Skip to main content

Make AI Coding AgentsFollow Your Process

AI DevKit installs the workflow layer your agent is missing: requirements, design, planning, implementation, tests, verification, memory, and review across the coding tools you already use.

Get StartedView on GitHub

Why AI DevKit?

Plans Before Code

Start with requirements, design decisions, and task plans before your agent edits files.

Verification Gates

Require fresh test or build evidence before the agent can claim implementation work is complete.

Durable Context

Keep decisions, conventions, and implementation notes available through docs and local memory.

One Setup, Many Agents

Configure Claude Code, Cursor, Codex, Gemini CLI, and other agents from one project config.

Supported Coding Agents

One config keeps the same workflow, memory, and skills across the coding agents your team already uses.

Quick Start

Install AI DevKit globally:

npm install -g ai-devkit

Initialize in your project:

ai-devkit init

or

npx ai-devkit@latest init

This creates a docs/ai/ workflow, installs agent commands, and wires skills such as memory and verification where supported.

In Cursor or Claude Code, type /new-requirement to get started.

Read the full documentation →

Ready to make your agent less reckless?

Add planning, memory, verification, and review to the AI coding tools you already use.

Get Started Now