Skip to content
Get packs

Edict

Multi-agent AI system modeled on China's Tang-era governance, with a live task dashboard, requires OpenClaw.

GitHub Download source
Stars
17k
Forks
1.8k
Licence
MIT
Language
Python
Last update
1 months ago
Type
AI model / agent

What it is

Edict is a set of about 12 AI agents arranged in roles borrowed from an ancient Chinese imperial administration: one sorts incoming requests, others plan, review, and dispatch work, and six execution departments carry it out. A web dashboard shows tasks on a kanban board, agent health, token usage, and an audit trail of each task's path. It is built to run on top of OpenClaw, which the README lists as a requirement.

  • Run the Docker demo to view the kanban dashboard with preloaded sample data
  • Submit a task through a connected chat channel and follow it through planning, review, and execution in the dashboard
  • Compare model settings for each agent from the Models panel
  • Browse and add agent Skills from the Skills panel

Great for

  • Developers already running OpenClaw who want a multi-agent workflow with visible task history
  • People who want a review step before agent output is executed, as the README describes
  • Teams who want to watch and pause agent tasks from a dashboard
  • Users who want to switch the LLM model used by each agent from a UI

Not ideal for

  • People without OpenClaw installed, since the README lists it as required and the dashboard depends on it for live data
  • Non-technical users, because setup involves shell scripts, API keys, and an agent configuration
  • Anyone needing Windows support, since the README lists macOS and Linux as the prerequisites
Step-by-step guide: how to use EdictInstall, first run, examples, tips and FAQs
Edict alternativesEdict vs PonytailEdict vs NanoClawEdict vs SuperAGICrewAI vs EdictEdict vs MetaGPT

How to get it

  1. 1

    Clone and build from source

    You'll need: OpenClaw installed · Python 3.10+ · macOS or Linux · Node.js 18+ (optional, for building the React frontend)

    Clone the repository

    git clone https://github.com/cft0808/edict.git

    Enter the project folder

    cd edict

    Make the installer executable and run it

    chmod +x install.sh && ./install.sh
  2. 2

    Run it

    • Run the Docker demo with: docker run -p 7891:7891 cft0808/sansheng-demo
    • Open http://localhost:7891 in a browser
    • For a full setup, first add an API key with: openclaw agents add taizi, then rerun ./install.sh
    • Start with: chmod +x start.sh && ./start.sh, then open http://127.0.0.1:7891
  3. 3

    Get the most out of it

    • Use the Monitor and Task Detail panels to follow each task's full path and spot stalled work
    • Set different LLM models per agent in the Models panel, which the README says takes about 5 seconds to apply
    • Enable Feishu push for news digests if you use Feishu, as described in the News panel
    • Use the Template Library to start tasks from preset templates with parameter forms

Downloads

Source code (.zip)main branch · straight from GitHubOfficial website & docsopenclaw.ai

Strengths

  • Dashboard shows task status, heartbeats, and full task history in one place
  • README describes a review stage that can send work back for rework before execution
  • Configuration is set up by an install script and can run as a systemd service

Trade-offs

  • Depends on OpenClaw, so it is not useful on its own without that separate system
  • README is written mainly in Chinese, and the English README is a separate file the excerpt did not include
  • No formal release is listed, and some claims in the README (such as comparisons with other frameworks) are the author's own

How it compares

ProjectStarsTypeLevelUpdatedBest for
edict (this)17kAI model / agentadvanced1 months agoDevelopers already running OpenClaw who want a multi-agent workflow with visible task history
crewAI60.4kLibraryintermediate4 days agoPython developers prototyping multi-agent workflows
MetaGPT71.1kFrameworkintermediate10 months agoDevelopers experimenting with multi-agent LLM workflows
paperclip110kAppadvanced4 days agoDevelopers already running several AI coding agents who want one place to track them
CowAgent47.8kAppintermediate4 days agoPeople who want a self-hosted personal assistant reachable from their everyday chat apps
agents42.1kDev toolbeginner4 days agoDevelopers who already use one of the listed AI coding tools and want ready-made specialist agents

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.