TradeMemory Protocol is a Python-based MCP server that provides AI agents with persistent, outcome-weighted memory for trading decisions. It journals each trade, stores contextual information, and uses an outcome-weighted recall engine to surface relevant memories when needed. Tools available include memory recall and pattern discovery to inform strategy adjustments, as well as a memory-weighted sizing approach (Kelly-from-memory) to inform position sizing. You can run simulated demonstrations, integrate with Claude, Cursor, or other MCP clients, and reference the Open Claw skill integration. To use it, start the server and connect your MCP client to the tradememory server endpoint; then issue prompts like recording trades, recalling memories, or requesting performance reflections to guide future decisions.

Prerequisites:

• Python 3.10+ installed on your system
• Git installed
• Access to install Python packages (pip)

Install steps:

1. Clone the repository git clone https://github.com/mnemox-ai/tradememory-protocol.git
2. Change to the project directory cd tradememory-protocol
3. Install in editable mode pip install -e .
4. Run the server locally python -m src.tradememory.server

Optional Docker-based setup:

• Build and run with Docker (if you prefer containerized setup): docker compose up -d

  or build and run manually

  docker build -t tradememory . docker run -p 8000:8000 -e ANTHROPIC_API_KEY=your-key tradememory

Integration notes:

• To connect with Claude Desktop/Code, add the MCP server entry using uvx with the tradememory-protocol package name as shown in the documentation.
• For other MCP clients, add a server mapping for tradememory with command set to python and the path to the server module as shown above.

Tips and considerations:

• The server exposes an MCP API over HTTP on localhost:8000 by default; ensure network access is allowed in your environment.
• Set ANTHROPIC_API_KEY or other required API keys as environment variables if your memory model requires external services.
• This server supports legacy tooling compatibility (v0.3.x tools) for recall and memory exploration.
• If you encounter port conflict, modify the server start parameters or environment to run on a different port if supported by your deployment.
• When using with Claude or Cursor, follow the Quick Start snippets in the README to configure your mcp client mappings.