Summary
Large language models are poor at picking profitable trades on their own, but with the right tools, skills, and MCP servers they become far more useful for algorithmic traders. This video walks through seven free GitHub repositories and skills that one trader uses every day to research, backtest, and monitor strategies with AI agents.
The list begins with a TradingView indicator search MCP that lets Claude browse open-source indicators instead of defaulting to basic RSI or moving-average ideas. It then covers a unified TradingView MCP server that controls most of the platform from inside Claude or ChatGPT, along with a warning that TradingView may not tolerate automation and that using such servers could put an account at risk. A free alternative, TradingKit, is shown backtesting strategies across markets such as gold, silver, and Bitcoin.
Other picks include an official broker MCP server for account reports and equity tracking, the paid Nansen Smart Money skill for crypto market context, and a position-sizing tool. The video also warns about fake broker skills and explains why API keys should be exported to a local environment instead of pasted into a chat. It is a practical look at how MCP servers and agent skills can plug into Claude, Codex, and other agents for trading research.
šŗ Source: Trading with DaviddTech Ā· Published October 26, 2026
š·ļø Format: Roundup
