Repolode
Submit
coleam00

coleam00/Archon

Open-source workflow engine that runs YAML-defined AI coding and other processes with agents, scripts and human approval gates.

View on GitHubWebsite
Stars
23,729
Forks
3,611
Open issues
180
Last commit
6 hours ago

What is it?

Archon lets you describe how work gets done as YAML workflows. Each workflow is a graph of nodes: AI agent prompts handle judgment, bash commands and scripts handle deterministic checks, loops repeat steps until a test command passes, and wait nodes pause the run so a person can approve or reject. The goal is to make AI-assisted coding follow the same plan, checks and gates every time instead of depending on how a single agent run goes.

The bundled sdlc pack implements a software delivery pipeline: it takes an issue through planning, implementation, review, CI waiting and merging, with queues that pick up ready work and attention items that surface only when a decision is needed. The engine itself is generic, so the same approach can be used for content production, research or operations. Archon runs Claude Code, Codex and Pi through their own CLIs and SDKs, stores run logs and cost data in a local SQLite database by default (PostgreSQL optional), and offers a CLI, a web console and chat integrations for Slack, Telegram, GitHub and Discord.

Extension happens through plugins for workflow packs, agent providers, code forges and chat surfaces. Plugins are installed from GitHub repositories with the archon plugin add command. The project is written in TypeScript and runs on Bun.

Key features

  • 01Define multi-step workflows in YAML with agent, bash, script, loop, wait and child-workflow nodes.
  • 02Loops repeat a step until a shell command such as your test suite succeeds.
  • 03Pauses runs for human approval or rejection through an attention queue.
  • 04Runs each repository task in its own git worktree so several can proceed in parallel.
  • 05Supports Claude Code, Codex and Pi as agent providers.
  • 06Ships with an sdlc pack that takes GitHub issues to merged pull requests.

Use cases

  • Automating issue-to-pull-request delivery in a repository while keeping review and approval gates.
  • Standardizing how AI coding agents plan, implement and test changes across a team.
  • Building repeatable non-code processes such as content production or research pipelines with agent steps.
  • Self-hosting an audit trail of agent runs, outcomes and costs for a team.

Quick start

  1. 1

    Install Claude Code and sign in, or install Codex

  2. 2

    Install Archon with Homebrew

    brew install coleam00/archon/archon
  3. 3

    List the available workflows in your git repository

    archon workflow list
  4. 4

    Run a bundled investigation workflow with a question

    archon workflow run archon-investigate "Why does the login test fail?"

Things to know

  • Requires an agent provider such as Claude Code or Codex, using your own accounts.
  • Sends anonymous usage telemetry events by default.

Topics

#ai-agents#workflow-engine#yaml#claude-code#automation#cli#typescript#developer-tools#ai#bun

FAQ

What is Archon?
Open-source workflow engine that runs YAML-defined AI coding and other processes with agents, scripts and human approval gates. Archon lets you describe how work gets done as YAML workflows. Each workflow is a graph of nodes: AI agent prompts handle judgment, bash commands and scripts handle deterministic checks, loops repeat steps until a test command passes, and wait nodes pause the run so a person can approve or reject. The goal is to make AI-assisted coding follow the same plan, checks and gates every time instead of depending on how a single agent run goes.
Is Archon free to use? What license does it have?
Archon is open source under the MIT license. Check the license terms before using it in commercial products.
Is Archon still maintained?
Actively maintained. The last commit to Archon was on Oct 20, 2026.
What can I use Archon for?
Automating issue-to-pull-request delivery in a repository while keeping review and approval gates. Standardizing how AI coding agents plan, implement and test changes across a team. Building repeatable non-code processes such as content production or research pipelines with agent steps. Self-hosting an audit trail of agent runs, outcomes and costs for a team.
How do I install Archon?
Install Claude Code and sign in, or install Codex Install Archon with Homebrew: brew install coleam00/archon/archon List the available workflows in your git repository: archon workflow list Run a bundled investigation workflow with a question: archon workflow run archon-investigate "Why does the login test fail?"
Where can I get Archon?
The source code and installation instructions are on GitHub: https://github.com/coleam00/Archon

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects