Powered by RAG + multi-agent workflows, delivering high-quality automated long-document generation for architecture design, automotive repair, healthcare, bidding & tendering, and more.
OpenSpec is an enterprise-grade intelligent platform for professional long-document generation.
In specialized fields such as architecture design, automotive repair, healthcare, and bidding & tendering, professionals need to produce large volumes of structurally complex, technically rigorous, and standards-compliant long documents. These documents typically share the following characteristics:
- Lengthy — Often tens or even hundreds of pages, with multiple chapters and sub-chapters;
- Highly specialized — Content must reference industry standards and regulatory provisions, with no room for fabrication;
- Strictly formatted — Must follow specific document templates and layout standards;
- Repetitive — Documents of the same type share similar structures, but details vary by project.
OpenSpec was built to solve this pain point. Users simply input basic project information, and the system will:
- Intelligent Retrieval — Retrieve relevant standards, historical cases, and reference materials from the knowledge base;
- Chapter-by-Chapter Generation — A multi-agent workflow (Researcher + Auditor) generates standards-compliant professional content chapter by chapter;
- Human-AI Collaboration — Supports chapter-level review, rewriting, and supplementation for efficient human-AI collaboration;
- One-Click Export — Generate PDF, Markdown, AutoCAD title blocks, and other professional formats.
OpenSpec 2-Minute Video Introduction (Architecture Design Scenario): https://www.bilibili.com/video/BV1DoFUzBEmW/?share_source=copy_web&vd_source=d91cce476d06006159a799f4db6b9171
| Domain | Typical Documents |
|---|---|
| Architecture Design | Construction drawing design notes, feasibility study reports |
| Automotive Repair | Repair technical manuals, fault diagnosis reports |
| Healthcare | Clinical trial reports, diagnosis & treatment standard documents |
| Bidding & Tendering | Technical proposals, tender document preparation |
| More | Any structured long document that needs to be generated from a knowledge base |
 |
 |
 |
 |
Try OpenSpec online: https://archspec.aizzyun.com/
- Email:
test@qq.com - Password:
test
| Component | Version |
|---|---|
| Docker | >= 20.10 |
| Docker Compose | >= 2.0 |
# 1. Clone the repository
git clone https://github.com/zhuzhaoyun/OpenSpec.git
cd OpenSpec
# 2. Copy and edit environment variables
cp deploy/docker/.env.example deploy/docker/.env
# Edit .env and fill in required configurations (RAGFlow, LLM API Key, etc.)
# 3. Start all services
cd deploy/docker
docker compose up -dOnce started, visit http://localhost to use the platform.
| Variable | Description | Required |
|---|---|---|
RAGFLOW_API_KEY |
RAGFlow API key | Yes |
RAGFLOW_BASE_URL |
RAGFlow service URL | Yes |
DASHSCOPE_API_KEY |
LLM API key (default: Qwen) | Yes |
LANGFUSE_SECRET_KEY |
Langfuse secret key (Prompt management) | No |
LANGFUSE_PUBLIC_KEY |
Langfuse public key | No |
LANGFUSE_BASE_URL |
Langfuse service URL | No |
For the full list of environment variables, see deploy/docker/.env.example.
Expand to view local development guide
cd apps/web
npm install
npm run dev
# Visit http://localhost:5173cd apps/agent
pip install -r requirements.txt
cp ../../deploy/docker/.env.example .env
# Edit .env and fill in required configurations
uvicorn app:app --reload --port 5000 --host 0.0.0.0cd apps/backend
mvn spring-boot:run| Feature | Open Source | Enterprise |
|---|---|---|
| Intelligent Retrieval & Chapter Generation | ✅ | ✅ |
| Human-AI Collaborative Editing | ✅ | ✅ |
| PDF / Markdown Export | ✅ | ✅ |
| AutoCAD Plugin (File Import) | ❌ | ✅ |
| Training & Q&A Support | ❌ | ✅ |
| Data Processing Consulting | ❌ | ✅ |
For enterprise edition details or business cooperation, feel free to reach out:
-
Email:
dlutyaol@qq.com -
WeChat: Scan the QR code below
Copyright (c) 2024-2026 OpenSpec Contributors.
Licensed under The GNU General Public License v3.0 (GPLv3).