worktrees
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes common development tools such as git, npm, pip, poetry, cargo, and go to manage worktrees and dependencies. These actions are performed to facilitate isolated feature development and are appropriate for the skill's stated intent.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data by searching
CLAUDE.mdfor worktree preferences. The ingestion point is theCLAUDE.mdfile, which lacks explicit boundary markers to delimit instructions. The capability inventory associated with this surface includes shell command execution and file writes. However, the skill employs sanitization by using a specific grep pattern (worktree.*director) to limit the data extracted to configuration values only. - [SAFE]: The skill enforces repository hygiene by checking if worktree directories are ignored by git. If a directory is not ignored, the skill automatically updates the
.gitignorefile and commits the change, preventing accidental data exposure or repository pollution.
Audit Metadata