Vibe-Trading is an AI-powered multi-agent finance workspace designed to translate natural language requests into executable trading strategies, research insights, and portfolio analyses across global markets. It empowers users, from engineers to researchers, by automating complex financial tasks, offering comprehensive trading capabilities, and providing deep market insights through an advanced agent framework.

How It Works

Vibe-Trading employs a ReAct agent core orchestrated by a Directed Acyclic Graph (DAG) for multi-agent "Swarm Intelligence." It features 68 specialized finance skills across seven categories, enabling modular functionality for tasks ranging from data sourcing to quantitative analysis. The system integrates with multiple LLM providers and utilizes a robust data handling layer with five sources and automatic fallback mechanisms, ensuring data availability across diverse markets. Strategies can be automatically generated, tested against historical data, and exported to Pine Script v6 for TradingView.

Quick Start & Requirements

Installation is straightforward via PyPI (pip install vibe-trading-ai), Docker, or as an MCP plugin. Local installation requires Python 3.11+ and an LLM API key (or local Ollama setup). Docker setup takes approximately 2 minutes, while local installation is estimated at 5 minutes. Key commands include vibe-trading (interactive CLI/TUI), vibe-trading serve (web UI), and vibe-trading-mcp (MCP server). Links to demos and the GitHub repository are available.

Highlighted Details

• Extensive functionality with 68 specialized finance skills covering data sourcing, strategy generation, analysis, asset classes, crypto, flow, and tooling.
• 29 pre-configured agent swarm team presets offer ready-to-use financial workflows for investment, trading, and risk management.
• Robust cross-market backtesting supports A-shares, HK/US equities, crypto, futures, and forex, with per-market rules and Pine Script v6 export for TradingView.
• Supports numerous LLM providers (e.g., OpenAI, Groq, Gemini, Ollama) with configurable timeouts and automatic fallback for data sources, ensuring broad accessibility.

Maintenance & Community

The project encourages community contributions via its CONTRIBUTING.md file and tags good first issues. Vibe-Trading is part of the HKUDS agent ecosystem, which includes projects like ClawTeam and OpenSpace, suggesting active development and integration within a broader AI framework. OpenSpace skills evolve autonomously, enhancing the project's capabilities over time.

Licensing & Compatibility

Vibe-Trading is released under the MIT License, which is permissive and generally compatible with commercial use and closed-source linking.

Limitations & Caveats

This tool is strictly for research, simulation, and backtesting purposes. It does not provide investment advice and does not execute live trades. Past performance is not indicative of future results.