autopilot
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill automates version control tasks by executing local scripts and standard CLI tools.
- Evidence: Uses
scripts/pull,scripts/checkout,git,sl(Sapling), andgh(GitHub CLI) to manage branches, commits, and pull requests. - [INDIRECT_PROMPT_INJECTION]: The skill monitors workspace content and performs automated write/push actions, which could be exploited via malicious code injection.
- Ingestion points: Monitors the local file system and checkouts to detect 'settled' units (SKILL.md).
- Boundary markers: Requires the user to explicitly invoke
/autopilotto start the mode and relies on a 'settled' state (tests/lint passing). - Capability inventory: Possesses the ability to write to the local repository (commit), push to remotes, and submit pull requests.
- Sanitization: Primarily relies on the success of external test suites and linters rather than sanitizing the code content itself.
Audit Metadata