MCP Brain

Your Personal AI Memory — works with Claude, ChatGPT, Gemini, Cursor, and any MCP client.

MCP Brain is a persistent memory layer that lets your AI remember you across every platform. Store observations, decisions, ideas, tasks, and preferences — then retrieve them from any AI tool via the MCP protocol.

How It Works

[Your AI Client] ←MCP Protocol→ [MCP Brain Server] ←→ [Supabase Postgres + pgvector]

1. You tell your AI something worth remembering
2. The AI calls create_memory via MCP
3. Your memory is embedded (OpenAI text-embedding-3-small) and stored
4. Later, from ANY AI client, search your memories semantically
5. Your AI has context about you that persists forever

Stack

Component	Technology
MCP Server	Python + FastMCP
Database	Supabase Postgres + pgvector
Auth	Supabase Auth (OAuth 2.1 — works with all MCP clients)
Embeddings	OpenAI text-embedding-3-small (1536 dimensions)
Search	Hybrid: pgvector similarity + Postgres full-text + RRF ranking
Billing	Stripe ($7.99/month)
Package Manager	UV (no pip)

MCP Tools

Tool	Description
create_memory	Store a new memory with auto-embedding and classification
search_memories	Hybrid semantic + full-text search across your brain
list_memories	Browse recent memories, filter by type
delete_memory	Remove a specific memory
brain_stats	View your memory statistics

Quick Start

Prerequisites

• Python 3.12+
• UV (curl -LsSf https://astral.sh/uv/install.sh | sh)
• Supabase project (free tier works)
• OpenAI API key

Setup

# Clone and enter the project
cd ~/Projects/mcp-brain

# Install dependencies with UV
uv sync

# Copy environment template
cp .env.example .env
# Edit .env with your Supabase and OpenAI credentials

# Run database migrations
# (paste migrations/001_initial_schema.sql into Supabase SQL Editor)

# Start the server
uv run python -m src.main

Connect from Claude Desktop

Add to your Claude Desktop MCP config:

{
  "mcpServers": {
    "brain": {
      "url": "http://localhost:8080/mcp",
      "transport": "streamable-http"
    }
  }
}

Connect from Claude Code

claude mcp add brain --transport http http://localhost:8080/mcp

Development

# Install with dev dependencies
uv sync --all-extras

# Run tests
uv run pytest

# Lint
uv run ruff check src/

# Format
uv run ruff format src/

Architecture

See docs/RESEARCH.md for comprehensive market research, competitive analysis, and technical feasibility assessment.

License

MIT