MCP

MCP Interface · Let AI Agents Control StratCraft

Connect any MCP-compatible AI agent to StratCraft. Run backtests, list strategies, and generate signals with natural language commands.

What is MCP?

Model Context Protocol (MCP) is an open standard by Anthropic that lets AI models interact with external tools and services as structured function calls. StratCraft implements an MCP server, turning your strategy system into a tool that any compatible AI agent can call.

Example Commands

Natural language commands that AI agents can execute against StratCraft.

Run a backtest on AAPL with RSI-14 strategy from 2020 to 2024
List my top 10 strategies sorted by Sharpe ratio
Generate 20 momentum strategy variants and backtest them
Show me which strategies have Sharpe above 1.5 and max drawdown below 15%
Compose the top 50 signals into a portfolio using FrankWolfe optimization

MCP Tool Calls

MCP

# Claude can call StratCraft as a tool
run_backtest(
  strategy="rsi_14",
  symbol="AAPL",
  period="2020-2024"
)

list_strategies(
  sort_by="sharpe",
  limit=10
)

generate_strategy(
  description="momentum + mean reversion hybrid"
)

compose_portfolio(
  method="frankwolfe",
  top_n=50
)

Supported Agents

Works with any MCP-compatible AI model.

Claude (Anthropic)

GPT-4 (OpenAI)

Gemini (Google)

Local LLMs via Ollama

Setup

One configuration file to connect any MCP agent.

Point your MCP client at the StratCraft MCP server. No additional infrastructure required — the server runs locally alongside StratCraft.

Начните свою фабрику сигналов

Уровень бесплатного пользования включает механизм бэктестинга C++, обнаружение режима и данные YFinance + Dukascopy — все, что вам нужно, чтобы начать масштабное строительство.

Star on GitHub See AI Engine →