tidy
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPERSISTENCECOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project files (code, documentation, and comments) to identify redundant logic and dead code. This creates a vulnerability surface where instructions embedded in project comments or data could influence the agent's behavior during the cleanup process.
- Ingestion points: Project source code, documentation files, and
AGENTS.md. - Boundary markers: Absent.
- Capability inventory: File deletion, file modification, and command execution (git hook testing).
- Sanitization: The instructions include a mandatory human-in-the-loop step, requiring the agent to list changes and receive user confirmation before modification.
- [PERSISTENCE]: The skill installs or modifies git pre-commit hooks (potentially using tools like Husky or Lefthook) to enforce token limits. While this creates an automated script that runs on future git operations, it is a standard development workflow practice for maintaining project health and is explicitly disclosed.
- [COMMAND_EXECUTION]: The skill performs file system operations including deleting files and modifying configuration. It also involves executing git commands to test the newly installed hooks. These operations are within the scope of a 'tidy' utility and include instructions to verify changes with the user first.
Audit Metadata