开源MCP工具：MaverickMCP

Prerequisites

• Python 3.12+: Core runtime environment
• uv: Modern Python package manager (recommended)
• TA-Lib: Technical analysis library for advanced indicators
• Redis (optional, for enhanced caching)
• PostgreSQL or SQLite (optional, for data persistence)

Installing TA-Lib

TA-Lib is required for technical analysis calculations.

macOS and Linux (Homebrew):

brew install ta-lib

Windows (Multiple Options):

Option 1: Conda/Anaconda (Recommended - Easiest)

conda install -c conda-forge ta-lib

Option 2: Pre-compiled Wheels 1. Download the appropriate wheel for your Python version from: - cgohlke/talib-build releases - Choose the file matching your Python version (e.g., TA_Lib-0.4.28-cp312-cp312-win_amd64.whl for Python 3.12 64-bit) 2. Install using pip:

pip install path/to/downloaded/TA_Lib-X.X.X-cpXXX-cpXXX-win_amd64.whl

Option 3: Alternative Pre-compiled Package

pip install TA-Lib-Precompiled

Option 4: Build from Source (Advanced) If other methods fail, you can build from source: 1. Install Microsoft C++ Build Tools 2. Download and extract ta-lib C library to C:\ta-lib 3. Build using Visual Studio tools 4. Run pip install ta-lib

Verification: Test your installation:

python -c "import talib; print(talib.__version__)"

Installing uv (Recommended)

```bash

Install dependencies and create virtual environment in one command

uv sync

Installation

Option 1: Using uv (Recommended - Fastest)

```bash

Create virtual environment and install

python -m venv .venv source .venv/bin/activate # On Windows: .venv\Scripts\activate pip install -e .

Ultra-fast one-liner (no installation needed)

uvx ty check . # Run ty directly without installing ```

Docker (Optional)

For containerized deployment:

```bash

Using uv in Docker (recommended for faster builds)

docker build -t maverick_mcp . docker run -p 8003:8003 --env-file .env maverick_mcp

Or start with docker-compose

docker-compose up -d ```

Note: The Dockerfile uses uv for fast dependency installation and smaller image sizes.

Quick Start

Usage Examples

Backtesting Example

Once connected to Claude Desktop, you can use natural language to run backtests:

"Run a backtest on AAPL using the momentum strategy for the last 6 months"

"Compare the performance of mean reversion vs trend following strategies on SPY"

"Optimize the RSI strategy parameters for TSLA with walk-forward analysis"

"Show me the Sharpe ratio and maximum drawdown for a portfolio of tech stocks using the adaptive ML strategy"

"Generate a detailed backtest report for the ensemble strategy on the S&P 500 sectors"

Technical Analysis Example

"Show me the RSI and MACD analysis for NVDA"

"Identify support and resistance levels for MSFT"

"Get full technical analysis for the top 5 momentum stocks"

Portfolio Management Example (NEW)

"Add 10 shares of AAPL I bought at $150.50"

"Show me my portfolio with current prices"

"Compare my portfolio holdings"  # No tickers needed!

"Analyze correlation in my portfolio"  # Auto-detects your positions

"Remove 5 shares of MSFT"

Portfolio Optimization Example

"Optimize a portfolio of AAPL, GOOGL, MSFT, and AMZN for maximum Sharpe ratio"

"Calculate the correlation matrix for my tech portfolio"

"Analyze the risk-adjusted returns for energy sector stocks"

Test Examples - Validate All Features

Test the comprehensive research capabilities and parallel processing improvements with these examples:

Copy environment template

cp .env.example .env

Copy environment template

cp .env.example .env

Configuration

Configure MaverickMCP via .env file or environment variables:

Essential Settings:

• REDIS_HOST, REDIS_PORT - Redis cache (optional, defaults to localhost:6379)
• DATABASE_URL - PostgreSQL connection or sqlite:///maverick_mcp.db for SQLite (default)
• LOG_LEVEL - Logging verbosity (INFO, DEBUG, ERROR)
• S&P 500 data automatically seeds on first startup

Required API Keys:

• TIINGO_API_KEY - Stock data provider (free tier available at tiingo.com)

Optional API Keys:

• OPENROUTER_API_KEY - Strongly Recommended for Research: Access to 400+ AI models with intelligent cost optimization (40-60% cost savings)
• EXA_API_KEY - Recommended for Research: Web search capabilities for comprehensive research
• OPENAI_API_KEY - Direct OpenAI access (fallback)
• ANTHROPIC_API_KEY - Direct Anthropic access (fallback)
• FRED_API_KEY - Federal Reserve economic data
• TAVILY_API_KEY - Alternative web search provider
• ADANOS_API_KEY - Optional Adanos Market Sentiment API access for stock sentiment from Reddit, X / FinTwit, News, and Polymarket (docs)
• ADANOS_API_BASE_URL - Optional Adanos API base URL override; defaults to https://api.adanos.org

When configured, use the data_get_adanos_market_sentiment MCP tool for ticker-level sentiment (ticker="AAPL") or market-wide sentiment (ticker=null). Optional sources values are reddit, x, news, and polymarket.

Performance:

• CACHE_ENABLED=true - Enable Redis caching
• CACHE_TTL_SECONDS=3600 - Cache duration

Copy and configure environment

cp .env.example .env

Research timeouts → Check logs, increase timeout settings

```

Fast Development Options

• Hot Reload: uv run python tools/hot_reload.py - Auto-restart on changes
• Fast Startup: ./tools/fast_dev.sh - < 3 second startup
• Quick Testing: uv run python tools/quick_test.py --test stock - Test specific features
• Experiment Harness: Drop .py files in tools/experiments/ for auto-execution

Add your Tiingo API key (free at tiingo.com)

#### Option 2: Using pip (Traditional)

Add your Tiingo API key (free at tiingo.com)

```

- HTTP endpoint: http://localhost:8003/mcp/

- SSE endpoint: http://localhost:8003/sse/

- HTTP endpoint: http://localhost:8003/mcp/ (streamable-http - use with mcp-remote)

- SSE endpoint: http://localhost:8003/sse/ (SSE - direct connection only, not mcp-remote)

Quick Development Workflow

make dev               # Start everything
make stop              # Stop services
make tail-log          # Follow server logs
make test              # Run tests quickly
make experiment        # Test custom analysis scripts

Alternative: via pip

pip install uv ```

Alternative startup methods

./scripts/start-backend.sh --dev # Script-based startup ./tools/fast_dev.sh # Ultra-fast startup (< 3 seconds) uv run python tools/hot_reload.py # Auto-restart on file changes

Alternative: Direct pytest (if activated in venv)

pytest # Unit tests only pytest --cov=maverick_mcp # With coverage pytest -m "" # All tests (requires PostgreSQL/Redis) ```

Alternative: Direct commands (if activated in venv)

ruff check . # Linting ruff format . # Formatting ty check . # Type checking

Troubleshooting