The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various system and development commands to manage the workspace. This includes git commands for worktree management and manifest inspection, as well as package managers (npm, cargo, pip, poetry, go) and test runners (npm test, pytest, cargo test, go test). These executions are context-dependent and triggered by the presence of standard project files.
[INDIRECT_PROMPT_INJECTION]: The skill reads and processes data from local project files which could be influenced by external actors in a collaborative environment. It specifically reads CLAUDE.md to determine directory preferences and inspects standard manifest files (package.json, requirements.txt, etc.) to decide which setup commands to run.
Ingestion points: Reads .worktrees/, worktrees/, CLAUDE.md, .gitignore, package.json, Cargo.toml, requirements.txt, pyproject.toml, and go.mod.
Boundary markers: None present; the skill treats the existence and content of these files as trusted triggers for command execution.
Capability inventory: The skill has the capability to execute shell commands, modify the .gitignore file, and commit changes to the repository.
Sanitization: There is no explicit sanitization of the paths or configurations read from the project files before they are used in shell commands.

using-git-worktrees