Open Source

Archon: The Open-Source Harness Builder Making AI Coding Deterministic and Repeatable

There’s a fundamental problem with AI coding agents: when you ask one to “fix this bug,” what happens depends on the model’s mood. It might skip planning. It might forget to run tests. It might write a pull request description that ignores your template. Every run is different.

Archon, a new open-source project trending on GitHub with over 16,000 stars, aims to fix this. It’s being described as “what Dockerfiles did for infrastructure and GitHub Actions did for CI/CD??xcept for AI coding workflows.”

The Vision: Make AI Coding Owned by You

Archon is a workflow engine for AI coding agents. The concept is elegantly simple: define your development processes as YAML workflows??lanning, implementation, validation, code review, PR creation??nd run them reliably across all your projects.

The workflow defines the phases, validation gates, and artifacts. The AI fills in the intelligence at each step, but the structure is deterministic and owned by you.

Key Features That Set Archon Apart

  • Repeatable: Same workflow, same sequence, every time. Plan, implement, validate, review, PR??ach run produces consistent results.
  • Isolated: Every workflow run gets its own git worktree. You can run five fixes in parallel with no conflicts.
  • Fire and Forget: Kick off a workflow, go do other work. Come back to a finished PR with review comments.
  • Composable: Mix deterministic nodes (bash scripts, tests, git ops) with AI nodes (planning, code generation, review).
  • Portable: Define workflows once in .archon/workflows/, commit them to your repo. They work the same from CLI, Web UI, Slack, Telegram, or GitHub.

Built-In Workflows for Common Tasks

Archon ships with seventeen default workflows covering common development scenarios:

  • archon-assist: General Q&A, debugging, exploration with full Claude Code agent capabilities
  • archon-fix-github-issue: Classify issue ??investigate/plan ??implement ??validate ??PR ??smart review ??self-fix
  • archon-idea-to-pr: Feature idea ??plan ??implement ??validate ??PR ??5 parallel reviews ??self-fix
  • archon-smart-pr-review: Classify PR complexity ??run targeted review agents ??synthesize findings

Or define your own. Default workflows are great starting points??opy one and customize it. Workflows are YAML files in .archon/workflows/, and commands are markdown files in .archon/commands/.

How It Works in Practice

Here’s a typical interaction:

You: Use archon to add dark mode to the settings page

Agent: I’ll run the archon-idea-to-pr workflow for this.
??Creating isolated worktree on branch archon/task-dark-mode…
??Planning…
??Implementing (task 1/4)…
??Tests failing – iterating…
??Tests passing after 2 iterations
??PR ready: https://github.com/you/project/pull/47

The Web Dashboard

For teams preferring a visual interface, Archon includes a web dashboard offering Chat, Dashboard for monitoring workflows, Workflow Builder for visual DAG editing, and Workflow Execution for step-by-step progress viewing.

Why This Matters

The proliferation of AI coding agents has been a double-edged sword. They dramatically increase productivity, but they also introduce unpredictability. Archon addresses this by externalizing process into version-controlled files. Your development methodology becomes code, not convention.

The project is MIT-licensed and contributions are welcome. Full documentation is available at archon.diy, including a ten-chapter tutorial called “The Book of Archon.”

Join the discussion

Your email address will not be published. Required fields are marked *