#4545th of 54 in Agent and MCP starters

danghuangshang

Multi-agent team on OpenClaw, modeled on Ming ministries, run from Discord

Stars
2.7k
License
MIT
Last commit
May 2026
Last release
Mar 2026
Language
TypeScript
Stars, last 29 weeks+229
21 Apr 2026: 2,488 stars28 Apr 2026: 2,509 stars5 May 2026: 2,520 stars12 May 2026: 2,539 stars19 May 2026: 2,551 stars26 May 2026: 2,575 stars2 Jun 2026: 2,592 stars9 Jun 2026: 2,602 stars16 Jun 2026: 2,611 stars23 Jun 2026: 2,621 stars30 Jun 2026: 2,625 stars7 Jul 2026: 2,629 stars14 Jul 2026: 2,642 stars21 Jul 2026: 2,645 stars28 Jul 2026: 2,649 stars4 Aug 2026: 2,652 stars11 Aug 2026: 2,660 stars18 Aug 2026: 2,665 stars25 Aug 2026: 2,671 stars1 Sept 2026: 2,674 stars8 Sept 2026: 2,675 stars15 Sept 2026: 2,680 stars22 Sept 2026: 2,684 stars29 Sept 2026: 2,689 stars6 Oct 2026: 2,693 stars13 Oct 2026: 2,696 stars20 Oct 2026: 2,702 stars27 Oct 2026: 2,709 stars3 Nov 2026: 2,711 stars10 Nov 2026: 2,717 stars

2,488 on 21 Apr 20262,717 on 10 Nov 2026

From GitHub's daily star counts.

Checked on GitHub.

Overview

Installer and config templates that set up 14 to 18 role-based agents (dispatcher, prompt refiner, six ministries, code reviewer) on the OpenClaw framework. You @mention an agent in Discord or Feishu and tasks route through the hierarchy. Three org presets (Ming, Tang, modern CEO) can be switched by script. A React dashboard on port 18795 and a Hermes Agent runtime option are included.

Who it is for: Self-hosters running a team of role-based LLM agents through Discord or Feishu

Strengths

  • Three org presets switchable with scripts/switch-regime.sh, each agent with its own memory
  • Supports Anthropic, OpenAI, Gemini and DeepSeek; models can differ per agent
  • Memory backup script uses sqlite3 .backup with daily, weekly and monthly retention
  • Multi-arch Docker image (amd64, arm64) plus local, remote and lite install scripts

Weaknesses

  • Depends on the OpenClaw framework; not usable standalone
  • Discord multi-bot setup needs allowBots set to mentions or bots loop
  • Main README is Chinese; English and Japanese versions are separate files
  • Install relies on curl-piped shell scripts; README advises a cloud server

What it needs

  • TypeScript, Anthropic, OpenAI, Gemini, DeepSeek
  • Needs OpenClaw, Discord bot token or Feishu app, LLM API key, Docker (optional), Notion (optional)
  • Docker

Quick answers

Is danghuangshang open source?

Yes. Its code is on GitHub under the MIT license.

Can you run danghuangshang with Docker?

Yes. Its repository has a Docker setup: a Dockerfile, a Compose file or a published image.

Also in Agent and MCP starters

See all 54
Also in Agent and MCP starters
RankProjectScore
1deepseek-harnessPlugin-based agent harness from DeepSeek with a local Web UI278.8k stars+32,022 (+13%) in 30 daysMIT98 out of 100
2composioTypeScript and Python SDKs that give agents authenticated tools across 1000+ apps31k starsMIT92 out of 100
3conductorDurable workflow engine that orchestrates microservices and LLM agents from JSON32.5k stars+251 (+0.8%) in 30 daysApache-2.091 out of 100
#4mcp-useTypeScript framework for building MCP servers with React Views and an Inspector Live demo ↗ (opens in a new tab)10.9k stars+131 (+1.2%) in 30 daysMIT87 out of 100
#5a2uiProtocol and renderers for agents to stream native UI instead of text Live demo ↗ (opens in a new tab)17.1k stars+418 (+2.5%) in 30 daysApache-2.086 out of 100