architecture
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local architecture checker script (scripts/check_architecture.py) to verify project structure and dependencies. It also invokes repomix via npx to package the repository for mapping.
- [EXTERNAL_DOWNLOADS]: Fetches the repomix package via npx. This is a well-known tool used for packing repositories to provide context for AI agents.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted code from the project repository to perform structural analysis.
- Ingestion points: scripts/check_architecture.py reads source code and package manifests.
- Boundary markers: No explicit delimiters are used for the script output.
- Capability inventory: The checker script is read-only; no network or file-write operations are performed during analysis.
- Sanitization: The script uses custom regex-based comment stripping to ensure only actual code imports are analyzed.
Audit Metadata