agent-shared-memory
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a robust 'Propose-Decide-Apply' workflow that prevents the AI from silently modifying project memory. All state changes require explicit human approval.
- [SAFE]: It uses cryptographic measures (HMAC-SHA256) to authorize decisions, explicitly stating that the signing secret must not be exposed to the delegated executor (the AI).
- [SAFE]: The memory model is append-only, ensuring that older records remain auditable and preventing data deletion or tampering without an audit trail.
- [SAFE]: The skill handles concurrency safely using atomic file operations and locking mechanisms to prevent race conditions during memory updates.
- [SAFE]: External references and artifact policies are clearly defined, using a
.coord/scratch directory that is gitignored by default to prevent accidental exposure of coordination metadata.
Audit Metadata