exa-search

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

• Third-party content exposure detected (high risk: 0.90). The required workflow uses the Exa MCP server tools (web_search_exa / get_code_context_exa) to fetch and ingest public web/code content at runtime (e.g., arbitrary pages, GitHub/StackOverflow/docs snippets), which are outsider-authored free text that can be passed into the agent’s LLM context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill's MCP configuration runs "npx -y exa-mcp-server" at runtime, which fetches and executes remote npm package code (e.g. https://www.npmjs.com/package/exa-mcp-server or https://registry.npmjs.org/exa-mcp-server), so external content is executed and can directly control the agent.