← All tools

agent-claim-output-verifier

Map an autonomous agent claim or output to supplied evidence and return SUPPORTED, PARTIALLY_SUPPORTED, NOT_PROVEN, or CONTRADICTED with explicit unsupported elements, contradictions, missing evidence, and recommended verification.

Price

$0.45 USDC equivalent

Endpoint

POST /v1/agent-claim-output-verifier

Network

Base mainnet · eip155:8453

Asset

USDC

MCP tool

agent-claim-output-verifier

External payability

PASS · 2026-10-08T19:55:01.298Z

When an agent should use it

• verify an AI agent claim against evidence
• check whether agent output is supported or contradicted
• map autonomous agent claims to evidence
• detect unsupported claims in agent output

Input schema

{
  "type": "object",
  "properties": {
    "claim": {
      "type": "string",
      "minLength": 1,
      "maxLength": 12000
    },
    "evidence": {
      "type": "array",
      "maxItems": 64,
      "items": {
        "oneOf": [
          {
            "type": "string",
            "minLength": 1,
            "maxLength": 8000
          },
          {
            "type": "object",
            "properties": {
              "id": {
                "type": "string",
                "maxLength": 256
              },
              "statement": {
                "type": "string",
                "minLength": 1,
                "maxLength": 8000
              },
              "relation": {
                "type": "string",
                "enum": [
                  "SUPPORTS",
                  "CONTRADICTS",
                  "NEUTRAL"
                ]
              },
              "source": {
                "type": "string",
                "maxLength": 1000
              },
              "observed_at": {
                "type": "string",
                "maxLength": 128
              }
            },
            "required": [
              "statement"
            ],
            "additionalProperties": false
          }
        ]
      }
    },
    "context": {
      "type": "string",
      "maxLength": 8000
    },
    "required_standard": {
      "type": "string",
      "maxLength": 2000
    }
  },
  "required": [
    "claim"
  ],
  "additionalProperties": false
}

Example request

{
  "claim": "The deployment is healthy and payment reconciliation succeeded.",
  "evidence": [
    {
      "id": "health",
      "statement": "Health endpoint returned 200",
      "relation": "SUPPORTS"
    },
    {
      "id": "stripe",
      "statement": "No reconciliation result was collected",
      "relation": "NEUTRAL"
    }
  ],
  "required_standard": "Every material claim element must have direct evidence"
}

Example output

{
  "verdict": "PARTIALLY_SUPPORTED",
  "confidence": 0.5,
  "supported_elements": [
    "The deployment is healthy"
  ],
  "unsupported_elements": [
    "payment reconciliation succeeded"
  ],
  "contradictions": [],
  "missing_evidence": [
    "Independent evidence supporting or contradicting: payment reconciliation succeeded"
  ]
}

x402 flow

POST /v1/agent-claim-output-verifier
→ HTTP 402 + PAYMENT-REQUIRED
→ buyer authorizes locally
→ retry same request with PAYMENT-SIGNATURE
→ one governed execution + machine-readable JSON

Discovery: OpenAPI · x402 · llms.txt

A valid 402/payability probe is operational evidence, not proof of independent demand, future settlement, or semantic correctness.