Exa MCP Server

Instructions

Implementation Reference

• src/tools/index.ts:206-215 (handler)

  The handler function for the 'web_search_exa' tool. It validates input parameters using webSearchSchema, performs the Exa search via client.search, formats the results using formatSearchResults, and returns the content in MCP format.

  case 'web_search_exa': {
      const params = webSearchSchema.parse(args);
      const results = await client.search(params);
      return {
          content: [{
              type: "text",
              text: formatSearchResults(results)
          }]
      };
  }

• src/tools/index.ts:18-33 (schema)

  Zod schema used for input validation of the web_search_exa tool parameters.

  const webSearchSchema = z.object({
      query: z.string().describe("Search query for finding information on the web"),
      num_results: z.number().optional().default(10).describe("Number of results to return (default: 10, max: 100)"),
      include_domains: z.array(z.string()).optional().describe("Only include results from these domains"),
      exclude_domains: z.array(z.string()).optional().describe("Exclude results from these domains"),
      start_crawl_date: z.string().optional().describe("Start date for crawled content (YYYY-MM-DD)"),
      end_crawl_date: z.string().optional().describe("End date for crawled content (YYYY-MM-DD)"),
      start_published_date: z.string().optional().describe("Start date for published content (YYYY-MM-DD)"),
      end_published_date: z.string().optional().describe("End date for published content (YYYY-MM-DD)"),
      use_autoprompt: z.boolean().optional().default(true).describe("Let Exa AI optimize the search query"),
      type: z.enum(['keyword', 'neural', 'magic']).optional().default('neural').describe("Search type: 'keyword' for exact match, 'neural' for semantic search, 'magic' for best results"),
      category: z.string().optional().describe("Filter by content category (news, blog, research, etc.)"),
      include_text: z.boolean().optional().default(false).describe("Include extracted text content from pages"),
      include_highlights: z.boolean().optional().default(false).describe("Include highlighted snippets from pages"),
      include_summary: z.boolean().optional().default(false).describe("Include AI-generated summaries of pages")
  });

• src/tools/index.ts:82-154 (registration)

  Tool registration definition including name, description, and JSON input schema for the web_search_exa tool, returned by getToolDefinitions for MCP listTools handler.

  name: 'web_search_exa',
  description: 'Search the web using Exa AI for up-to-date information',
  inputSchema: {
      type: "object",
      properties: {
          query: {
              type: "string",
              description: "Search query for finding information on the web"
          },
          num_results: {
              type: "number",
              description: "Number of results to return (default: 10, max: 100)",
              default: 10
          },
          include_domains: {
              type: "array",
              items: { type: "string" },
              description: "Only include results from these domains"
          },
          exclude_domains: {
              type: "array", 
              items: { type: "string" },
              description: "Exclude results from these domains"
          },
          start_crawl_date: {
              type: "string",
              description: "Start date for crawled content (YYYY-MM-DD)"
          },
          end_crawl_date: {
              type: "string",
              description: "End date for crawled content (YYYY-MM-DD)"
          },
          start_published_date: {
              type: "string",
              description: "Start date for published content (YYYY-MM-DD)"
          },
          end_published_date: {
              type: "string",
              description: "End date for published content (YYYY-MM-DD)"
          },
          use_autoprompt: {
              type: "boolean",
              description: "Let Exa AI optimize the search query",
              default: true
          },
          type: {
              type: "string",
              enum: ["keyword", "neural", "magic"],
              description: "Search type: 'keyword' for exact match, 'neural' for semantic search, 'magic' for best results",
              default: "neural"
          },
          category: {
              type: "string",
              description: "Filter by content category (news, blog, research, etc.)"
          },
          include_text: {
              type: "boolean",
              description: "Include extracted text content from pages",
              default: false
          },
          include_highlights: {
              type: "boolean",
              description: "Include highlighted snippets from pages",
              default: false
          },
          include_summary: {
              type: "boolean",
              description: "Include AI-generated summaries of pages",
              default: false
          }
      },
      required: ["query"]
  }

• src/server.ts:30-37 (registration)

  Default enabled tools list in ExaServer constructor, including 'web_search_exa' for server tool registration.

  this.enabledTools = enabledTools || [
      'web_search_exa',
      'company_research_exa',
      'crawling_exa',
      'linkedin_search_exa',
      'deep_researcher_start',
      'deep_researcher_check'
  ];