The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes Git and shell commands to automate the lifecycle of Git worktrees. These operations are restricted to the local filesystem and repository context.
[EXTERNAL_DOWNLOADS]: The skill identifies and invokes Node.js package managers (npm, yarn, pnpm, bun) to install project dependencies when initializing new worktree environments.
[SAFE]: Safety protocols are deeply integrated into the workflows, including validation of repository status and user prompts before copying environment files or removing directories.
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing Git branch names which may be attacker-controlled in malicious repositories.
Ingestion points: Branch names are ingested via git branch and git worktree list commands in various modes and templates.
Boundary markers: Data is typically processed as individual lines and displayed with status indicators, though explicit instructions to ignore embedded commands are not present.
Capability inventory: The skill possesses capabilities to execute shell commands, install packages, and remove directories (e.g., in modes/mode3-cleanup.md).
Sanitization: Shell scripts in the templates/ directory use standard quoting for variables to mitigate common command injection vectors.

git-worktree-setup