7.7 KiB
The BMAD-Method (Breakthrough Method of Agile (ai-driven) Development)
Welcome to the BMAD Method! This repository documents the V2 methodology (stable) and introduces the BETA-V3 (experimental).
- Stable Version (V2): Resources located in
CURRENT-V2/. This is the recommended version for general use. - Experimental Beta (V3): Resources located in
BETA-V3/. Contains new features and agents currently under testing.
Quick Links:
- V2 Setup Instructions (For IDEs/Gems/GPTs)
- BETA-V3 Setup Instructions (For IDEs/Web/Tasks)
- Community Discussion Forum
- Contributing Guidelines
- License
Legacy V1 has been archived.
What is the BMad Method?
The BMad Method is a revolutionary approach that elevates "vibe coding" to "Vibe CEOing." Unlike the spontaneity of pure vibe coding for quick prototypes, this method provides a structured yet flexible framework to plan, execute, and manage software projects using a team of specialized AI agents. It helps you build faster, cheaper, and more effectively by leveraging AI-driven processes throughout the entire product development lifecycle.
The V2 release (in CURRENT-V2/) established a robust workflow with distinct agent roles, standardized templates, and comprehensive checklists. BETA-V3 (in BETA-V3/) builds upon this foundation, refining processes and introducing new capabilities currently under evaluation.
The comprehensive, step-by-step approach transforms a product idea into a fully implemented application by:
- Structuring the development process into distinct AI persona-based phases.
- Generating detailed, high-quality artifacts at each phase.
- Using a sequential workflow with clear handoffs.
- Enabling AI to code the application based on well-defined specifications.
Guiding Principles
These principles apply across versions, with V3 enhancing certain aspects:
- Environment Recommendations: Generally, initial planning phases (Analyst, PM, early Architect work) are often well-suited for Web UIs (Gems/Custom GPTs) due to their conversational nature and cost-effectiveness. Technical implementation, file management (like V3's POSM Librarian), specialized tasks (like V3's RTE-Agent or IDE Tasks), and development are typically better suited for an IDE environment with custom agent support.
- No Rules Required (Flexibility): Agents primarily reference project documents (PRDs, architecture docs, coding standards, etc.) rather than relying heavily on proprietary AI platform rules.
- Iterative & Collaborative: Emphasizes a step-by-step, interactive process where agents collaborate with the user, pausing for input and validation at key decision points.
- Documentation Focus (Enhanced in V3): V3 places a stronger emphasis on creating a granular, indexed
docs/structure to improve context for AI agents and human developers.
Agent Overview (V2 Core Agents)
The CURRENT-V2/ system features a team of specialized AI agents. See CURRENT-V2/agents/ for IDE modes and CURRENT-V2/gems-and-gpts/ for Web UI versions.
Analyst (Business Analyst, Research Assistant, Brainstorming Coach)
Starting point for new/unclear ideas. Transforms concepts into a structured Project Brief.
Product Manager (PM)
Transforms ideas/briefs into detailed product plans. Creates the Product Requirements Document (PRD) with Epics and User Stories. V2 includes checklists for validation.
Architect
Defines the overall technical blueprint based on the PRD. Creates the Technical Architecture Document.
Product Owner (PO)
Validates the MVP plan (PRD, Architecture, Epics) against a checklist before development begins.
Scrum Master (SM)
Technical bridge generating detailed, executable Story Files from approved plans, one at a time, ready for development.
Developer Agent (Dev)
Implements features based on assigned Story Files, adhering to project architecture and standards.
New/Enhanced Agents in BETA-V3
BETA-V3 (in BETA-V3/) introduces new roles and enhances existing ones:
0. BMAD Method Advisor (0-bmad.md)
A dedicated guide explaining V3 concepts, including the new agents, IDE Tasks, and workflow nuances.
4. Design Architect (4-design-architect.md)
(Engage if project has a UI) Specializes in UI/UX (UI/UX Spec) and frontend technical strategy (Frontend Architecture Doc).
5. Technical POSM (5-posm.md)
Replaces the separate V2 PO and SM with enhanced capabilities: Phases:
- Master Checklist: Validates all planning/design docs.
- Librarian: Creates granular
docs/structure &index.md(IDE Recommended). - Story Creator: Generates developer-ready Story Files.
6. Release Train Engineer (RTE-Agent) (6-rte.md)
Manages significant mid-project changes/pivots. Uses rte-checklist.md for analysis and drafts artifact updates, outputting a Sprint Change Proposal.
(Note: The V3 PM and Architect agents also have refined phases and use updated V3 templates/checklists.)
Step-by-Step Process (V2 Typical Flow)
- Analyst: (Optional) -> Project Brief.
- PM: Project Brief/idea -> PRD (Epics, Stories).
- Architect: PRD -> System Architecture Document.
- PO: Validates PRD, Architecture, Epics -> Validation Report/Approval.
- SM: Approved plans -> Generates Story File (one at a time).
- Developer Agent: Implements approved story.
- Repeat 5 & 6 until MVP complete.
- Ongoing Advisory: Architect & PM provide support.
BETA-V3 Process Modifications
BETA-V3 refines the flow, particularly after initial planning:
- The Design Architect is inserted after the Architect if a UI is required.
- The POSM replaces the PO and SM:
- After planning docs are ready (PRD, Arch, Design Arch docs), POSM (Master Checklist Phase) validates everything.
- After validation/updates, POSM (Librarian Phase) organizes the
docs/structure. - Then, POSM (Story Creator Phase) can (but not recommended from the ide) generates stories one by one for for a full epic for the stories folder and dev agent pick up.
- The RTE-Agent can be engaged at any point after story development begins if a major issue or pivot occurs - the RTE (Release Train Engineer) can help figure out what needs to change and how to proceed, and help update all artifacts - best run from the ide since we are mid project.
(Refer to the BETA-V3 agent definitions for detailed flows).
IDE Tasks (BETA-V3 Feature)
Located in BETA-V3/tasks/, these self-contained instruction sets allow IDE agents (in V3 mode) to perform specific one-off jobs on demand (e.g., run checklist, create next story, shard docs), reducing the need for complex agent modes.
Tooling & Setup
V2 (Stable - in CURRENT-V2/)
- Templates:
CURRENT-V2/docs/templates/ - Checklists:
CURRENT-V2/docs/checklists/ - Web UI Agents (Gems/GPTs):
CURRENT-V2/gems-and-gpts/ - IDE Agents:
CURRENT-V2/agents/ - Setup Instructions:
CURRENT-V2/instructions.md
BETA-V3 (Experimental - in BETA-V3/)
- Templates:
BETA-V3/templates/ - Checklists:
BETA-V3/checklists/ - Web UI Agents:
BETA-V3/web-agent-modes/ - IDE Agents:
BETA-V3/ide-agent-modes/ - Tasks:
BETA-V3/tasks/ - Setup Instructions:
BETA-V3/instruction.md
Demonstration Walkthrough
The effectiveness of the BMAD Method's interactive approach is demonstrated in the V2 Video Walkthrough. The V2-FULL-DEMO-WALKTHROUGH folder contains the full transcripts and artifacts.
A BETA-V3 specific walkthrough is planned along with multiple sample projects.