ValueCell

ValueCell is an open-source, multi-agent platform for financial applications, published by the ValueCell-ai organization under the Apache 2.0 license. It provides investment agents for research, news tracking, and strategy-driven trading, available as a hosted service at valuecell.ai and as a desktop app for macOS and Windows. The README states that sensitive information, such as API secrets, is stored locally on the user's device.

What It Is

ValueCell is a multi-agent system for stock selection, research, tracking, and trading. Users configure an LLM provider, then interact with agents through a web interface. The README describes agents including a DeepResearch Agent that retrieves and analyzes fundamental documents to produce summaries, a Strategy Agent that runs multi-strategy crypto trading, and a News Retrieval Agent that delivers personalized scheduled news. The README says more agents are planned.

How Live Trading Works

Users add an AI model API key, set up exchange API credentials, and combine the model with an exchange to create a custom strategy. They can then start or stop traders and monitor performance in real time. The README lists Binance, Hyperliquid, and OKX as tested exchanges. Coinbase, Gate.io, MEXC, and others are partially tested. Trading currently supports contract trading only, with spot implemented as 1X contracts.

Models, Data, and Integrations

The README lists LLM providers including OpenRouter, SiliconFlow, Azure, OpenAI-compatible endpoints, Google, OpenAI, and DeepSeek. Market data covers the US, crypto, Hong Kong, and China markets. Agents can integrate with LangChain and Agno through the A2A protocol.

Hosted Product and Dynamic Programming

The product site describes in-depth research, algorithmic backtesting, and a dynamic programming layer. That layer covers intent orchestration with ReAct scheduling, skill routing, coding agents for non-standard modeling, and a sandboxed execution loop. Mobile apps are listed as coming soon.

Running It Locally

Developers can clone the repository and run start.sh on Linux or macOS, or start.ps1 on Windows. The web UI is served locally at port 1420. The README warns that investing involves risk.