codebase-onboarding
Pass
Audited by Gen Agent Trust Hub on Apr 3, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill reads various codebase manifest files, configuration files, and source code to generate an onboarding guide and a CLAUDE.md instruction file. This process creates a surface where malicious instructions embedded in a repository's metadata (e.g., package.json description) could be incorporated into the agent's persistent project instructions, potentially subverting its behavior.\n
- Ingestion points: manifest files (package.json, pyproject.toml, etc.), configuration files (next.config.*, etc.), and source code entry points identified in Phase 1.\n
- Boundary markers: None specified for the untrusted content being processed during reconnaissance.\n
- Capability inventory: Writing a CLAUDE.md file to the repository root, which defines the 'Project Instructions' that govern the agent's future behavior within that environment.\n
- Sanitization: No explicit sanitization or validation of the extracted content is mentioned before it is included in the generated artifacts.
Audit Metadata