exa-research

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill uses official SDK patterns for Exa.ai (exa_py and exa-js) and follows security best practices by referencing environment variables (EXA_API_KEY) for authentication rather than hardcoding secrets.
  • [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and process untrusted data from the web via the Exa API, which is a standard surface for indirect prompt injection.
  • Ingestion points: Untrusted content enters the system via exa.answer(), exa.search_and_contents(), and exa.research() as seen in SKILL.md and the reference documentation.
  • Boundary markers: The provided implementation patterns do not explicitly define boundary markers or 'ignore' instructions for the ingested text.
  • Capability inventory: The skill facilitates network operations (via the Exa library) and data processing/formatting for research reports.
  • Sanitization: references/citations.md includes utility functions for URL validation and simple keyword-based relevance checks, though it lacks robust prompt-level sanitization for ingested web content.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 06:47 PM
Security Audit — agent-trust-hub — exa-research