The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various system and development commands to manage git worktrees (git worktree add, git rev-parse), modify project configuration (.gitignore), and initialize environments (npm install, cargo build, go mod download). It also runs test suites (npm test, cargo test, pytest, go test) based on the detected project type.
[EXTERNAL_DOWNLOADS]: The automated execution of package managers (npm, pip, poetry, go) triggers the download of external dependencies defined in the project's configuration files (e.g., package.json, requirements.txt).
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through project-controlled files.
Ingestion points: The skill reads and acts upon configuration found in CLAUDE.md and repository-level files like package.json or Cargo.toml (SKILL.md).
Boundary markers: No boundary markers or specific validations are used when reading preferences from CLAUDE.md or when executing scripts defined in the repository.
Capability inventory: The skill has extensive capabilities including file system modification, git commits, network access via package managers, and arbitrary command execution through build and test scripts.
Sanitization: There is no sanitization or human-in-the-loop requirement before executing installation or test scripts found in the repository, assuming the project files are safe to process.

using-git-worktrees