Clear Thought MCP Server

Systematic thinking, mental models, and debugging approaches — as MCP tools for any AI client.

What it is / why it exists

An MCP server that gives an AI assistant a toolbox of structured reasoning frameworks — mental models, design patterns, systematic debugging, decision analysis, the scientific method, and more — so it reasons deliberately instead of improvising. Each mode is an explicit, inspectable tool the model calls, with typed (Zod-validated) inputs and a formatted, chain-of-thought-friendly response.

Twelve tools, one server, zero required configuration.

Related MCP server: Clear Thought 1.5

Links

• Live MCP endpoint: https://clear-thought-mcp-server.oriz.in (Streamable HTTP, MCP 2.0)

• GitHub Pages: https://chirag127.github.io/Clear-Thought-MCP-server/

• Smithery: https://smithery.ai/server/@chirag127/clear-thought-mcp-server

• Repo: https://github.com/chirag127/Clear-Thought-MCP-server

⭐ If this is useful, please star the repo — it helps others find it.

Architecture

flowchart LR
    Client["MCP Client<br/>(Claude, Cursor, opencode, ...)"]
    subgraph Server["clear-thought-mcp-server"]
        direction TB
        T1["sequentialthinking"]
        T2["mentalmodel"]
        T3["designpattern"]
        T4["programmingparadigm"]
        T5["debuggingapproach"]
        T6["collaborativereasoning"]
        T7["decisionframework"]
        T8["metacognitivemonitoring"]
        T9["scientificmethod"]
        T10["structuredargumentation"]
        T11["visualreasoning"]
        T12["stochasticthinking"]
    end
    Client -->|"Streamable HTTP (MCP 2.0) /mcp<br/>or stdio"| Server
    Server -->|"formatted reasoning result"| Client

Tools

Features

• 12 reasoning tools, each with a typed Zod input schema and formatted output

• Two transports: stdio (local clients) and Streamable HTTP (MCP 2.0, remote)

• Zero required configuration — no keys, no state, no external calls

• Installable via npx, Smithery, or the hosted endpoint

Tech stack

• TypeScript + Node.js (ESM, >=18)

• @modelcontextprotocol/sdk ^1.29

• zod for input validation, chalk for formatting

• vitest for tests, tsc build

Repo structure

Clear-Thought-MCP-server/
├── src/
│   ├── index.ts              # server bootstrap, tool registration, stdio + HTTP transports
│   ├── models/interfaces.ts  # shared TypeScript interfaces
│   └── tools/                # one module per reasoning tool (mentalModelServer.ts, ...)
├── dist/                     # compiled JS (tsc output)
├── test/                     # vitest suites
├── Dockerfile                # container image
├── smithery.yaml             # Smithery deploy config
└── package.json

Quick start

Run with npx (stdio)

npx clear-thought-mcp-server

Install via Smithery

npx -y @smithery/cli install @chirag127/clear-thought-mcp-server --client claude

MCP client config

Hosted (Streamable HTTP):

{
  "mcpServers": {
    "clear-thought": {
      "url": "https://clear-thought-mcp-server.oriz.in/mcp"
    }
  }
}

Local (stdio via npx):

{
  "mcpServers": {
    "clear-thought": {
      "command": "npx",
      "args": ["-y", "clear-thought-mcp-server"]
    }
  }
}

Run the HTTP transport yourself

npm install
npm run build
HTTP_PORT=3779 npm start   # serves Streamable HTTP at http://localhost:3779/mcp

Register it

• MCP Registry: https://registry.modelcontextprotocol.io

• Smithery: @chirag127/clear-thought-mcp-server — https://smithery.ai/server/@chirag127/clear-thought-mcp-server

Configuration

No configuration is required. Optional environment variables:

Part of the oriz family

One of ~80 sites and tools in the oriz family. It reasons; the Stochastic Thinking MCP Server handles probabilistic decision-making, and knowledge-mcp serves the knowledge base.

Contributing

Issues and PRs welcome. Conventional commits are the changelog.

License

MIT — see LICENSE.

Author

Chirag Singhal · chirag@oriz.in · @chirag127

Status

Stable (v1.1.2). Roadmap: more mental models, richer visual-reasoning output.