build-with-exa
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides instructions to install official SDKs (exa-py for Python and exa-js for Node.js). These are standard dependencies for the Exa platform and originate from the vendor's official package registries.- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted external content from the web through search results and page extraction. This is a common surface for indirect prompt injection.
- Ingestion points: Data enters the agent context via API responses containing search highlights, full page text, or synthesized answers.
- Boundary markers: The skill encourages the use of systemPrompt and outputSchema to separate instructions from data and define strict formats for extracted information.
- Capability inventory: The skill enables the agent to perform web searches, retrieve page content, generate grounded answers, and run long-running research tasks via the Agent API.
- Sanitization: Guidance is provided on using structured JSON output and schema validation to ensure extracted data conforms to expected formats.- [DATA_EXPOSURE]: The skill provides standard documentation for API authentication, using environment variables and placeholders for the EXA_API_KEY. It does not contain hardcoded secrets or instructions that would lead to credential leakage.
Audit Metadata