add /research functionality to MCP server

[lloyd]

the new EXA /research API is super interesting! having it exposed via MCP would allow one to experiment with it within an LLM environment like claude. This PR closes issue #37 by adding three new MCP functions:

• exa_research_create_task
• exa_research_get_task
• exa_research_list_tasks

Example (claude code):

> can you start a research task to find out what google's latest gemini models are and their string tags, release date, and costs?

● I'll create a research task to gather comprehensive information about Google's latest Gemini models, their identifiers, release dates,
  and pricing.

● exa:exa_research_create_task (MCP)(instructions: "Research Google's latest Gemini AI models and provide comprehensive information
                                    including:\n\n1. Model names and string tags/identifiers (e.g., gemini-1.5-pro, gemini-1.5-flash,
                                    etc.)\n2. Release dates for each model\n3. Pricing information (cost per token, per request, or other
                                     pricing structure)\n4. Key capabilities and differences between models\n5. Any recent updates or new
                                     releases in 2024-2025\n6. API availability and access methods\n\nFocus on the most current and
                                    official information from Google AI, Google Cloud documentation, and reputable tech news sources.
                                    Include both consumer and enterprise/API pricing where available.", inferSchema: true)
  ⎿ {
      "taskId": "01jxdrfcxvxjy7s55nyy2pvfn5",
      "status": "Task created successfully. Use exa_research_get_task to check status and retrieve results.",
      "instructions": "Research Google's latest Gemini AI models and provide comprehensive information including:\n\n1. Model names and
    string tags/identifiers (e.g., gemini-1.5-pro, gemini-1.5-flash, etc.)\n2. Release dates for each model\n3. Pricing information (cost
     per token, per request, or other pricing structure)\n4. Key capabilities and differences between models\n5. Any recent updates or
    new releases in 2024-2025\n6. API availability and access methods\n\nFocus on the most current and official information from Google
    AI, Google Cloud documentation, and reputable tech news sources. Include both consumer and enterprise/API pricing where available."
    }


● Research task created! The task ID is 01jxdrfcxvxjy7s55nyy2pvfn5. Let me check the status and results:

● exa:exa_research_get_task (MCP)(taskId: "01jxdrfcxvxjy7s55nyy2pvfn5")
  ⎿ {
      "id": "01jxdrfcxvxjy7s55nyy2pvfn5",
      "status": "running",
      "instructions": "Research Google's latest Gemini AI models and provide comprehensive information including:\n\n1. Model names and
    string tags/identifiers (e.g., gemini-1.5-pro, gemini-1.5-flash, etc.)\n2. Release dates for each model\n3. Pricing information (cost
     per token, per request, or other pricing structure)\n4. Key capabilities and differences between models\n5. Any recent updates or
    new releases in 2024-2025\n6. API availability and access methods\n\nFocus on the most current and official information from Google
    AI, Google Cloud documentation, and reputable tech news sources. Include both consumer and enterprise/API pricing where available.",
    … +50 lines (ctrl+r to expand)


● The research task is currently running. Let me check again in a moment to see if it's completed:

● exa:exa_research_get_task (MCP)(taskId: "01jxdrfcxvxjy7s55nyy2pvfn5")
  ⎿ {
      "id": "01jxdrfcxvxjy7s55nyy2pvfn5",
      "status": "running",
      "instructions": "Research Google's latest Gemini AI models and provide comprehensive information including:\n\n1. Model names and
    string tags/identifiers (e.g., gemini-1.5-pro, gemini-1.5-flash, etc.)\n2. Release dates for each model\n3. Pricing information (cost
     per token, per request, or other pricing structure)\n4. Key capabilities and differences between models\n5. Any recent updates or
    new releases in 2024-2025\n6. API availability and access methods\n\nFocus on the most current and official information from Google
    AI, Google Cloud documentation, and reputable tech news sources. Include both consumer and enterprise/API pricing where available.",
    … +50 lines (ctrl+r to expand)


● The research task is still running. While we wait for it to complete, let me also do a quick direct search for Google Gemini models to
  get some immediate results:

● exa:web_search_exa (MCP)(query: "Google Gemini AI models latest 2024 2025 pricing API string tags identifiers", numResults: 5)
  ⎿  {
       "requestId": "b0f0d698fc5f600b88271589e9115d3a",
       "autopromptString": "Google Gemini AI models latest 2024 2025 pricing API string tags identifiers",
       "autoDate": "2024-01-01T00:00:00.000Z",
     … +66 lines (ctrl+r to expand)

● Great! I found some current information. Let me check if the research task has completed:

● exa:exa_research_get_task (MCP)(taskId: "01jxdrfcxvxjy7s55nyy2pvfn5")
  ⎿ {
      "id": "01jxdrfcxvxjy7s55nyy2pvfn5",
      "status": "completed",
      "instructions": "Research Google's latest Gemini AI models and provide comprehensive information including:\n\n1. Model names and
    string tags/identifiers (e.g., gemini-1.5-pro, gemini-1.5-flash, etc.)\n2. Release dates for each model\n3. Pricing information (cost
     per token, per request, or other pricing structure)\n4. Key capabilities and differences between models\n5. Any recent updates or
    new releases in 2024-2025\n6. API availability and access methods\n\nFocus on the most current and official information from Google
    AI, Google Cloud documentation, and reputable tech news sources. Include both consumer and enterprise/API pricing where available.",
    … +260 lines (ctrl+r to expand)


● Perfect! The research task has completed successfully. Here's a comprehensive summary of Google's latest Gemini models:

  Google Gemini Models - Latest Information

  Current Production Models (2024-2025)

  Gemini 1.5 Flash 002 (gemini-1.5-flash-002)
  - Released: September 24, 2024
  - Pricing: $0.075 per 1M input tokens, $0.30 per 1M output tokens
  - Features: Fast, efficient model with improved factuality, multilingual support (102 languages), 2x faster output, 3x lower latency

  Gemini 1.5 Pro 002 (gemini-1.5-pro-002)
  - Released: September 24, 2024
  - Pricing: $1.25 per 1M input tokens (up to 128K), $5.00 per 1M output tokens; $2.50/$10 for longer contexts
  - Features: 2M token context window, enhanced math/reasoning (~20% improvement), vision capabilities, fine-tuning support

  Gemini 2.5 Pro (gemini-2.5-pro)
  - Released: March 2025
  - Pricing: $1.25 per 1M input tokens (up to 200K), $10 per 1M output tokens; $2.50/$15 for longer contexts
  - Features: Most expensive model yet, industry-leading performance, 1M token context, "Deep Think" reasoning mode, native audio output

  Gemini 2.5 Flash (gemini-2.5-flash)
  - Released: May 2025 (preview), June 2025 (GA)
  - Pricing: Optimized for efficiency and lower cost
  - Features: 20-30% fewer tokens needed, improved reasoning, multimodal capabilities, native audio output