tavily
Pass
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted web content from the Tavily API, creating a vulnerability surface for indirect prompt injection.
- Ingestion points: The
scripts/search.mjsscript retrieves web snippets, andscripts/extract.mjsfetches raw content from arbitrary URLs, which are then presented to the agent. - Boundary markers: The output provided to the agent lacks delimiters or specific instructions to distinguish retrieved data from trusted system prompts.
- Capability inventory: The scripts utilize the Node.js
fetchAPI to communicate with external endpoints and retrieve data. - Sanitization: Content retrieved from the API is printed to the agent's context without sanitization, filtering, or escaping of potentially malicious instructions embedded in the web data.
- [EXTERNAL_DOWNLOADS]: The skill performs network requests to the Tavily API (
api.tavily.com) to retrieve search results and extract content from remote websites.
Audit Metadata