use-worktrees

Pass

Audited by Gen Agent Trust Hub on Aug 21, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to follow setup instructions and install dependencies from the repository being processed, creating a surface for indirect prompt injection from untrusted files.
  • Ingestion points: Untrusted setup instructions and dependency manifests in target repositories.
  • Boundary markers: Absent.
  • Capability inventory: Git command execution, filesystem access outside the repository root, and shell command execution for setup.
  • Sanitization: Absent.
  • [COMMAND_EXECUTION]: The skill uses git worktree commands to manage checkouts in the ~/worktrees/ directory, which involves filesystem operations outside the immediate repository checkout.
  • [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill recommends installing dependencies from the worktree based on repository instructions, which could lead to the execution of untrusted code if the repository is malicious.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 21, 2026, 03:25 PM
Security Audit — agent-trust-hub — use-worktrees