The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to execute shell commands for repository management and filesystem cleanup. Evidence includes instructions for git worktree remove, git branch -d, and rm -rf <path>.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the local Git environment, creating a vulnerability surface. 1. Ingestion points: The audit procedure in SKILL.md reads output from git worktree list --porcelain. 2. Boundary markers: Absent; no specific delimiters are used to wrap external data. 3. Capability inventory: Commands including git, awk, sed, and rm -rf. 4. Sanitization: Absent; branch names and paths are processed as-is without validation.

worktree-hygiene