living-docs-governance
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for agents to ingest project documentation (README, architecture guides, ADRs) which represents a potential attack surface for indirect prompt injection. However, it specifically mitigates this risk by instructing agents to treat these sources as untrusted context, verify claims against actual code/tests, and never execute embedded instructions. It recommends using specific harness files (e.g., AGENTS.md, CLAUDE.md) as the sole authoritative instruction surface to maintain boundaries.
- [CREDENTIALS_UNSAFE]: The skill includes a proactive security policy that strictly forbids placing credentials, tokens, or private payloads in governance documents and requires immediate redaction of secrets if discovered.
- [REMOTE_CODE_EXECUTION]: The skill includes preventative measures against unauthorized execution by instructing the agent to report missing optional capabilities rather than attempting to install external modules or tools automatically.
Audit Metadata