Skip to content

README.en.md

Latest commit

 

History

History
219 lines (154 loc) · 8.35 KB

README.en.md

File metadata and controls

219 lines (154 loc) · 8.35 KB
LabOS Banner

LabOS

Next-Gen AI Research Automation Platform

Full research lifecycle coverage: Ideation · Literature Survey · Hypothesis Design · Experiment Execution · Paper Writing

Local-first · Multi-model · Fully open-source

License: AGPL-3.0 Python 3.8+ FastAPI SQLite PRs Welcome

中文 · Quick Start · Screenshots · Contributing · Roadmap

✦ Core Capabilities

Research Pipeline · 4-stage automation  │  AI Chat · Multi-model streaming  │  Experiment Execution · SSH to remote GPU servers  │  Codex Integration · AI writes experiment code

Memory System · Cross-experiment knowledge persistence  │  Multi-model Config · Separate profiles for chat / code / paper / experiment  │  Stage Reports · Survey → Analysis → Experiment  │  Local-first · Your data stays with you

Demo Video

Screenshots

📊 Dashboard — Project overview, quick navigation to chat, projects, or new experiments
LabOS Dashboard
📁 Projects — Multi-project management with independent experiments, memory, and paper library per project
LabOS Projects
🧪 Experiment Pipeline — 4-stage pipeline status, stage approvals, real-time logs
LabOS Experiment Pipeline

Features

Feature Description
🔬 4-Stage Research Pipeline Ideation → Design → Execution → Paper, each stage produces independent reports
💬 Chat-Driven AI assistant with streaming; create projects directly from conversations
🤖 Multi-LLM Profiles Configure different models and APIs for chat, code analysis, paper writing, experiment design
🖥️ Remote Execution SSH to GPU servers (AutoDL, etc.) with real-time log streaming
⚡ Codex CLI Integration Full-auto mode, JSONL streaming, AI writes experiment code
🧠 Memory System Cross-experiment knowledge persistence, project-level memory retrieval
✅ Stage Approval Approve → next stage / Revise & rerun / Reject & terminate
⚙️ Fully Configurable All settings exposed via Web UI; works with any OpenAI-compatible API
💾 Local-First SQLite storage, all data stays on your machine

Quick Start

git clone https://github.com/YUANXICHE98/LabOS.git
cd LabOS
bash start.sh

Or manually:

pip install -r requirements.txt
cd src && python api_server.py

Open your browser at http://localhost:8000

First-Time Setup

  1. Go to Settings → Configure your LLM API endpoint and key (any OpenAI-compatible API)
  2. (Optional) Configure SSH server for remote experiment execution
  3. Go to Chat → Start chatting → Create a project from the conversation
  4. Or go to Projects → Create a project manually → Launch an experiment

Multi-LLM Configuration

LabOS supports independent LLM configurations per task type:

Task Type Use Case Recommended Models
General Chat Daily research discussions DeepSeek-Chat / GPT-4o
Code Analysis Code review, experiment code generation DeepSeek-Coder / Claude
Paper Analysis Literature review, paper writing GPT-4o / Claude
Experiment Design Hypothesis generation, experiment planning DeepSeek-Chat / Claude

Configure via Settings > LLM Profiles — each task type gets its own Base URL + API Key + Model.

Project Structure

LabOS/
├── src/
│   ├── api_server.py      # FastAPI backend — all API endpoints and pipeline logic
│   ├── index.html          # Main page (SPA)
│   ├── app.js              # Frontend — UI logic, API calls, SSE streaming
│   └── style.css           # Styles
├── docs/
│   ├── screenshots/        # Screenshots
│   └── videos/             # Demo videos
├── start.sh                # One-click launcher
├── requirements.txt        # Python dependencies
├── CONTRIBUTING.md         # Contribution guide
├── GOVERNANCE.md           # Contributor governance & incentives
└── LICENSE                 # AGPL-3.0

Tech Stack

Layer Technology
Backend Python / FastAPI / uvicorn
Database SQLite (zero-config, local file)
Frontend Vanilla HTML + CSS + JavaScript (no build step)
Remote Execution Paramiko (SSH)
LLM Calls httpx (OpenAI-compatible protocol)
Real-time Server-Sent Events (SSE)

Open Core Model

LabOS follows an open core + paid add-ons model:

🆓 Free & Open Source (this repo)

All code in this repo is permanently free and open source: full research pipeline, chat, project management, LLM config, memory system, experiment execution.

💎 Paid Add-ons (coming soon)

  • Skill Library — Verified research methodologies, experiment paths, and best practices. Think of it as a knowledge base of "what actually works"
  • Premium Integrations — Pre-built connectors for more cloud GPU platforms and HPC clusters
  • Priority Support — Direct access to the dev team

The platform itself will always be open source. The real value is in verified research methodologies — battle-tested paths that save weeks of trial and error.

Contributor Incentives

We value every contribution. See GOVERNANCE.md for details.

Tier Requirement Incentive
🌱 Contributor 1 merged PR Contributors Wall recognition
🌿 Active Contributor 3+ PRs Free Skill Library access + Beta early access
🌳 Core Contributor 10+ PRs or 1 major feature 30% revenue share + paper co-authorship
💰 Bounty Tasks 💰 bounty label Crypto / Sponsors cash rewards

Code isn't the only way to contribute — docs, translations, bug reports, research methodologies, and design all count equally.

PRs welcome! See CONTRIBUTING.md, browse Issues to find tasks that interest you.

Support the Project

⭐ Star this repo — Help more people discover LabOS  │  🍴 Fork & Contribute  │  💰 Sponsor — Fund continued development

Cryptocurrency

Chain Address
ETH / ERC-20 (USDT, USDC, etc.) 0xc6B4720835E6C3CB58618B4df26B64F595C30202

Alipay

Alipay QR Code

GitHub Sponsors

Click the "Sponsor" button at the top of this repository.

Contributors

AGPL-3.0 — Modifications must be open-sourced. Network services must provide source code. Derivatives must reference the upstream repo.

Made with ❤️ for the research community