mcp-tradingview

This MCP server exposes TradingView indicators and OHLCV data via a FastMCP v2 interface powered by the tradingview_scraper library. It provides tools to fetch full indicator snapshots, filter specific indicators, and stream historical OHLCV data. Clients like Claude Desktop can connect over stdio, SSE, or HTTP transports. Use the standard MCP commands exposed by the server to query indicators for a given symbol and exchange, retrieve historical data, or export results for later analysis. The server emphasizes a clear separation between indicator metadata and actual data, with optional export functionality for long-running experimentation.

Prerequisites:

• Python 3.11+ installed on your system
• uv (AstraLuv) installed

Manual setup steps:

1. Create a Python virtual environment and install the package in editable mode:

uv venv --python 3.11 && source .venv/bin/activate
uv pip install -e .

2. Run the MCP server in stdio mode (default):

uv run mcp-tradingview

To run over SSE or HTTP transport, use:

uv run mcp-tradingview --transport sse --host 0.0.0.0 --port 8000
uv run mcp-tradingview --transport http --host 0.0.0.0 --port 8000 --path /mcp

Note: The setup script (setup.sh) is provided in automated setups to configure Claude Desktop automatically. If you use that, follow the script prompts and then restart Claude Desktop to load the new MCP configuration.

Environment variables FASTMCP_HOST, FASTMCP_PORT, and FASTMCP_PATH override default bind settings when provided. When exporting results, large archives should be kept out of version control (export/ directory is intended for runtime outputs). If you modify Claude Desktop configuration, restart Claude Desktop to apply changes. For debugging, enable verbose logging in the Python app and reference export/ fixtures during testing. The tradingview_scraper utilities power the indicators and OHLCV data retrieval; ensure external dependencies for scraping are accessible in your environment.