What is ThreadMind?

ThreadMind MCP is a Model Context Protocol server that organizes AI conversations into hierarchical thread trees.

The Problem

When you work with an AI coding assistant, every new message includes the entire conversation history as context. This means:

• Token waste — repeating thousands of tokens of already-processed information
• Context limits — hitting the model's context window, losing early messages
• Tangled topics — authentication logic mixed with UI discussions mixed with deployment questions

Message 1 → Message 2 → ... → Message 80 → Message 81
                                              ↑
                        All 80 messages sent as context
                        = tens of thousands of tokens

The Solution

ThreadMind replaces linear history with a tree of summarized threads:

main (summary: 200 tokens)
├── auth (summary: 150 tokens)
│   ├── auth-ui (summary: 100 tokens) ← active
│   └── auth-api (summary: 120 tokens)
└── dashboard (summary: 180 tokens)

When you're working on auth-ui, only the direct ancestor chain is sent as context:

main → auth → auth-ui = ~450 tokens

That's it. The dashboard branch and auth-api sibling are not included — they're irrelevant to your current focus.

Key Benefits

Drastically Reduce Token Usage

Instead of sending 50+ messages of raw conversation, you send 2-5 concise summaries. Typical reduction: 80-95% of tokens.

Explore Multiple Approaches

Want to try a different architecture? Create a branch:

main
├── approach-a (REST API)
│   └── implementation
└── approach-b (GraphQL)     ← switch here to explore
    └── implementation

Each branch maintains its own context. Switch freely without contaminating other explorations.

Collaborate with Your Team

In team mode, thread trees are shared via git:

• Pull teammates' threads to see their progress
• Branch from their work to build on it
• Each person manages their own summaries

Stay Organized

Every topic gets its own thread. No more scrolling through hundreds of messages to find that one decision about the database schema.

How It Integrates

ThreadMind is an MCP server that works alongside your AI client:

┌─────────────┐       MCP Protocol       ┌──────────────┐
│  AI Client   │◄────────────────────────►│  ThreadMind  │
│ (Claude Code,│    tools & resources     │  MCP Server  │
│  ChatGPT,    │                          │              │
│  Gemini...)  │                          │  .threadmind/│
└─────────────┘                           └──────────────┘

The AI client calls ThreadMind's tools (create threads, update summaries, get context) and reads its resources (assembled context, tree visualization). ThreadMind handles all the storage and context assembly logic.

Next Steps

• Getting Started — install and configure ThreadMind
• Projects — understand project management
• Threads — learn about thread trees
• Context Assembly — how context is built