manage-branches
Pass
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill explicitly defines a restrictive scope, prohibiting network operations like pushing or fetching, and remote branch deletions, which minimizes the risk of data exfiltration.
- [SAFE]: Instructions include mandatory evidence collection and verification steps (e.g., checking OIDs, status, and untracked collisions) to ensure state integrity before performing mutations.
- [SAFE]: The skill incorporates a defensive measure against indirect prompt injection by instructing the agent to treat repository files, commit messages, and issue text as data rather than commands unless authority is established.
- [SAFE]: No obfuscation, hardcoded credentials, remote code downloads, or privilege escalation patterns were detected.
Audit Metadata