AI DevKit CLI

The command-line interface for AI DevKit — make AI coding agents follow a repeatable engineering workflow in your project.

Use this package when you want the ai-devkit command to install agent skills, memory, verification gates, MCP configuration, and docs/ai/ workflow files into a project.

Features

• Workflow layer — Requirements, design, planning, implementation, testing, verification, and review
• AI environment setup — One-command configuration for Cursor, Claude Code, Codex, Gemini CLI, Devin, and other agents
• Skill management — Install reusable AI skills that change agent behavior
• Persistent memory — Store project decisions and conventions so agents can reuse them across sessions

Installation

# Run directly (no install needed)
npx ai-devkit@latest init

# Or install globally
npm install -g ai-devkit

Quick Start

# Set up your project interactively
ai-devkit init

# Set up from template (no step-by-step prompts when template is complete)
ai-devkit init --template ./ai-devkit.init.yaml

This will:

1. Create a .ai-devkit.json configuration file
2. Set up your AI coding agent environment, including skills and MCP servers where supported
3. Generate docs/ai/ workflow docs for requirements, design, planning, implementation, and testing
4. Give your agent a repeatable process instead of relying on one-off chat instructions

After initialization, your repo gets project-local files you can review and commit:

your-project/
├── .ai-devkit.json
├── .claude/        # or .cursor/, .codex/, etc. based on selected agents
│   ├── skills/
│   └── settings.json
└── docs/ai/
    ├── requirements/
    ├── design/
    ├── planning/
    ├── implementation/
    └── testing/

In your AI editor, ask the agent to use the dev-lifecycle skill to clarify the feature before editing code.

Common Commands

# Initialize project
ai-devkit init

# Initialize project from YAML/JSON template
ai-devkit init --template ./ai-devkit.init.yaml

# Install/reconcile project setup from .ai-devkit.json
ai-devkit install

# Overwrite all existing install artifacts without extra prompts
ai-devkit install --overwrite

# Add a development phase
ai-devkit phase requirements

# Validate workspace docs readiness
ai-devkit lint

# Validate a feature's docs and git branch/worktree conventions
ai-devkit lint --feature lint-command

# Emit machine-readable output for CI
ai-devkit lint --feature lint-command --json

# Probe capacity for every supported provider (codex, z.ai, OpenAI, Claude)
ai-devkit capacity

# Probe a single provider's capacity
ai-devkit capacity zai

# Claude uses CLAUDE_CODE_OAUTH_TOKEN, the active profile file, or the
# default-profile macOS Claude Code Keychain item
ai-devkit capacity claude

# Emit the JSON report (single object for one provider, array for many)
ai-devkit capacity codex --json

# Install a skill
ai-devkit skill add <skill-registry> [skill-name]

# List skills installed across known global environment paths
ai-devkit skill list --global

# Limit global listing to selected environments
ai-devkit skill list --global --env claude codex

# Store project knowledge for future agent sessions
ai-devkit memory store

On macOS, the operating system may request Keychain access the first time the Claude fallback is used. Custom CLAUDE_CONFIG_DIR profiles use only their own credentials file and never the global Keychain item.

UI Formatting Standards

Shared CLI presentation helpers live in src/util/. Commands should use status.ts for status labels and chalk colors, time-format.ts for relative or Mon D · HH:mm time labels with an injectable now clock, and pluralize.ts for count labels. Keep JSON output independent from these terminal-only helpers so machine-readable shapes stay stable.

Template example:

version: 1
environments:
  - codex
  - claude
phases:
  - requirements
  - design
  - planning
  - implementation
  - testing
skills:
  - registry: codeaholicguy/ai-devkit
    skill: dev-lifecycle
  - registry: codeaholicguy/ai-devkit
    skill: verify
  - registry: codeaholicguy/ai-devkit
    skill: memory
  - registry: codeaholicguy/ai-devkit
    skill: tdd

Documentation

📖 For the full user guide, workflow examples, and best practices, visit:

ai-devkit.com/docs

License

MIT