guard-gate-schema.md

Guard Gate Schema

Guard modules emit normalized findings and gates.

.agent-context/runs/<task-id>/iterations/<nnn>/
  guard.findings.json
  guard.gates.json
  decision.json

Finding

interface GuardFinding {
  id: string;
  source: "policy" | "hallucination" | "regression";
  kind: "forbidden" | "required" | "risk" | "info";
  status: "failed" | "missing" | "warning" | "satisfied" | "passed";
  severity: "error" | "warning" | "required" | "info";
  message: string;
  evidence: string[];
  requiredCommands: string[];
}

Gate

interface GuardGate {
  id: string;
  guard: "context" | "boundary" | "evidence" | "hallucination" | "regression";
  blocking: boolean;
  action: "repack" | "rollback" | "human-review" | "run-tests" | "repair" | "block";
  evidence: string[];
  findingIds: string[];
}

Gates, executor failures, policy findings, Loop decisions, and risk signals are normalized into HarnessDecisionCandidate records before arbitration. Candidate ordering is deterministic and uses the shared action priority:

rollback > block > repack > repair > run-tests > human-review > finalize

The highest-priority candidate becomes the selected action. Remaining candidates are retained as supporting candidates; their reasons, required commands, and artifacts are merged into the final decision with stable deduplication. decision.json and the orchestrator Markdown report record the selected candidate, selected priority, and supporting candidates, so the result does not depend on Guard Gate array order.