tavily-best-practices
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [SAFE]: No malicious patterns or behaviors detected. The skill provides integration guidelines for the Tavily API and emphasizes secure implementation.
- [INDIRECT_PROMPT_INJECTION]: The skill involves processing untrusted web data. Evidence: 1. Ingestion points:
search,extract,crawl, andresearchmethods inSKILL.mdand reference files. 2. Boundary markers: Documentation explicitly warns to treat returned content as untrusted. 3. Capability inventory: Not applicable (documentation resource). 4. Sanitization: Author recommends verifying content against user requests. - [EXTERNAL_DOWNLOADS]: The documentation references official packages such as 'tavily-python', '@tavily/core', 'langchain-tavily', 'pydantic-ai-slim', 'agno', 'crewai', 'anthropic', 'openai', and 'llama-index' from standard registries.
Audit Metadata