explaingit

bmad-code-org/bmad-method

📈 Trending47,554JavaScriptAudience · developerComplexity · 3/5ActiveSetup · moderate

TLDR

A structured workflow framework that guides AI-assisted software development from requirements through testing using specialized agent personas and agile practices.

Mindmap

mindmap
  root((repo))
    What it does
      Organizes AI coding
      Defines agent roles
      Structures workflows
    Key features
      Product manager agent
      Architect agent
      Developer agent
      UX agent
    How to use
      Install via npm
      Invoke agents per stage
      Scale by complexity
    Use cases
      Solo dev projects
      Non-trivial apps
      Professional process
    Tech stack
      Node.js
      Python
      AI IDEs

Things people build with this

USE CASE 1

Build a full-stack web app from scratch with AI assistance, following structured requirements and architecture phases instead of ad-hoc prompting.

USE CASE 2

Develop a game using the optional game development module to handle design, mechanics, and implementation in a systematic way.

USE CASE 3

Create a solo startup MVP with professional engineering practices, requirements gathering, system design, and testing, without hiring a team.

Tech stack

Node.jsPythonJavaScript

Getting it running

Difficulty · moderate Time to first run · 30min

Requires Node.js and Python runtime setup plus dependency installation across multiple language ecosystems.

License could not be detected automatically. Check the repository's LICENSE file before use.

In plain English

BMAD-METHOD is a framework for organizing AI-assisted software development using a structured, agile-inspired workflow. The problem it addresses is that while AI coding assistants like Claude Code or Cursor can write code, most developers use them in an ad-hoc way, asking one-off questions or generating individual files, without a systematic process for going from an idea all the way through requirements, architecture, implementation, and testing. BMAD provides that process. The framework works by giving you a set of specialized AI agent personas to invoke at each stage of a project: a product manager agent for requirements gathering, an architect agent for system design, a developer agent for implementation, a UX agent for interface design, and others totaling over a dozen specialized roles. These agents follow structured workflows adapted from agile software development practices (agile means iterative, short-cycle development rather than one big upfront plan). A help skill called bmad-help can be invoked at any point to get guidance on what the next step should be. The framework is installed via an npm command that sets up configuration files in your project directory, after which you interact with it through your AI IDE. It scales its planning depth based on project complexity, a bug fix gets a lightweight process while an enterprise system gets more formal phases. Optional modules extend it for game development, test architecture, and creative brainstorming. You would use BMAD-METHOD when building a non-trivial software project with AI assistance and wanting more structure than free-form prompting provides, particularly for solo developers who want a professional engineering process without a full team. It runs on Node.js and Python and integrates with AI coding tools like Claude Code and Cursor.

Copy-paste prompts

Prompt 1
Set up BMAD-METHOD in my project and walk me through the product manager agent workflow to define requirements for a task management app.
Prompt 2
I have a bug in my code. Use the BMAD-METHOD developer agent to diagnose it, propose a fix, and suggest tests to verify the solution.
Prompt 3
Help me design the architecture for a real-time chat application using the BMAD-METHOD architect agent, including database schema and API structure.
Prompt 4
Use the BMAD-METHOD UX agent to review and improve the user interface design for my e-commerce checkout flow.
Open on GitHub → Explain another repo

Generated 2026-05-18 · Model: sonnet-4-6 · Verify against the repo before relying on details.