using-git-worktrees

Pass

Audited by Gen Agent Trust Hub on Apr 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes various system commands for Git operations (worktree management, configuration checks), directory navigation, and file system interactions. It also triggers project-specific test runners (npm test, cargo test, pytest, go test) to verify workspace integrity.
  • [EXTERNAL_DOWNLOADS]: Automatically invokes standard package managers (npm, pip, poetry, go mod) to fetch and install project dependencies from official remote registries when setting up a new worktree.
  • [PROMPT_INJECTION]: The skill identifies configuration preferences by reading content from CLAUDE.md, which creates a surface for indirect prompt injection if the file is modified by untrusted sources.
  • Ingestion points: Reads CLAUDE.md using grep to extract worktree directory settings (SKILL.md).
  • Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are used when processing the external configuration data.
  • Capability inventory: High-privilege shell operations including git worktree creation, dependency installation, and arbitrary test execution.
  • Sanitization: No validation or sanitization is performed on the data extracted from CLAUDE.md before it influences the agent's workspace path selection.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 2, 2026, 04:26 PM