BMAD Method 🤖 Agent Free

What it does

BMad Method (Breakthrough Method for Agile AI-Driven Development) is an open-source workflow for turning an idea or change request into working software with an AI coding agent while keeping the important decisions explicit. Its central idea is right-sized planning: a clear, small change goes straight to a build step, while a large or fuzzy initiative gets research, a product brief or PRD, UX and architecture work, and a breakdown into epics and stories first. It works on new and existing codebases.

What is inside / how it works

Recent versions are delivered as Agent Skills. The core module adds general skills such as bmad-help, which inspects your project and suggests the next step, plus brainstorming, structured review and a multi-agent discussion mode. The BMad Method module adds planning and delivery skills like bmad-spec and bmad-build, and five named agent personas: an analyst, a product manager, an architect, a developer (who also handles QA test generation, code review and sprint planning) and a UX designer. Each agent has a menu of short codes that start its workflows. Optional official modules add a skill builder, a test architect, an unattended epic loop, creative-thinking partners and a game development studio.

Works with

The documentation names Claude Code, Cursor, Windsurf, Codex, Auggie, Amp, Cline, IBM Bob and Antigravity, among others; the installer prints the current list with npx bmad-method install --list-tools.

How to install or start

Run the installer in your project (Node.js 20.12 or later; uv is needed for some skills):

npx bmad-method install

Alternatives are npx skills add bmad-code-org/BMAD-METHOD, or the Claude Code and Codex plugin marketplace at bmad-code-org/bmad-plugins. Then ask bmad-help what to do next.

Pricing

Free and open source under the MIT licence, with no paywalled workflows. BMad and BMAD-METHOD are trademarks of BMad Code, LLC.

Who should use it

Solo developers and teams who want product and architecture thinking written down before an agent writes code, especially for multi-epic projects. For tiny fixes the process can feel heavy, although the build-first path keeps it short. The project changes quickly, so older tutorials may describe personas and commands that no longer exist.