MCP Think Tank

v1.0.0Knowledge & Memorystable

MCP Think Tank is a powerful Model Context Protocol (MCP) server designed to enhance the capabilities of AI assistants like Cursor and Claude. It provides a structured environment for enhanced reasoning, persistent memory, and responsible tool usage.

mcp-think-tankmcpai-integration
Share:
62
Stars
0
Downloads
0
Weekly
0/5

What is MCP Think Tank?

MCP Think Tank is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to mcp think tank is a powerful model context protocol (mcp) server designed to enhance the capabilities of ai assistants like cursor and claude. it provides a structured environment for enhanced reasoni...

MCP Think Tank is a powerful Model Context Protocol (MCP) server designed to enhance the capabilities of AI assistants like Cursor and Claude. It provides a structured environment for enhanced reasoning, persistent memory, and responsible tool usage.

This server falls under the Knowledge & Memory category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • upsert_entities
  • create_relations
  • add_observations
  • delete_entities
  • delete_observations

Use Cases

Enhanced reasoning and persistent memory for agents
Responsible tool usage with structured environment
flight505

Maintainer

LicenseMIT License
Languagetypescript
Versionv1.0.0
UpdatedMay 15, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

NPM

npx -y mcp-think-tank

Manual Installation

npx -y mcp-think-tank

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use MCP Think Tank

MCP Think Tank is a comprehensive MCP server that enhances AI assistants with structured reasoning, persistent knowledge graph memory, task planning, and web research capabilities. It provides a 'think' tool for deep structured reasoning with optional self-reflection, a full knowledge graph for storing and retrieving entities and relationships across sessions, a task management system with dependencies and priorities, and Exa-powered web search for grounding responses in current information. Developers and power users integrate it with Cursor and Claude Desktop to give their AI workflows long-term memory and more disciplined problem-solving.

Prerequisites

  • Node.js installed (for npx execution)
  • An Exa API key from exa.ai (required for web search tools)
  • An MCP-compatible client such as Claude Desktop or Cursor
  • A writable directory for the memory file (MEMORY_PATH)
  • npm 5.2+ for npx usage
1

Obtain an Exa API key

Sign up at exa.ai and create an API key. This is required to use the exa_search and exa_answer web research tools. The knowledge graph and reasoning tools work without it.

2

Decide on a memory file location

Choose an absolute path where the knowledge graph will be stored as a JSONL file. This file persists across sessions and is the source of long-term memory for the AI.

3

Add to your MCP client configuration

Add mcp-think-tank to your Claude Desktop or Cursor MCP configuration, setting MEMORY_PATH to your chosen file location and EXA_API_KEY to your key.

4

Restart your MCP client

Reload Claude Desktop or Cursor so it picks up the new server. On first run, npx will download the mcp-think-tank package automatically.

5

Use the think tool for complex reasoning

Instruct the AI to use the 'think' tool when you want structured multi-step analysis. Enable 'selfReflect: true' to have the model critique its own reasoning before finalizing an answer.

6

Build and query the knowledge graph

Use upsert_entities and add_observations to store facts, create_relations to link concepts, and search_nodes or memory_query to retrieve context. The memory persists in the JSONL file between sessions.

MCP Think Tank Examples

Client configuration

Claude Desktop configuration using npx to run mcp-think-tank with persistent memory and Exa web search enabled.

{
  "mcpServers": {
    "think-tank": {
      "command": "npx",
      "args": ["-y", "[email protected]"],
      "env": {
        "MEMORY_PATH": "/Users/you/projects/memory.jsonl",
        "EXA_API_KEY": "your-exa-api-key-here",
        "TOOL_LIMIT": "25",
        "CACHE_TOOL_CALLS": "true"
      }
    }
  }
}

Prompts to try

Example prompts that exercise MCP Think Tank's reasoning, memory, task planning, and web research capabilities.

- "Use the think tool to break down this architecture decision, then save your conclusions using upsert_entities"
- "Query my memory from the last 48 hours and show me the highest-priority task to resume"
- "Search the web for the latest best practices for React Server Components, analyze with think, and store key concepts in the knowledge graph"
- "Create a task plan for migrating our database to PostgreSQL, with dependencies between steps"

Troubleshooting MCP Think Tank

The exa_search tool returns an error about authentication

Verify that EXA_API_KEY is set correctly in the env section of your MCP config and that the key is valid at exa.ai. The thinking and memory tools work independently of Exa and will still function without a valid key.

Memory from previous sessions is not appearing in responses

Check that MEMORY_PATH points to an absolute path and that the file exists and is readable. Run the show_memory_path tool to confirm the server is using the correct location. If the path is relative, convert it to absolute.

The AI exceeds the tool call limit and stops mid-task

Increase TOOL_LIMIT in your environment config (default is 25). For complex multi-step tasks, explicitly instruct the AI to use plan_tasks first to sequence work, which reduces redundant tool calls.

Frequently Asked Questions about MCP Think Tank

What is MCP Think Tank?

MCP Think Tank is a Model Context Protocol (MCP) server that mcp think tank is a powerful model context protocol (mcp) server designed to enhance the capabilities of ai assistants like cursor and claude. it provides a structured environment for enhanced reasoning, persistent memory, and responsible tool usage. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install MCP Think Tank?

Install via npm with the command: npx -y mcp-think-tank. Then add the server configuration to your AI client's JSON config file (e.g., claude_desktop_config.json or .cursor/mcp.json).

Which AI clients work with MCP Think Tank?

MCP Think Tank works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is MCP Think Tank free to use?

Yes, MCP Think Tank is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.

Browse More Knowledge & Memory MCP Servers

Explore all knowledge & memory servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "mcp-think-tank": { "command": "npx", "args": ["-y", "mcp-think-tank"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use MCP Think Tank?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides