using-git-worktrees
Pass
Audited by Gen Agent Trust Hub on Feb 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill automatically executes shell commands for environment setup and verification, including
npm install,pip install,poetry install, andgo mod download. It further executes test suites such asnpm test,cargo test,pytest, andgo test ./...immediately after worktree creation. - [EXTERNAL_DOWNLOADS]: Orchestrates downloads from well-known public registries (NPM, PyPI, Crates.io, and Go modules) to install project dependencies.
- [PROMPT_INJECTION]: Susceptible to indirect prompt injection through the
CLAUDE.mdfile. - Ingestion points: The skill reads
CLAUDE.mdusinggrepto find worktree directory preferences. - Boundary markers: None; the skill searches for patterns and adopts the found path without validation.
- Capability inventory: Includes
git worktree add, directory creation, repository-level commits (.gitignoremodification), and arbitrary command execution via package managers and test runners. - Sanitization: No sanitization is performed on the directory paths or branch names retrieved from the environment or configuration files before they are used in shell commands.
Audit Metadata