jaimeparker/exa-search

Exa Search

Neural search for web content, code, companies, and people via the Exa MCP server.

When to Activate

• User needs current web information or news.
• Searching for code examples, API docs, or technical references.
• Academic Research: Finding specific research papers or academic discussions.
• Deep Literature Review: When a query requires a multi-step, synthesized report (Deep Researcher).
• User says "search for", "look up", "find", or "what's the latest on".

🛡️ Interaction Protocol - Important

Before performing any search operation, the Agent MUST follow this workflow:

1. Assess Task Complexity:
   • For simple queries: Ask about Category and Domain filters.
   • For complex topics (e.g., reviews, comparative analysis): Propose the "Deep Researcher" mode (explaining that it generates an exhaustive, synthesized report asynchronously).
2. Actively Inquire Filtering Needs: Unless the user has already specified conditions in their prompt, the Agent must ask:
   • Category Filter: Should the search be limited to research paper, news, github, or tweet?
   • Domain Filter: Should specific sites be included or excluded (e.g., include arxiv.org)?
3. Provide Suggestions: Suggest the most appropriate category and domain filters based on the query context.
4. Confirm and Execute: Adjust tool selection and parameters based on the user's response before proceeding.

MCP Requirement

Exa MCP server must be configured in settings.json. To enable all capabilities, ensure the HTTP URL or transport configuration includes the necessary tools if using a custom proxy, or simply use the standard remote URL: https://mcp.exa.ai/mcp?exaApiKey=YOUR_KEY&tools=web_search_exa,web_search_advanced_exa,get_code_context_exa,crawling_exa,deep_researcher_start,deep_researcher_check

For simple search needs, use: https://mcp.exa.ai/mcp?tools=get_code_context_exa,people_search_exa

Core Tools & Parameters

1. Basic & Advanced Search

• web_search_exa: Simple neural search.
• web_search_advanced_exa: Supports advanced filtering like includeDomains, excludeDomains, and startPublishedDate.
• get_code_context_exa: Optimized for finding technical code and documentation.

2. Content Extraction

• crawling_exa(url: string): Retrieves the full text content of a specific URL, rather than just search snippets.

3. Deep Researcher Mode

• deep_researcher_start(query: string): Initiates an autonomous research agent. Returns a taskId.
• deep_researcher_check(taskId: string): Polls for status and retrieves the final synthesized report with citations.

Usage Patterns

1. Pattern: Simple Search

User: "Find me the latest progress on xxx." Agent: "I can search that for you. To make the results more precise, would you like to limit the search to research papers? I also suggest including arxiv.org and paperswithcode.com to find implementation details. Does that sound good?"

2. Pattern: Deep Research Task

User: "Compare the pros and cons of major xxx algorithms in 2026." Agent: "This is a complex research topic. I suggest using the 'Deep Researcher' mode. It will take about 1-2 minutes to autonomously synthesize a detailed report comparing multiple sources. Should I start this research task for you?"

Related Skills

• deep-research — For generating long-form, exhaustive research reports.
• paper-finder — Use Exa as a backend for semantic discovery beyond arXiv.