null,"activationCount":0,"lastUsedAt":null,"lastSuccessfullyRanAt":null,"lastHealthCheckAt":"2026-09-15T00:39:52.579Z","isFirstParty":false}