ArchonDev

AI Development Governance — Stop Babysitting Your AI Agent

Two Ways to Use ArchonDev

Option 1: Full CLI (Recommended)

The complete AI development system. It manages your entire development process so you can focus on the big picture — not constantly correcting your AI.

# macOS/Linux:
npm install -g archon && archon

# Windows PowerShell:
npm install -g archon; archon

What you get:

• AI that reads and respects your architecture before writing code
• Adversarial planning (Architect proposes, Sentinel critiques)
• Automatic quality gates before changes are applied
• Learning persistence — mistakes are remembered and avoided
• Dependency tracking — prevent regressions with "what-breaks-what" map
• Bug reporting with root cause analysis
• AI-powered code review for any codebase
• Multi-provider key support with adversarial features

Option 2: Lite Package

Copy governance files into any project. Works with your existing AI tools (Cursor, Claude Code, Windsurf, VS Code + Copilot).

Download Lite Packages →

What you get:

• ARCHITECTURE.md template with best practices
• IDE-specific rule files (.cursorrules, CLAUDE.md, etc.)
• Progress tracking templates
• DEPENDENCIES.md — Track file-level dependencies to prevent regressions
• First-Run Walkthrough — Guided onboarding when AI detects your governance files
• Code Review Mode — Structured code review without changing your code
• Local Database — Track atoms and learnings in SQLite (no CLI required)
• Memory Management — Context handoff protocol for long sessions
• Task Extraction Protocol — AI confirms all items before starting, nothing gets forgotten
• Pre-Deploy Accessibility — WCAG 2.2 AA check before going live, legal liability warnings
• Works with any AI coding assistant

Commands

| Command | Description | |---------|-------------| | archon | Interactive mode — just run and follow prompts | | archon init | Initialize in your project | | archon plan <description> | Create a work item with AI planning (extracts and confirms multi-item requests) | | archon execute <atom-id> | Execute with quality gates | | archon list | List all work items | | archon show <atom-id> | Show details | | archon watch | Live TUI dashboard with status and credits | | archon bug report <title> | Bug report with root cause analysis | | archon review init | Initialize AI-powered code review | | archon review analyze | Scan project and populate review tasks | | archon review run | Run AI review on pending tasks | | archon credits | View credit balance | | archon credits add | Purchase credits | | archon credits budget | Set monthly budget and alerts | | archon keys add <provider> | Add your own API key — BYOK (Bring Your Own Key) | | archon keys list | Show configured API keys by provider | | archon preferences | Interactive settings menu (billing, models, keys, usage) | | archon models | List available AI models | | archon deps list | View file dependency rules | | archon deps add | Add a new dependency rule | | archon deps check --files <list> | Check for downstream impacts | | archon deps graph | Generate Mermaid dependency diagram | | archon a11y check | Run WCAG 2.2 AA accessibility audit | | archon a11y fix | Auto-fix common accessibility issues | | archon a11y badge | Add accessibility compliance badge | | archon a11y pre-deploy | Interactive pre-deployment check |

Pricing

| Tier | Cost | What You Get | |------|------|--------------| | Free | $0 | Ultra-cheap models (GPT-5-nano, GPT-4.1-nano, Gemini 2.5 Flash-Lite) — $0 cost to us | | Credits | Pay as you go | All models, 10% service fee, deposit any amount | | BYOK (Bring Your Own Key) | $0 | Use your own API keys, unlimited usage | | Claude Subscription (coming soon) | Your existing sub | Use Claude Pro/Max subscription instead of API keys |

No subscriptions. No commitments. Start free.

Claude Pro/Max Subscription Support (Coming Soon)

If you have an existing Claude subscription, you'll soon be able to use it with ArchonDev instead of API keys:

• Claude Pro ($20/mo) — Use your Pro subscription
• Claude Max 5x ($100/mo) — Higher limits for power users
• Claude Max 20x ($200/mo) — Maximum throughput

This works similarly to how Claude Code allows subscription-based access. We're currently waiting on Anthropic to provide a public OAuth API for third-party applications. For now, use BYOK (bring your own API key) — we'll notify you when subscription support is available.

How It Works

1. Define Your Rules — Create ARCHITECTURE.md with boundaries and invariants
2. AI Reads Rules First — Every session starts by understanding your architecture
3. Changes Are Validated — Quality gates check code before it's applied
4. Learnings Persist — Insights saved for future sessions

Working with Existing Projects

Have a project created by another AI agent? ArchonDev can review it first, then govern future changes.

# Step 1: Review existing code
cd your-existing-project
archon review init        # Create local review database
archon review analyze     # Scan project structure
archon review run --all   # AI reviews all features
archon review export > review-report.md

# Step 2: Set up governance
archon init --analyze     # Creates ARCHITECTURE.md

# Step 3: Fix issues with governed workflow
archon plan "fix critical issues from review"
archon execute <atom-id>

The CLI detects existing projects and suggests this workflow automatically.

Documentation

• archondev.io — Main website with animated AI problems story
• AI Coding Problems Research — Market research on AI coding assistant issues

ArchonDev by Jumping Ahead Corp.