Thinking Partner MCP

Name: Thinking Partner MCP
Author: nwant

@nwant

Repository

About Thinking Partner MCP

MCP server for unified AI thinking partner across Claude Desktop and Code

Basic information

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "thinking-partner": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-thinking-partner"
      ]
    }
  }
}

Tools

No tools detected

We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.

Overview

What is Thinking Partner MCP?

Thinking Partner MCP is a Model Context Protocol (MCP) server that creates a unified AI thinking partner across Claude Desktop and Claude Code. It enables seamless context sharing and conversation continuity between architectural thinking and implementation work.

How to use Thinking Partner MCP?

Install dependencies with npm install, then configure both Claude Desktop (npm run setup-desktop) and Claude Code (npm run setup-code). Start the development server with npm run dev. For multi-machine sync, set environment variables like THINKING_PARTNER_GIT_REMOTE in a .env file.

Key features of Thinking Partner MCP

set_focus: Set a shared work topic across both tools.
get_context: Retrieve relevant context for the current conversation.
log_design_decision: Capture architectural choices with reasoning.
log_implementation_finding: Record discoveries made during coding.
bridge_conversation: Connect conversations between Desktop and Code.
Automatic Git synchronization for context across multiple machines.

Use cases of Thinking Partner MCP

Morning design session in Claude Desktop to explore architecture and log decisions.
Implementation session in Claude Code that automatically pulls in design context.
Next-day continuity where both tools remember reasoning and open questions.
Cross-session pattern recognition through persistent context storage.

FAQ from Thinking Partner MCP

What tools does Thinking Partner MCP provide?

It offers five tools: set_focus, get_context, log_design_decision, log_implementation_finding, and bridge_conversation.

How does context persist between sessions?

Context is stored persistently in a local JSON file (data/context.json), which is automatically read and written by the server.

Does Thinking Partner MCP support syncing across multiple machines?

Yes, by setting environment variables in a .env file (e.g., THINKING_PARTNER_GIT_REMOTE), the server can automatically pull before reading and commit/push after saving changes.