edict
🏛️ 三省六部制 · OpenClaw Multi-Agent Orchestration System — 9 specialized AI agents with real-time dashboard, model config, and full audit trails
What is edict?
🏛️ 三省六部制 · OpenClaw Multi-Agent Orchestration System — 9 specialized AI agents with real-time dashboard, model config, and full audit trails
Overview of edict
What it is
Edict is an open-source multi-agent orchestration system inspired by the ancient Chinese "Three Departments and Six Ministries" (三省六部) governance structure. It coordinates 12 AI agents (11 business roles + 1 compatibility role) through a strict division of power and checks-and-balances workflow. The system includes a real-time dashboard (Junji Chu Kanban), model configuration, and full audit trails. It is built on OpenClaw and released under the MIT license.
Key features
- 12-agent architecture with role-based permissions: Crown Prince (message triage), Three Departments (planning, review, dispatch), and Seven Ministries (execution). Each agent has independent workspace, skills, and model settings.
- Real-time Kanban dashboard with 10 panels: task board, department monitor, memorial archive, template library, official statistics, news aggregation, model config, skills config, session monitor, and ceremony animation.
- Institutional review mechanism: the Menxia (门下) Department reviews all plans before execution and can reject (封驳) substandard proposals, forcing rework.
- Full audit trail: every task (旨意) is archived as a memorial with a five-stage timeline (imperial decree → Zhongshu → Menxia → Six Ministries → reply).
- Live intervention: tasks can be paused, cancelled, or resumed from the dashboard.
- Hot-swappable LLM models: each agent can be assigned a different model via the dashboard, with automatic Gateway restart (~5 seconds).
- Agent health monitoring: heartbeat and activity detection displayed on the dashboard.
- News aggregation and push: daily tech/finance news with category subscription and Feishu push.
Use cases
- Complex multi-step tasks that require quality assurance at each stage (e.g., content generation, research reports, project planning).
- Teams that need full observability and auditability of AI agent workflows.
- Scenarios where different AI models are needed for different roles (e.g., planning vs. execution).
- Prototyping and demonstrating multi-agent collaboration with a visual dashboard.
Getting started
Docker one-click start:
docker run -p 7891:7891 cft0808/sansheng-demo
Open http://localhost:7891 to see the dashboard with preloaded demo data.
Full installation (requires OpenClaw, Python 3.10+, macOS/Linux):
git clone https://github.com/cft0808/edict.git
cd edict
chmod +x install.sh && ./install.sh
Then start:
chmod +x start.sh && ./start.sh
Or start components separately:
bash scripts/run_loop.sh &
python3 dashboard/server.py
Open http://127.0.0.1:7891.
License and pricing
License: MIT. The project is open source (pricing type: open_source). No pricing information is stated beyond the license.
Based on the project's public README and repository metadata.