BMAD Method

BMAD-METHOD (Breakthrough Method of Agile AI-Driven Development) is an open-source framework hosted on GitHub that organizes AI-assisted software development through a structured multi-agent methodology. It defines specialized agent personas (such as Analyst, Architect, Developer, and QA) that collaborate through structured prompts and documents to guide a project from ideation to implementation. The method emphasizes context engineering — carefully managing what information each agent receives — to maximize the quality and coherence of AI-generated outputs. It is designed to work with popular AI coding assistants and chat interfaces, making it accessible without requiring custom infrastructure.

• Multi-agent role system — Defines distinct agent personas (Analyst, Architect, PM, Dev, QA) each with scoped responsibilities and prompt templates to simulate a full development team.
• Structured project documents — Uses a set of living documents (PRD, architecture doc, stories) as the shared context that agents read and update throughout the project lifecycle.
• Context engineering — Carefully controls what context is fed to each agent at each stage, reducing hallucinations and keeping outputs aligned with project goals.
• Iterative story-driven workflow — Breaks work into user stories that are planned, implemented, and reviewed in cycles, mirroring agile sprint practices.
• Model-agnostic — Works with any capable LLM or AI coding assistant (Claude, GPT-4, Gemini, etc.) via chat or IDE integrations.
• Open-source and community-driven — Fully open on GitHub under a permissive license; contributions, issues, and discussions are welcomed from the community.
• Getting started — Clone or fork the repository, follow the setup guide in the README, and begin by running the Analyst agent prompt against your project idea to generate an initial PRD.