The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute local git commands for repository analysis and branch deletion.
The commands executed include git branch -d, git branch -D, and git worktree remove.
Security is maintained through explicit quoting of branch variables ("$branch") to prevent shell expansion attacks from branch names containing special characters.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection as it ingests untrusted data from the local environment (git branch names and commit messages).
Ingestion points: Data enters the agent context through git branch, git log, and git worktree list results.
Boundary markers: The skill architecture enforces two distinct confirmation gates using AskUserQuestion, preventing the agent from acting on instructions potentially embedded in git metadata without human oversight.
Capability inventory: The skill has the capability to delete local file system data via git branch -D and git worktree remove.
Sanitization: The skill explicitly instructs the agent to quote branch names to prevent command injection and provides a gated workflow that requires the user to review all proposed deletions before execution.

git-cleanup