TraderAlice/OpenAlice

OpenAlice – AI‑powered trading workspace

What it is – OpenAlice is a locally‑run application that lets a single user combine large‑language‑model agents (Claude, Codex, Pi, etc.) with market data, research tools, and brokerage accounts. It treats each research or trading project as a Workspace (a Git‑backed folder) so the whole workflow – prompts, code, data files, and results – stays version‑controlled and inspectable.

Core ideas

Concept	What it does
Agents + market tools	Connects supported LLM agents to live price/ fundamentals/ news feeds and quantitative libraries. Agents can be asked to fetch data, run analyses, or generate code.
Workspaces	Each workspace is a directory‑based Git repo that stores the conversation history, research files, and any generated code. Specialized workspaces (Chat, AutoQuant, Auto Prediction) have their own UI “Studio”.
Tracked & Issues	Links research artefacts to assets (stocks, ETFs, etc.) and lets you turn follow‑up tasks into scheduled Issues, similar to a todo list that runs automatically.
Inbox	Central view of reports, questions and updates generated by agents, with a one‑click path back to the originating session.
Unified Trading Account & Trading‑as‑Git	Optional broker integration (paper, demo or real accounts). Agents can draft trade orders; the user reviews and commits them via a Git‑style workflow. Execution is marked as beta.

How you get it

• Desktop app (macOS/Windows) – includes the “Pi” model and a UI. Download from the GitHub releases page.
• CLI / Docker – for servers or headless use. Installation instructions are in docs/cli-installer.md and the Docker guide.
• After install, add credentials for the LLM you want to use (Claude, OpenCode, etc.) and optionally configure a broker.

Typical workflow

1. Create a Workspace for a stock or idea (e.g., “NVDA thesis”).
2. Ask an agent via the “Ask Alice” UI to research fundamentals, sector trends, price behaviour, and to save the output.
3. Iterate – add follow‑up questions, turn the work into a recurring Issue (e.g., weekly scan), or launch AutoQuant to run quantitative experiments.
4. Review results in Inbox, open the generated report, and, if satisfied, let the agent propose a trade.
5. Approve trade through the Trading‑as‑Git interface; the order is sent to the connected broker.

Who it’s for

• Individual quantitative traders or hobbyists who want a single place to orchestrate LLM‑driven research and automated trading.
• Developers who prefer their research artefacts to be version‑controlled and reproducible.
• Anyone comfortable running software locally and managing API keys for LLMs and brokers.

Safety notes

• Trading execution is experimental; the README warns to start with simulators or paper accounts.
• The software stores broker credentials encrypted under ~/.openalice but does not guarantee loss‑prevention.

Project health

• Open‑source under AGPL‑3.0.
• Uses a modern JavaScript/TypeScript stack (pnpm, dev script).
• Active community channels: Discord, QQ, X (Twitter) and a public issue tracker.
• Star history badge suggests modest community interest.

Bottom line – OpenAlice is a genuine AI‑orchestrated trading platform that stitches together LLM agents, market data, research management, and broker APIs into a locally hosted, Git‑backed workflow. It is squarely in the AI‑agent‑for‑finance space and provides concrete tooling rather than just a curated list or tutorial.