bmad-code-org/BMAD-METHOD

What it solves

Coding assistants are good at implementation but often turn unstated assumptions into code. BMad Method keeps you in control by making important decisions explicit and preserving them as context for later work. It also solves the problem of process bloat: small changes go directly to implementation, while larger initiatives get deeper planning, so the method fits both weekend prototypes and long-lived systems.

How it works

BMad ships as a set of skills and workflows you install into an AI coding tool that supports skills (like Claude Code, Codex, or others). You start by running bmad setup, then invoke bmad-build with a change request. The method's agents and structured workflows guide you through planning, implementation, review, and learning, with specialized perspectives from product, architecture, UX, development, and testing available when needed. Context is carried forward in briefs, specifications, and architecture documents, so you don't re-explain decisions in every chat. The stable v6.12 line installs via npm (npx bmad-method install), while the V7 preview alpha uses the Skills CLI or plugin marketplaces.

Who it's for

Solo developers and small teams who use AI coding assistants and want a structured way to deliver software without losing control. It's also useful for teams starting from nothing or adding structure to an existing codebase, and for anyone who wants to move a change from idea to reviewed, working code with explicit planning that scales to the complexity of the work.

Highlights

• Right-sized process: go straight to implementation for clear changes, or add deeper planning for larger initiatives.
• Durable context: product and technical decisions carry forward instead of being re-explained in every chat.
• Specialized perspectives: brings in product, architecture, UX, dev, and testing expertise when it helps.
• Works with new or existing codebases, from prototyped ideas to inherited systems.
• Free and open source (MIT), with no paywalled workflows.