Run TradingAgents as a desktop app.
TradingAgents, the open-source multi-agent framework from TauricResearch, runs from the command line and expects a Python environment. Trading Agents Lab is an open-source desktop application built on the same multi-agent approach, packaged for macOS, Windows, and Linux. You download it, point it at a model, and read a full diligence debate end to end. No virtualenv, no pip, no notebook.
macOS · Windows · Linux · BYO LLM key or run locally · AGPL-3.0
What stays, what the desktop adds.
The agent architecture is the same idea you came for: specialised analysts and researchers study a stock, a trader and risk seats debate from opposing angles, and the run converges on a recommendation you can read in full. The desktop app layers a usable experience on top.
- No Python setup
Install a signed desktop build instead of cloning a repo and managing a Python environment.
- Bring your own model, or run local
OpenAI, Anthropic, OpenRouter, Google Gemini, xAI Grok, and MiniMax via your own key, or run fully offline through Ollama and LM Studio.
- Use a ChatGPT subscription, no per-token billing
Sign in with ChatGPT OAuth to power debates from your existing plan instead of paying the API per token.
- Read the debate
Every analyst argument, rebuttal, and the final call are laid out so you can follow the reasoning, not just the verdict. See Reading the debate.
- Local-only by design
No accounts, no telemetry. Keys live in your OS keychain and runs are stored on your machine. See Security and privacy.
Getting started.
Three docs cover the path from download to your first run:
Getting started
Install, pick a provider, run your first diligence on any ticker.
Local models
Run offline with Ollama or LM Studio, no API key and no per-token cost.
ChatGPT OAuth
Power debates from your paid ChatGPT plan instead of the per-token API.
New to the agent model itself? Start with How it works, then browse the full documentation.
Open source, like the project it builds on.
Trading Agents Lab is a fork of TradingAgents, released under AGPL-3.0. The full source is open to inspection, and you are free to study it, run it, and build on it under the same license.
For educational and research purposes only, not investment advice.