project-setup
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill installs shell and JavaScript scripts into the project's .claude/ directory and applies executable permissions using chmod +x. This is the intended functionality to provide project automation tools.
- [PERSISTENCE]: Project-specific hooks are registered in .claude/settings.json to trigger automation during various agent events (e.g., PreToolUse, Stop, SessionStart), ensuring the environment configuration remains active across sessions.
- [INDIRECT_PROMPT_INJECTION]: The skill processes project manifest files to auto-detect framework settings and populate templates.
- Ingestion points: Manifest files such as package.json, pyproject.toml, Cargo.toml, etc., as described in Phase 1 and references/detection-rules.md.
- Boundary markers: Phase 2 requires explicit user confirmation of all detected values before they are written to the configuration.
- Capability inventory: The skill uses Write, Edit, and Bash tools to modify project configuration files and install scripts.
- Sanitization: While content is interpolated into templates, the human-in-the-loop confirmation step serves as the primary defense against malicious manifest data.
Audit Metadata