The Agent Skills Directory

[SAFE]: The skill utilizes standard, non-destructive git commands such as git rev-parse --is-inside-work-tree and git worktree list to detect the user's environment and provide relevant guidance.
[SAFE]: The skill includes a dedicated 'Guardrails' section that explicitly instructs the agent never to execute destructive commands (like removing worktrees or deleting branches) without user confirmation, and to verify the current location before suggesting operations.
[INDIRECT_PROMPT_INJECTION]: The skill interacts with external data by reading local git branch names and file paths. While this presents a theoretical surface for indirect injection if a repository contained maliciously named branches, the skill's functionality is restricted to local development guidance and includes safety instructions for the agent to verify state, resulting in a safe implementation for its intended purpose.

worktree-guide