wtf is ai-openclaw-cli?

OpenFinClaw is a command-line tool and MCP server that turns a single natural-language prompt into a full quantitative trading research workflow: researching an idea, generating a trading strategy, backtesting it against historical data, and optionally paper trading it, all from inside an AI coding assistant such as Claude Code or Cursor. MCP, short for Model Context Protocol, is the standard that lets an AI agent call outside tools, this project uses it to plug its research and backtesting capabilities into more than twenty different AI platforms. The tool includes over sixty built-in analysis skills spanning technical, fundamental, sentiment, risk, timing, and factor-based approaches, and covers five markets including US stocks, Chinese A-shares, Hong Kong stocks, crypto, and forex. A user can type a plain request, such as asking for RSI divergence signals on a stock or a momentum strategy backtested over two years, and the agent streams back research, a strategy definition, and backtest performance metrics. There is also a browser-based playground that requires no installation or API key for trying it out first. Installation is done through a Node package manager command that runs an interactive setup wizard: it writes MCP configuration files for detected AI agents, saves an API key locally, and registers a skill file so Claude Code and Cursor automatically recognize quant-related requests. The command-line interface groups its commands into a research assistant, a strategy management set, and system utilities like an installer, a doctor diagnostic check, and an update command. A community feature lets users browse a leaderboard of published trading strategies ranked by backtested performance, fork any strategy to their own machine to modify it, then publish their version back for independent server-side verification, so results are not self-reported. The project is written in TypeScript, distributed as an npm package, and released under the MIT license.