source-driven-development
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest external content from official documentation websites (e.g., react.dev, docs.djangoproject.com) and local project files (e.g., package.json, requirements.txt). This creates a surface for indirect prompt injection if those sources were to contain malicious instructions targeting the agent's behavior. However, the instructions focus on authoritative, high-reputation sources and standard project metadata, minimizing the risk.
- Ingestion points: Dependency files (package.json, composer.json, requirements.txt, etc.) and external documentation URLs.
- Boundary markers: None explicitly defined in the instructions for the agent to separate ingested content from its system instructions.
- Capability inventory: The skill generates code recommendations and cites sources but does not include instructions for file writing, network exfiltration, or shell command execution beyond fetching documentation.
- Sanitization: None mentioned for the ingested text.
Audit Metadata