tavily-web-search

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill facilitates interactions with the Tavily Search API (https://api.tavily.com/search), a well-known and legitimate service for AI-optimized web searches. It provides implementation examples using the standard Python requests library and curl for API communication.
  • [SAFE]: Authentication is handled securely via the $TAVILY_API_KEY environment variable. This approach avoids the risk of hardcoded credentials and follows standard security protocols for secret management.
  • [PROMPT_INJECTION]: The skill includes an inherent attack surface for indirect prompt injection as its core function is to ingest and synthesize external web content.
  • Ingestion points: Untrusted data enters the agent context via the results returned by the Tavily API call in SKILL.md.
  • Boundary markers: There are no explicit delimiters or specific instructions provided in the snippets to prevent the agent from following commands embedded in the search results.
  • Capability inventory: The skill utilizes network capabilities to communicate with the Tavily endpoint.
  • Sanitization: No sanitization or filtering logic for the retrieved web content is implemented in the provided examples.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 05:44 PM