OpenChainGraph · Cluster ③ AI Governance · Chain

Agent Audit Trail Conformance

Two-step linear chain: build an EU AI Act Art 12(2)-conformant AI decision log record then validate the IETF draft-sharif-agent-audit-trail-00 chain linkage. End-to-end agent audit trail conformance for high-risk FS AI systems.

EU AI Act Art 12(2) IETF AAT Linear · 2 steps OCG v0.8.0 Cluster ③ AI Governance
Computing…
Pending
DAG · 2 steps · art-236 → art-237 · no gate
STEP 1 Art 12(2) Decision Log Record
Builds an EU AI Act Art 12(2)-conformant AI decision log record. Captures model identity, input/output SHA-256 digests, override flags, retention months (≥6), and chain position. The output record digest (execution_hash) is passed downstream as sha256_prev_record to the IETF AAT validator, establishing cryptographic chain linkage between the two steps.
→ Handoff: execution_hash of this record becomes sha256_prev_record for IETF AAT validator at Step 2
🔗 art-236-build-ai-decision-log-record
MCP Call · build_ai_decision_log_record
{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "build_ai_decision_log_record",
    "arguments": {
      "model_id": "credit-score-model-v3",
      "model_version": "3.2.1",
      "decision_type": "CREDIT_APPROVAL",
      "input_digest": "sha256:e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
      "output_digest": "sha256:a665a45920422f9d417e4867efdc4fb8a04a1f3fff1fa07e998e86f7f7a27ae3",
      "decision_outcome": "APPROVED",
      "human_override": false,
      "chain_position": "first",
      "retention_months": 12,
      "subject_ref": "SYNTHETIC-REF-001"
    }
  },
  "id": 1
}
STEP 2 IETF AAT Chain-Linked Validator
Validates the IETF draft-sharif-agent-audit-trail-00 chain-linked audit record. Checks action_class, outcome, and trust_level enum membership; validates the sha256_prev_record 64-hex linkage format; verifies ECDSA signature presence. Emits conformance_result (CONFORMANT / PARTIAL / NON_CONFORMANT / EMPTY_INPUT) and aat_completeness_score (0–1). Terminal stage.
🏁 Terminal stage · emits conformance_result and aat_completeness_score
🔗 art-237-validate-agent-audit-trail
MCP Call · validate_agent_audit_trail
{
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "name": "validate_agent_audit_trail",
    "arguments": {
      "agent_identity": "did:web:credit-score-model.example.com",
      "timestamp": "2026-07-04T12:00:00Z",
      "action_class": "AUTHORIZE",
      "outcome": "SUCCESS",
      "trust_level": "HIGH",
      "sha256_prev_record": "a3b4c5d6e7f8a9b0c1d2e3f4a5b6c7d8e9f0a1b2c3d4e5f6a7b8c9d0e1f2a3b4",
      "ecdsa_signature": "0x3045...",
      "reason": "Credit score 750 exceeds approval threshold 680"
    }
  },
  "id": 2
}
Export Chain Artifact
Download a §4-compliant chain artifact with execution hash, step definitions, and regulatory basis. Hash is deterministic over canonical {policy_parameters, output_payload} per RFC 8785/JCS.