MCP TradingView Server

An MCP (Model Context Protocol) server that provides access to TradingView technical indicators using the tradingview_scraper library.

Architecture

graph TB
    Claude[Claude Desktop] --> MCP[FastMCP Server]
    
    subgraph Server["MCP TradingView Server"]
        MCP --> Tools[Tools]
        MCP --> Resources[Resources]
        
        subgraph Tools
            GI[get_indicators]
            GSI[get_specific_indicators]
            GHD[get_historical_data]
        end
        
        subgraph Resources
            RES["indicators/{symbol}"]
        end
    end
    
    subgraph Library["TradingView Scraper Library"]
        Ind[Indicators]
        Stream[Streamer]
        Converter[OHLCVConverter]
    end
    
    subgraph TV["TradingView API"]
        TechInd[(Technical Indicators)]
        OHLCV[(Real-time OHLCV Data)]
    end
    
    subgraph Export["Export Directory"]
        JSON[JSON Files]
    end
    
    Tools --> Ind
    Tools --> Stream
    Stream --> Converter
    Ind --> TechInd
    Stream --> OHLCV
    Tools --> JSON
    
    classDef cloud fill:#e1f5fe
    classDef database fill:#f3e5f5
    classDef file fill:#e8f5e8
    classDef component fill:#fff3e0
    
    class TV cloud
    class TechInd,OHLCV database
    class JSON file
    class Server,Library,Tools,Resources component

Loading

Note: The Streamer class provides native timeframe support, avoiding over-collection of data. The OHLCVConverter efficiently aggregates 1-minute data to requested timeframes when needed.

Features

• Fetch all technical indicators for any trading symbol
• Retrieve specific indicators by name
• Collect real-time OHLCV (candlestick) data with native timeframe support
• Support for multiple exchanges and all major timeframes (1m to 1M)
• Efficient data streaming using TradingView's native Streamer class
• Resource-based access to indicator data
• Export data to JSON files

Installation

Prerequisites

• Python 3.8 or higher
• uv (recommended for fast Python package and environment management)

Setup with uv

1. Install uv (if not already installed):

   # macOS/Linux
   curl -LsSf https://astral.sh/uv/install.sh | sh
   
   # Or with Homebrew
   brew install uv
   
   # Windows
   powershell -c "irm https://astral.sh/uv/install.ps1 | iex"

2. Create and setup the project:

   # Navigate to the project directory
   cd /path/to/mcp-tradingview-server
   
   # Create a virtual environment with Python 3.11
   uv venv --python 3.11
   
   # Activate the environment (optional, uv can work without activation)
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install the package:

   # Install in development mode with uv
   uv pip install -e .
   
   # Or install dependencies directly
   uv pip install mcp tradingview-scraper

MCP Server Configuration

Claude Desktop Setup

1. Locate Claude Desktop configuration:

   • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
   • Windows: %APPDATA%\Claude\claude_desktop_config.json
   • Linux: ~/.config/Claude/claude_desktop_config.json

2. Edit the configuration file:

   {
     "mcpServers": {
       "tradingview": {
         "command": "/path/to/mcp-tradingview-server/.venv/bin/python",
         "args": ["/path/to/mcp-tradingview-server/src/tradingview_server.py"],
         "cwd": "/path/to/mcp-tradingview-server"
       }
     }
   }

   Important:

   • Replace /path/to/mcp-tradingview-server with the ABSOLUTE path to this project
   • Use absolute paths only (e.g., /Users/username/projects/mcp-tradingview-server)
   • Do NOT use ~ or relative paths as Claude Desktop won't expand them

3. Alternative configuration using uv (requires full path):

   {
     "mcpServers": {
       "tradingview": {
         "command": "/Users/username/.local/bin/uv",
         "args": ["run", "python", "-m", "tradingview_server"],
         "cwd": "/path/to/mcp-tradingview-server/src"
       }
     }
   }

   Note: You must use the full path to uv (find it with which uv in terminal)

4. Restart Claude Desktop after updating the configuration.

Verifying the Setup

1. Test the server locally:

   cd /path/to/mcp-tradingview-server
   python test_server.py

   This will test all available functions:

   • Technical indicators retrieval
   • Specific indicators filtering
   • Resource-based access
   • Historical OHLCV data with different timeframes
   • Data export functionality

2. Check server is running in Claude:

   • Open Claude Desktop
   • Look for "tradingview" in the available MCP servers
   • Try using the tools with example prompts below

Usage

Example Prompts for Claude Desktop

Once configured, you can use these prompts in Claude Desktop:

Basic Indicators:

• "Get Bitcoin indicators using the tradingview server"
• "Show me all technical indicators for AAPL"
• "What are the current indicators for EURUSD?"

Specific Indicators:

• "Fetch RSI and MACD for Tesla using the tradingview tool"
• "Get the RSI, Stochastic, and Bollinger Bands for BTCUSD"
• "Show me momentum indicators for ETHUSD"

With Timeframes:

• "Get 4-hour indicators for Bitcoin on Binance"
• "Show daily indicators for SPY on NYSE"
• "Fetch 15-minute indicators for GBPUSD"

Historical Data:

• "Get historical OHLC data for Bitcoin"
• "Collect 200 candlesticks for Tesla and save to file"
• "Stream real-time price data for EURUSD"
• "Get OHLCV data for Ethereum on Binance"

Technical Analysis:

• "Analyze SOLUSDT using tradingview indicators on the 1-hour timeframe"
• "What do the technical indicators say about MSFT?"
• "Get indicators for gold (XAUUSD) and interpret them"

Tool Interaction Overview

graph TD
    User[👤 Trader/Analyst] --> TA[📊 Technical Analysis]
    User --> HD[📈 Historical Data]
    User --> RA[⚡ Resource Access]
    
    subgraph Tools["🛠️ MCP Tools"]
        GI[get_indicators]
        GSI[get_specific_indicators]
        GHD[get_historical_data]
        Resource["indicators/{symbol}"]
    end
    
    subgraph Sources["📡 Data Sources"]
        TVTI[(TradingView<br/>Technical Indicators)]
        TVRT[(TradingView<br/>Real-time Stream)]
    end
    
    subgraph Outputs["📤 Output Formats"]
        JSON[📄 JSON Export]
        Console[🖥️ Console Output]
        MCP_OUT[📋 MCP Response]
    end
    
    TA --> GI
    TA --> GSI
    HD --> GHD
    RA --> Resource
    
    GI --> TVTI
    GSI --> TVTI
    GHD --> TVRT
    Resource --> TVTI
    
    GI --> Console
    GSI --> Console
    GHD --> JSON
    GHD --> Console
    Resource --> MCP_OUT
    
    classDef userAction fill:#e3f2fd
    classDef tool fill:#f3e5f5
    classDef source fill:#e8f5e8
    classDef output fill:#fff3e0
    
    class TA,HD,RA userAction
    class GI,GSI,GHD,Resource tool
    class TVTI,TVRT source
    class JSON,Console,MCP_OUT output

Loading

Key Features:

• GHD (get_historical_data): Supports all timeframes (1m, 5m, 15m, 30m, 1h, 2h, 4h, 1d, 1w, 1M)
• GSI (get_specific_indicators): Efficient filtering - only requested indicators returned
• TVRT: WebSocket streaming for real-time data with native timeframe support

Available Tools

get_indicators

Retrieve all technical indicators for a symbol.

Parameters:

• symbol (required): Trading symbol - Must be a complete trading pair (e.g., "BTCUSD", "EGLDUSDT", "AAPL")
  • For crypto: Use full pairs like "BTCUSD", not just "BTC"
  • For stocks: Ticker symbols work (e.g., "AAPL", "TSLA")
• exchange: Exchange name (default: "BINANCE")
• timeframe: Timeframe (default: "1h") - Options: "1m", "5m", "15m", "30m", "1h", "4h", "1d", "1w", "1M"
• all_indicators: Fetch all indicators (default: true)
• export_result: Export to JSON file (default: false)

get_specific_indicators

Retrieve specific indicators for a symbol.

Parameters:

• symbol (required): Trading symbol
• indicators (required): List of indicators to fetch (e.g., ["RSI", "MACD", "BB"])
• exchange: Exchange name (default: "BINANCE")
• timeframe: Timeframe (default: "1h")
• export_result: Export to JSON file (default: false)

get_historical_data

Retrieve real-time OHLCV (Open, High, Low, Close, Volume) data stream with automatic timeframe conversion.

Parameters:

• symbol (required): Trading symbol (e.g., "BTCUSD", "AAPL")
• exchange: Exchange name (default: "BINANCE")
• timeframe: Timeframe for candles (default: "1h") - Options: "1m", "5m", "15m", "30m", "1h", "2h", "4h", "1d", "1w", "1M"
• max_records: Maximum number of OHLC records to collect in the requested timeframe (default: 100)
• export_result: Export to JSON file (default: false)

How it works: Uses TradingView's native Streamer class to efficiently collect data in your specified timeframe directly, without needing to over-collect and convert data.

Historical Data Flow

sequenceDiagram
    participant Claude as Claude Desktop
    participant Tool as get_historical_data
    participant Stream as Streamer
    participant Converter as OHLCVConverter
    participant WS as TradingView WebSocket
    participant Export as Export Directory

    Claude->>+Tool: get_historical_data<br/>symbol=BTCUSD<br/>timeframe=4h<br/>max_records=10
    
    Tool->>+Stream: Streamer(export_result=True)
    
    Stream->>+Converter: OHLCVConverter<br/>target_timeframe=4h
    Converter-->>-Stream: timeframe validation ✓
    
    Stream->>+WS: WebSocket connection
    Stream->>WS: create_series<br/>timeframe=1m, records=10
    WS-->>-Stream: 1-minute OHLCV data stream
    
    Stream->>+Converter: convert 1m data to 4h
    Note over Converter: Efficiently aggregates:<br/>• High = max(highs)<br/>• Low = min(lows)<br/>• Open = first open<br/>• Close = last close<br/>• Volume = sum(volumes)
    Converter-->>-Stream: aggregated 4h candles
    
    alt export_result=true
        Stream->>+Export: save JSON file
        Export-->>-Stream: file saved
    end
    
    Stream-->>-Tool: ohlc and indicator data
    
    Tool-->>-Claude: Response with success,<br/>timeframe, records count,<br/>data array, and export file
    
    Note over WS: Real-time streaming<br/>avoids historical API limits

Loading

Returns:

• success: Operation status
• symbol: Trading symbol
• exchange: Exchange name
• timeframe: The timeframe of the returned data
• records_collected: Number of records in the requested timeframe
• data: Array of OHLCV candles
• export_file: Export filename (if exported)

Available Resources

• indicators/{symbol}: Get current indicator values as a formatted text resource

Example Usage

# Get all indicators for Bitcoin
result = await get_indicators("BTCUSD", exchange="BINANCE", timeframe="4h")

# Get specific indicators
result = await get_specific_indicators(
    "AAPL", 
    indicators=["RSI", "MACD"], 
    exchange="NASDAQ"
)

# Get historical OHLCV data
result = await get_historical_data(
    "BTCUSD",
    exchange="BINANCE",
    timeframe="4h",  # Will collect 1m data and convert to 4h candles
    max_records=50,
    export_result=True
)

Claude AI Integration

This MCP server includes a CLAUDE.md file that provides detailed documentation for Claude AI, including:

• Tool usage examples
• Best practices for indicator analysis
• Common prompt patterns
• Exchange and timeframe recommendations

When using this server with Claude Desktop, Claude will automatically reference this documentation to provide better assistance with technical analysis.

Supported Exchanges

The server supports all exchanges available in TradingView, including:

• BINANCE
• COINBASE
• KRAKEN
• NASDAQ
• NYSE
• And many more...

Timeframes

• 1m, 5m, 15m, 30m (minutes)
• 1h, 4h (hours)
• 1d (daily)
• 1w (weekly)
• 1M (monthly)

Troubleshooting

Common Issues

1. "spawn python ENOENT" or "spawn uv ENOENT" error:

   • This means Claude Desktop can't find the command in its PATH
   • Solution: Use ABSOLUTE paths in your configuration
   • Example: Instead of "command": "python", use "command": "/Users/yourname/project/.venv/bin/python"
   • Find the correct path by running which python or pwd in your project directory

2. "Module not found" error:

   • Ensure you're in the correct directory with the .venv
   • Verify installation with: uv pip list | grep tradingview

3. MCP server not showing in Claude:

   • Check the configuration file path is correct
   • Ensure ALL paths in the config are absolute (no ~ or relative paths)
   • Restart Claude Desktop after configuration changes

4. Permission errors:

   • Make sure the Python script is executable: chmod +x src/tradingview_server.py

Debug Mode

To run the server in debug mode for troubleshooting:

cd /path/to/mcp-tradingview-server
uv run python -m tradingview_server

# Or with activated environment
source .venv/bin/activate
python -m tradingview_server

Contributing

Feel free to submit issues and enhancement requests!

License

This project is provided as-is for educational and research purposes.