BMad Method: Breakthrough Method of Agile AI-driven Development

repository·main·Indexed 13 days ago

https://github.com/bmad-code-org/bmad-method

An Agile AI-Driven Development (AiDD) framework version 6.11.0 designed to transform ideas and change requests into working software. It features a core delivery loop via `bmad-build`, a suite of specialized modules (Builder, Creative Intelligence, Test Architect, Loop, and Game Dev Studio), and Web Bundles for Google Gemini Gems and ChatGPT Custom GPTs.

Tokens
209.7K
Snippets
329
Records
793
Agent score
99%

What's inside BMad Method

  1. Understand the Product Brief Coach Protocol

    main

    The Product Brief Coach Protocol is a framework for coaching users through the creation, updating, or validation of a product brief. The protocol emphasizes a conversational approach to 'draw the brief out' of the user, ensuring the final document is honest, right-sized, and feels like the user's own creation.

    Key operational principles include:

    • Core Stance: Prioritize conversation over speed. Push back on unexamined assumptions and ease up as the brief firms up.
    • Canvas Usage: Use a two-section Canvas structure:
      1. Brief: The primary 1-2 page deliverable (starts with status: draft).
      2. Addendum: A repository for depth that doesn't fit the brief (e.g., rejected alternatives, technical constraints, sizing data) and a Decisions subsection for a paper trail of scope cuts and overrides.
    • Visuals: Use Mermaid (for quadrantChart, flowchart LR, mindmap, or flowchart) and HTML tables (for matrices and success criteria) to convey meaning faster than prose.
  2. Automate development loops with bmad-build-auto

    main

    bmad-build-auto is an unattended automation tool for the Build implementation model. It automates a single development-loop iteration consisting of:

    1. Clarify: Clarify incoming intent.
    2. Spec: Create (or find and resume) a spec file.
    3. Implement: Implement the change.
    4. Review: Review the result.
    5. Finish: Write a terminal status to the spec file or a fallback result artifact.

    Architectural Boundary: bmad-build-auto owns the implementation run and the spec artifact it produces. It does not own your backlog policy. If a review finds issues outside the scope of the current story, the skill records them as deferred findings in the spec and stops. An external orchestrator is responsible for deciding whether to queue, deduplicate, or escalate those findings.

    # Example conceptual invocation
    # (Actual CLI command depends on the orchestrator/environment)
    bmad-build-auto --intent "Implement the user login validation logic"
  3. Design Specification for Drift (shadcn/ui + Next.js + Tailwind)

    main

    Drift is a design system specification for a task tracker built on shadcn/ui, Next.js, and Tailwind CSS. The design philosophy follows a "brand-layer delta" approach: it inherits 80% of its visual properties from shadcn/ui defaults and only explicitly overrides specific brand tokens (colors, typography, and shapes) to maintain a disciplined, tool-like aesthetic.

    Core Design Principles

    • Minimalism: Inherit shadcn defaults for everything not explicitly defined in the brand layer.
    • Purposeful Accent: Use the accent color exclusively to indicate "live" or "in-focus" items. Never use it for decoration, chrome, or state badges.
    • Visual Restraint: Avoid gradients, chromatic flourishes, or custom destructive colors.
    • Typography: Use a serif font (Instrument Serif) only for display roles (hero text, project titles) to act as a punctuation mark; use Geist Sans for all body and UI text.
    • Layout: Maintain a single-column layout with a maximum content width of max-w-3xl (768px).
    name: Drift
    description: Focused task tracker for solo founders and small async teams. shadcn/ui on Next.js + Tailwind; this DESIGN.md specifies the brand-layer delta only.
    colors:
      primary: '#0F4C81'
      primary-foreground: '#FFFFFF'
      accent: '#F59E0B'
      accent-foreground: '#1A1208'
      primary-dark: '#5C8AC2'
      primary-foreground-dark: '#0A1A2A'
      accent-dark: '#FBC470'
      accent-foreground-dark: '#1A1208'
    typography:
      display:
        fontFamily: 'Instrument Serif'
        fontSize: 36px
        fontWeight: '400'
        lineHeight: '1.15'
        letterSpacing: -0.01em
      display-sm:
        fontFamily: 'Instrument Serif'
        fontSize: 24px
        fontWeight: '400'
        lineHeight: '1.2'
    rounded:
      sm: 4px
      md: 6px
      lg: 8px
  4. Understand the PRFAQ Coach Protocol

    main

    The PRFAQ Coach Protocol is a coaching methodology based on Amazon's 'Working Backwards' approach. It guides users through creating a Press Release (PR) and Frequently Asked Questions (FAQ) to ensure customer-first clarity before any product development begins.

    Core Components

    • Press Release (PR): An announcement of the finished product. If a compelling PR cannot be written, the product is not ready.
    • Customer FAQ: Validates the value proposition from the customer's perspective (external).
    • Internal FAQ: Addresses feasibility, trade-offs, and risks from the builder's perspective (internal).
    • The Verdict: A candid assessment of what is 'Forged in steel', what 'Needs more heat', and where there are 'Cracks in the foundation'.

    Operating Principles

    • Customer-first: Always redirect focus from technology/solutions to the customer's problem.
    • Specificity over fluency: Reject 'weasel words' (e.g., 'best-in-class', 'seamless', 'revolutionary') in favor of concrete claims.
    • Hardcore Coaching: The protocol uses direct coaching and hard questions to challenge vague answers.
  5. Use the bmad-correct-course skill for sprint change management

    main

    The bmad-correct-course skill is designed to manage significant changes during sprint execution. When a user triggers it (e.g., by saying "correct course" or "propose sprint change"), the skill analyzes the impact across all project artifacts (PRD, epics, architecture, UX) and produces a structured Sprint Change Proposal.

    Goal: Analyze triggering issues, assess impact across artifacts, and produce an actionable proposal with clear handoff.

    Role: You act as a Developer navigating change management.

  6. Explore the BMad Ecosystem

    main

    The BMad ecosystem consists of the core method and several specialized modules designed for different development needs:

    ModulePurpose
    BMad MethodCore package to plan and deliver software (prototypes to established codebases)
    BMad BuilderTool for building skills, workflows, and agents
    BMad Creative Intelligence SuiteCreative thinking partners for innovation, design thinking, and storytelling
    BMad Test ArchitectEnterprise-grade testing add-on
    BMad LoopAutomates the building, verifying, and retrospecting of entire epics unattended
    BMad Game Dev StudioIdeation and building for game frameworks like Unity, Unreal, Godot, and Phaser
  7. Use the BMad Forge Idea skill to pressure-test ideas

    main

    The bmad-forge-idea skill is designed to take a half-formed idea and pressure-test it through persona-driven interrogation. The goal is to strengthen the idea, reject it, or clarify thinking before investing significant resources.

    When to use it

    Invoke this skill when you want to:

    • Forge an idea
    • Pressure-test an idea
    • Stress-test your thinking
    • Harden an idea

    Core Principles

    • Question, don't lecture: Lead with questions, press on weak points, and avoid vague claims.
    • Avoid agreement/praise: Do not use agreement or praise to smooth the interaction; this lowers pressure and leads to shallower thinking. In attack mode, never agree with the idea until the mode ends.
    • One question at a time: Work through dependencies and drive unresolved branches to concrete decisions.
    • Use personas: Use different voices (installed agents, user-defined personas, or generated experts) to provide diverse perspectives and sharper objections.
  8. Overview of BMad Deep Recon services

    main

    BMad Deep Recon is a research director service designed to produce decision-grade artifacts. It offers three primary service modes that can be combined into a workflow:

    1. Draft: Composes a high-quality, deep-research prompt tailored to a specific research type (e.g., market, technical) for the user to run in external tools like ChatGPT, Gemini, Grok, or Perplexity.
    2. Process: Takes a finished research report (from an external tool or manual source), files it, extracts claims, and distills a succinct, cited summary with metadata that other BMad skills can consume without reprocessing.
    3. Run: Performs native research directly within the environment using parallel web fan-out. This mode requires web access.

    A common workflow is Draft (create prompt) $\rightarrow$ Run externally $\rightarrow$ Process (turn external output into a BMad-compatible artifact).

  9. What is Advanced Elicitation and when to use it

    main

    Advanced Elicitation is a structured second-pass reasoning technique used to improve LLM-generated content. Instead of using vague prompts like "make it better," you select a specific reasoning method that forces the AI to re-examine its output through a particular analytical lens.

    When to use it:

    • When you need alternatives to generated content.
    • When output seems sufficient but lacks depth.
    • To stress-test assumptions or identify weaknesses.
    • For high-stakes content requiring rigorous rethinking.

    In BMad workflows, advanced elicitation is presented as a decision point after an LLM has generated content.