The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes repository-local scripts located in the bin/ directory, such as bin/worktree-add, bin/worktree-up, and bin/worktree-env. This behavior is the primary intended purpose of the tool for supporting repo-aware workflows.\n- [COMMAND_EXECUTION]: User-provided inputs, including branch names and pull request numbers, are interpolated into shell commands for git fetch, git worktree add, and printf.\n- [PROMPT_INJECTION]: The skill processes untrusted input strings (branch names and PR numbers) which represents a potential surface for indirect prompt injection.\n
Ingestion points: User-supplied <branch-name> and pr:<number> inputs processed in SKILL.md.\n
Boundary markers: No explicit delimiters are used for the interpolated variables.\n
Capability inventory: Access to git CLI and the ability to execute any script within the repository's bin/ directory.\n
Sanitization: The skill performs slugification (removing special characters) when generating directory paths, although raw inputs are passed to git commands.

create-worktree