Cole Medin · 2026-04-10 · notable
Archon — deterministic workflow engine for AI coding agents
Open-source workflow engine that lets you define AI coding workflows as YAML with deterministic phases, isolated git worktrees, and composable nodes mixing scripts with AI tasks. 16.6k stars.
Define your AI coding workflow as YAML, run it the same way every time, and stop hoping the model remembers what to do next.
Key specs
| License | MIT |
|---|---|
| GitHub stars | 16.6k |
| Pre built workflows | 17 |
What is it?
Archon is an open-source workflow engine for AI coding agents by Cole Medin. Instead of prompting an LLM and hoping it follows a consistent process, you write a YAML file that declares the phases of your development workflow -- planning, implementation, validation, code review, PR creation -- and Archon executes them in order. The AI fills in the intelligence at each step, but the structure is deterministic and owned by you.
How does it work?
Each workflow is a directed graph of composable nodes. Nodes can be deterministic scripts, AI-powered tasks, or validation gates. Archon runs each workflow in an isolated git worktree so parallel runs do not interfere. It ships 17 pre-built workflows for common development tasks and supports custom authoring. The engine runs from a CLI, a web UI, or via Slack, Telegram, Discord, and GitHub integrations.
Why does it matter?
The biggest complaint about AI coding agents is that they are non-deterministic: the same prompt produces different results on different runs. Archon addresses this by separating the workflow structure (deterministic, version-controlled) from the AI inference (non-deterministic but constrained to a single node). For teams shipping AI-assisted development to production, this is the missing reliability layer.
Who is it for?
Engineering teams using AI coding agents who need repeatable, auditable workflows.
Try it
curl -fsSL https://archon.diy/install | bash