cs-feat
Pass
Audited by Gen Agent Trust Hub on Jul 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a controlled development workflow involving a 'Harness' that manages risk and authorized writes, and a 'Reviewer' that challenges the implementation. This multi-role structure is a security best practice for autonomous agents.
- [SAFE]: No external downloads or remote code execution from unknown sources were identified. The skill relies on 'project-standard verification commands' and local repository tools.
- [DATA_EXFILTRATION]: Access to the repository and internal state files (state.json, evidence.jsonl) is restricted to the local environment for feature implementation and verification. No sensitive paths or exfiltration patterns were found.
- [PROMPT_INJECTION]: The skill addresses potential indirect prompt injection from the repository through mandatory boundary markers ('allowed writes', 'write/side-effect boundary'), capability inventory, and rigorous verification requirements enforced by the Harness.
Audit Metadata