API Documentation Lookup
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWPROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection (LOW): The skill processes untrusted external data from a non-whitelisted domain (effect-ts.github.io), which could theoretically contain malicious instructions designed to influence the agent. Evidence: 1. Ingestion: Data is fetched via WebFetch from external URLs. 2. Boundaries: No explicit boundary markers or isolation instructions (e.g., 'ignore embedded commands') are defined in the retrieval prompts. 3. Capabilities: The fetched content influences agent reasoning and responses; however, no dangerous write, execute, or network-send capabilities are associated with this data pipeline. 4. Sanitization: No sanitization or filtering of the retrieved documentation content is observed.
Audit Metadata