searching-with-exa

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). This skill's INSTRUCTIONS.md explicitly instructs the agent to use Exa MCP tools such as web_search_advanced_exa (including categories like "personal site" and "tweet") and crawling_exa to fetch and ingest public web pages, social posts, and arbitrary URLs, which the agent is expected to read and use to drive summaries and follow-up actions (e.g., deep_researcher_start), exposing it to untrusted third-party content that could carry indirect prompt injections.