worktrees

Warn

Audited by Gen Agent Trust Hub on Apr 7, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Invokes the bd CLI tool to automate Git worktree creation, removal, and beads database integration.
  • [COMMAND_EXECUTION]: Executes shell commands to modify the project's .gitignore file and manage Git repository state.
  • [COMMAND_EXECUTION]: Uses standard development toolchains including npm, cargo, and go to install dependencies and build projects within worktrees.
  • [EXTERNAL_DOWNLOADS]: Fetches external packages and dependencies from public registries (NPM, Crates.io, Go Proxy) during workspace initialization.
  • [DATA_EXFILTRATION]: Suggests reading agent configuration from the user's home directory (~/.claude/my-project-instructions.md), involving access to files outside the specific project directory.
  • [PROMPT_INJECTION]: Processes untrusted project metadata files such as package.json, Cargo.toml, and go.mod which could contain malicious content designed to influence the agent's behavior during build or setup steps.
  • [PROMPT_INJECTION]: Category 8 analysis: Untrusted data enters via project files (package.json, Cargo.toml, go.mod in SKILL.md); no explicit boundary markers or sanitization are present; the skill possesses capabilities to execute subprocesses (npm, cargo, go in SKILL.md).
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 7, 2026, 03:50 PM