archon-cli

Run, manage, and author Archon AI coding workflows through the archon CLI.

23.3k|3.5k|Updated Feb 7, 2025
One-click install
npx skills add https://github.com/coleam00/Archon --skill archon-cli
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: archon-cli
Source: https://github.com/coleam00/Archon/tree/main/.claude/skills/archon-cli
Command: npx skills add https://github.com/coleam00/Archon --skill archon-cli

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

AI coding agents behave inconsistently when asked to fix bugs or build features, skipping planning, tests, or review steps. This Skill lets you drive Archon, a workflow engine that encodes development processes as YAML workflows with deterministic phases, validation gates, and isolated git worktrees, so every run follows the same repeatable structure.

Core Features & Use Cases

  • Run and monitor workflows: Discover available workflows with archon workflow list, launch detached runs on isolated branches, and use archon workflow wait to be notified exactly when a run finishes or needs a decision.
  • Manage run lifecycle: Inspect, approve, reject, respond to, cancel, abandon, or resume runs, including resolving interactive approval gates and judging finished runs by their artifacts rather than exit status.
  • Author and configure workflows: Design new YAML workflows with eleven node types (AI prompts, bash/script gates, loops, approvals, fan-out composition), write effective node prompts, and set up providers, model tiers, and config layers.
  • Use Case: Ask the agent to "fix issue #42 with archon" and it will validate the issue as a complete brief, launch the right workflow on a fix/issue-42 branch in an isolated worktree, monitor the run, and report the outcome with its review artifacts.

Quick Start

Ask the agent to run an Archon workflow on a specific task, for example: use archon to fix issue #42 on a new branch and let me know when the run needs my approval.

Frequently Asked Questions about archon-cli

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run an Archon workflow on a GitHub issue?

Run `archon workflow run <workflow> --branch fix/issue-N "<message>" --detach` from the repo root. First check the issue body states the problem, outcome, invariants, and acceptance criteria, since the input is the contract the whole run is measured against.

How do I approve or reject a paused Archon workflow run?

Use `archon workflow approve <run-id> "comment"` or `archon workflow reject <run-id> "<reason>"`. With `--json` the decision is only recorded and you must call `archon workflow resume <run-id>` separately; without it, approval records and auto-resumes.

What is the difference between archon workflow cancel and abandon?

`cancel` actively stops a live CLI-detached run by verifying the process tree is gone before marking it cancelled. `abandon` is state-only: it marks a paused or orphaned run cancelled in the database without killing anything, for crashed hosts or already-stopped runs.

Why does my Archon workflow run fail in the worktree but work locally?

Git worktrees carry only tracked files, so gitignored files like `.env` are missing. Add a `worktree.copyFiles` entry listing `.env` in the repo-level `.archon/config.yaml` so the file is copied into each new worktree.

Can I override the AI models for a single Archon workflow run?

Yes. Pass `--model <tier|@alias>=<provider>/<model>` at launch, for example `--model large=pi/minimax/minimax-m3`. This rebinds tiers for that run only without touching persistent config; verify with `--dry-run` first.

When should I not use an approval gate in an Archon workflow?

Reserve human approval gates for interactive sessions or irreversible, public-facing, expensive actions like deploys or deletions. Everything else should be decided by deterministic bash/script checks or an independent AI verifier returning a structured boolean.