agentsop-repo-state-gating
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill consists exclusively of markdown documentation and a JSON decision artifact. It does not perform any automated code execution, network requests, or sensitive file modifications.\n- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze repository metadata using commands like
git logandgit ls-files. While Git metadata is an external and untrusted data source, the analysis is limited to statistical counting and high-level architectural classification, which poses no risk of instruction override or malicious command execution.
Audit Metadata