Danghuangshang
wanikua/danghuangshang
Multi-agent team modeled on Ming dynasty government, built on OpenClaw with a Hermes Agent runtime option
Danghuangshang is a multi-agent collaboration system that models a Ming dynasty cabinet, with each AI agent playing a government role such as a ministry or the censorate. It is built on OpenClaw, and its README also lists an install method that uses the Hermes Agent runtime.
What Danghuangshang does
You act as the emperor and the agents act as ministers. In Discord or Feishu you mention an agent and it carries out the task. In the default Ming cabinet mode, a Directorate of Ceremonial agent takes the instruction, the Grand Secretariat refines the prompt into an execution plan, ministries carry it out, and a censorate agent reviews code pushed to GitHub. In Discord multi-bot mode each department is its own bot and can be addressed directly.
The installer offers three governance styles: the Ming cabinet with 18 agents, the Tang three-department system with 14 agents, and a modern company layout with 14 agents. The README lists collaboration, memory, 60+ skills, cron and sandboxing as core capabilities, and warns against setting allowBots to true on Discord because bots can trigger each other in loops. A Hermes Agent runtime install is listed as a method parallel to OpenClaw. The README is written mainly in Chinese, with English and Japanese versions linked.
Key features
- Three governance modes: Ming cabinet (18 agents), Tang three departments (14) and modern company (14)
- Mention an agent in Discord or Feishu to assign work
- A prompt-refinement step turns a request into an execution plan before dispatch
- Automatic review of code pushed to GitHub by a censorate agent
- Local install through git clone and scripts/full-install.sh
- Discord safety guidance, with allowBots set to mentions by default in the installer
When to use it
- Run a role-based AI team that takes tasks from a Discord server
- Have a dedicated reviewer agent check code pushed to GitHub
- Set up a company-style agent group with CEO and CTO roles for a small startup
Who it is for: Users who want a ready-made role-based multi-agent team on OpenClaw, or on the Hermes Agent runtime.
How it fits with Hermes Agent
Built mainly for OpenClaw, with a documented install method that uses the Hermes Agent runtime as a parallel option.
How to install Danghuangshang
These commands are copied from the project's README. Check the repository for the latest steps before you run them.
git clone https://github.com/wanikua/danghuangshang.git
cd danghuangshang
bash scripts/full-install.shRequirements: OpenClaw; the README advises beginners to use a cloud server rather than a personal computer, and chat access through Discord or Feishu
FAQ
What is Danghuangshang?
Danghuangshang is an open-source multi-agent system that arranges AI agents as the officials of a Ming dynasty cabinet. You give orders in Discord or Feishu and the agents divide the work.
Does Danghuangshang work with Hermes Agent?
Partly. It is built on OpenClaw, but the README lists an install method that uses the Hermes Agent runtime in parallel to OpenClaw. Check the repository README for the exact steps.
How do I install Danghuangshang?
Clone the repository, change into the folder and run bash scripts/full-install.sh. The README recommends this local install and suggests beginners use a cloud server.
Similar multi-agent for Hermes Agent
All multi-agentPre-alpha host agent that builds task DAGs and orchestrates built-in and third-party agents
HarnessRouter HarnessRouterSelf-hosted unified API that runs Codex, Claude Code, Hermes and other agent harnesses
Orkas-AI OrkasLocal-first desktop app where a Commander agent directs specialists and CLI agents such as Hermes
a5c-ai BabysitterWorkflow orchestrator for coding agents with quality gates, human approvals and a decision journal
traycerai TraycerOpen-source desktop app for running Hermes Agent, Claude Code, Codex and other agents in parallel
win4r ClawTeam-OpenClawMulti-agent swarm coordination for coding agents, with OpenClaw as default and Hermes Agent supported
Related guides: Build agent teams with the Hermes kanban board · Run multiple Hermes agents with profiles · Connect Hermes agents on several machines with Hermes Desktop