edict-multi-agent-orchestration

What This Skill Does

The Edict (三省六部) skill introduces a sophisticated multi-agent orchestration framework to the OpenClaw ecosystem, modeled after the historic Tang Dynasty government. It moves beyond standard linear agent workflows by implementing a rigorous checks-and-balances architecture. At its core, the system utilizes 12 specialized agents, including the Taizi (triage), Zhongshu (planning), and the critical Menxia (veto/quality assurance).

Unlike modern flat multi-agent systems, Edict enforces a structural hierarchy where the Menxia layer acts as a gatekeeper, capable of rejecting tasks and forcing the Zhongshu planning layer back to the drawing board. This ensures that only high-quality, verified directives reach the Six Ministries for execution. The skill comes bundled with a real-time React-based kanban dashboard that visualizes the movement of tasks, status updates, and audit trails for every interaction, making it ideal for complex, multi-stage project management.

Installation

To install this skill, run the following command in your terminal: clawhub install openclaw/skills/skills/adisinghstudent/edict-multi-agent-orchestration. Once installed, navigate to the local edict directory. Ensure you have Node.js and Docker installed for full functionality. Run chmod +x install.sh && ./install.sh to initialize the 12 agent workspaces, symlink data directories, and set up the openclaw.json configuration. Post-installation, initialize your API keys using openclaw agents add and restart the gateway via openclaw gateway restart to finalize the system integration.

Use Cases

• Complex Project Planning: Breaking down massive technical initiatives into distinct planning, review, and execution phases.
• High-Stakes Content Production: Ensuring rigorous editorial oversight by having the 'Menxia' agent audit draft content before final dispatch.
• Enterprise Process Automation: Modeling rigid corporate or government workflows that require multiple sign-offs and auditable trails.
• Software Development Lifecycle: Using the hubu, libu, and bingbu agents to manage specific modules of code generation, testing, and deployment sequentially.

Example Prompts

1. "taizi, analyze the current market trends in generative AI and create a strategic roadmap for our new product launch."
2. "zhongshu, draft a comprehensive technical architecture proposal for a scalable microservices backend, ensuring it passes the menxia review."
3. "taizi, initiate a audit trail review of the last 24 hours of tasks completed by the Bingbu agent."

Tips & Limitations

• Tips: Utilize the dashboard at http://localhost:7891 to identify bottlenecks in the agent pipeline. If a task is stuck, check the audit logs for the Menxia veto reason.
• Limitations: The system is resource-intensive due to running 12 persistent agents. Ensure your host machine has sufficient RAM. The mandatory veto layer can increase latency for simple, straightforward tasks compared to single-agent workflows.