research-task-mcp-server

Research Task MCP Server v2

A powerful MCP (Model Context Protocol) server that enables AI-powered research on ANY topic through an interactive configuration wizard. Originally designed for LLM evaluation tools, now completely flexible and customizable!

🚀 What's New in v2

• Interactive Configuration Wizard: Start with a simple description, get a customized research plan
• Works for ANY Topic: Market research, academic literature reviews, competitive analysis, technology assessment, and more
• Domain Templates: Pre-built templates that adapt to your specific needs
• Quality Review Agent: Built-in quality assurance that reviews research completeness
• Flexible Output Formats: Choose from comparison, deep dive, recommendations, synthesis, or executive summary
• Backward Compatible: All original tools still work for existing workflows

Features

• Real AI-powered research using Anthropic's Claude API
• Interactive configuration through conversational wizard
• Parallel execution of multiple research agents
• Comprehensive analysis across customizable dimensions
• Quality assurance with automated review and recommendations
• Automatic synthesis and actionable insights
• Rate limiting to respect API limits

Installation

npm install
npm run build

Configuration

1. Copy .env.example to .env:

   cp .env.example .env
   

2. Add your Anthropic API key:

   ANTHROPIC_API_KEY=your_api_key_here
   

Quick Start Guide

Step 1: Start Research Configuration

// Describe what you want to research
await configureResearch({
  initialDescription: "I want to research sustainable packaging solutions for e-commerce"
})

// The wizard will respond with clarifying questions

Step 2: Answer Questions

// Continue the conversation
await continueConfiguration({
  sessionId: "your-session-id",
  userResponse: "I'm looking at this from an investment perspective, focusing on North American markets"
})

Step 3: Generate Research Plan

// Once configuration is complete
await generateResearchPlan({
  sessionId: "your-session-id",
  includeQualityReview: true
})

// Returns a detailed research plan with dimensions, criteria, and quality checks

Step 4: Execute Research

// Run the research with AI agents
await runFlexibleResearch({
  configId: "your-config-id",
  executionMode: "parallel",
  includeQualityReview: true
})

// Returns comprehensive findings, synthesis, and recommendations

Example Research Scenarios

Market Research

"Research market for AI code assistants"
→ Market size, competition, customer needs, trends
→ Investment recommendations with opportunity scores

Academic Research

"Literature review on quantum computing applications"
→ Theoretical frameworks, methodologies, key findings
→ Research gaps and future directions

Competitive Analysis

"Analyze competitors in meal kit delivery"
→ Feature comparison, pricing, market positioning
→ Strategic recommendations and differentiation opportunities

Technology Assessment

"Evaluate Kubernetes vs Docker Swarm"
→ Technical capabilities, implementation factors
→ Adoption recommendations for your use case

Available Tools

New Flexible Research Tools

1. configureResearch - Start interactive research configuration
2. continueConfiguration - Continue configuration conversation
3. generateResearchPlan - Generate customized research plan
4. runFlexibleResearch - Execute research with quality review

Legacy Tools (for LLM evaluation research)

1. createResearchTask - Create a research task
2. defineEvaluationAreas - Define areas to evaluate
3. initializeSubagents - Set up research agents
4. runSubagentResearch - Execute AI research
5. synthesizeFindings - Synthesize and compare findings
6. getOpportunityRecommendation - Get recommendations

Claude Desktop Configuration

1. Find your Claude Desktop config file:

   • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
   • Windows: %APPDATA%\Claude\claude_desktop_config.json
   • Linux: ~/.config/Claude/claude_desktop_config.json

2. Add the research-task server to your config:

{
  "mcpServers": {
    "research-task": {
      "command": "node",
      "args": ["/Users/YOUR_USERNAME/research-task-mcp-server/dist/index-v2.js"],
      "env": {
        "ANTHROPIC_API_KEY": "sk-ant-api03-YOUR-API-KEY-HERE"
      }
    }
  }
}

3. Replace:

   • /Users/YOUR_USERNAME/ with your actual path
   • YOUR-API-KEY-HERE with your Anthropic API key

4. Restart Claude Desktop for changes to take effect

Research Templates

The system includes pre-built templates for:

• Market Research: Market size, competition, customers, trends
• Academic Research: Literature review, methodologies, findings, gaps
• Competitive Analysis: Competitor profiles, positioning, strategies
• Technology Assessment: Capabilities, readiness, implementation factors

Templates automatically adapt based on your specific requirements!

Quality Assurance

Every research project includes:

• Completeness checks across all dimensions
• Bias and consistency validation
• Confidence scoring
• Specific improvement recommendations
• Optional quality review by dedicated AI agent

Development

npm run dev    # Run with auto-reload
npm run build  # Build for production
npm start      # Run production build

Architecture

┌─────────────────────┐
│ Configuration       │ ← Interactive wizard understands your needs
│ Wizard              │
└──────────┬──────────┘
           │
┌──────────▼──────────┐
│ Research Templates  │ ← Adapts to your specific domain
│ & Plan Generation   │
└──────────┬──────────┘
           │
┌──────────▼──────────┐
│ Flexible Research   │ ← Multiple AI agents work in parallel
│ Agents              │
└──────────┬──────────┘
           │
┌──────────▼──────────┐
│ Quality Review      │ ← Ensures comprehensive, unbiased results
│ Agent               │
└──────────┬──────────┘
           │
┌──────────▼──────────┐
│ Synthesis &         │ ← Actionable insights and recommendations
│ Recommendations     │
└─────────────────────┘

Rate Limiting

The server automatically handles rate limiting:

• 10 requests per minute (default)
• Automatic queuing and retry
• Graceful handling of API limits

Extending the System

To add new research domains:

1. Add domain to ResearchDomain type in types-v2.ts
2. Create template in research-templates.ts
3. Define default dimensions and quality checks
4. Add domain-specific keywords for auto-detection

Requirements

• Node.js 18+
• Anthropic API key
• Claude Desktop (for MCP integration)