manage-git-changes
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a Python test script (
scripts/test_commit_snapshots.py) and a shell script (assets/examples/verify.sh) that execute Git commands usingsubprocess.runand direct shell execution. - These tools are used to verify the correct functioning of Git mechanics.
- They operate within isolated temporary directories to avoid impacting the host system.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from Git repositories, such as file contents and commit messages, which could contain instructions.
- Ingestion points: Git status, diffs, and logs in the working directory (described in SKILL.md and references/staging-and-commits.md).
- Boundary markers: The workflow requires explicit user authorization for operations like commit, rewrite, push, and tag.
- Capability inventory: Git repository management, file writing via commits/stashing, and network operations via push (SKILL.md).
- Sanitization: The skill instructs the agent to manually inspect diffs and run regex-based checks for secrets before staging (references/staging-and-commits.md).
Audit Metadata