using-git-worktrees

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • COMMAND_EXECUTION (MEDIUM): The skill executes multiple shell commands to set up project environments and run tests based on repository content (e.g., npm test, cargo test, pytest, go test ./...). This allows for the execution of arbitrary code defined in the project's configuration files.
  • EXTERNAL_DOWNLOADS (MEDIUM): Automatically invokes package managers (npm, pip, poetry, cargo, go) to download and install external dependencies. These operations fetch third-party code from public registries based on potentially untrusted local manifests like package.json or requirements.txt.
  • INDIRECT_PROMPT_INJECTION (LOW): The skill reads from CLAUDE.md to determine directory preferences without sanitization or boundary markers. An attacker could potentially influence the workspace location by providing malicious path configurations in CLAUDE.md.
  • Ingestion points: File read of CLAUDE.md using grep.
  • Boundary markers: Absent; the skill trusts the extracted string for directory selection.
  • Capability inventory: File system modification via git worktree add, cd, and directory creation; code execution via npm install and test runners.
  • Sanitization: None detected for the values extracted from CLAUDE.md.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 04:57 PM