agentgovernance/v1

Open file-format spec for approvals, audit logs, and autonomy policies. Vendor-neutral, third-party verifiable. Hash-chained.

agentgovernance/v1

agentgovernance/v1 is the trust layer — a vendor-neutral, filesystem-first, third-party-verifiable file format for recording approvals, append-only audit logs, and autonomy policies.

Domain-agnostic by design: the doctypes reference no business concepts. Any workflow that needs auditable decisions — clinician overrides, approve-to-publish flows, AI-agent action gates, board votes — adopts this spec directly.

The canonical spec is AIP-7 at agentproto.sh.

Doctypes

DoctypeFile pathPurpose
signature<artifact>/../signatures/<signer>-<isoDate>.signature.jsonUniversal approval primitive (one event per file)
audit-event<scope>/audit/audit-log.jsonl (one line per event)Append-only hash-chained event log
policy<scope>/policies/<slug>/POLICY.mdDeclarative autonomy rule

Hash-chain protocol

The audit log is end-to-end verifiable by any third party using the published hash-chain protocol. Tampering, truncation, and replay are all detectable. Implementations exist in TypeScript today; verifiers can be written in any language.

Conventions

  • Markdown canonical with YAML frontmatter (for POLICY.md)
  • JSON canonical for signature.json and audit-log.jsonl lines
  • Slug-based references, never database IDs
  • schema: agentgovernance/v1 on every doctype
  • Vendor-specific extensions under metadata.<vendor>.*
  • Git-native workspace layout

Sibling specs

Reference implementation

packages/governance/core — parser, validator, hash-chain implementation, policy evaluator.

Get involved