BMAD-METHOD/expansion-packs/bmad-technical-writing/data/bmad-kb.md

6.9 KiB

BMad Technical Writing Knowledge Base

Overview

BMad Technical Writing transforms you into a "Book Director" - orchestrating specialized AI agents through the technical book creation process. This expansion pack provides structured workflows for creating high-quality technical books with code examples, tutorials, and progressive learning paths.

When to Use BMad Technical Writing

Use this expansion pack for:

  • Writing technical books (PacktPub, O'Reilly, Manning, self-publish)
  • Creating comprehensive tutorials and course materials
  • Developing technical documentation with code examples
  • Updating existing technical books (2nd/3rd editions, version updates)
  • Incorporating technical reviewer feedback
  • Managing code example testing and maintenance

The Core Method

1. You Author, AI Supports

You provide:

  • Technical expertise and domain knowledge
  • Teaching insights and pedagogical decisions
  • Code examples and real-world experience

Agents handle:

  • Structure and organization
  • Consistency and quality assurance
  • Learning progression validation
  • Publisher compliance checking

2. Specialized Agents

Each agent masters one aspect:

  • Instructional Designer: Learning architecture, objectives, scaffolding
  • Code Curator: Example development, testing, version management
  • Tutorial Architect: Step-by-step instruction, hands-on learning
  • Technical Reviewer: Accuracy verification, best practices (Sprint 2)
  • Technical Editor: Polish, clarity, consistency (Sprint 2)
  • Book Publisher: Submission packaging, formatting (Sprint 2)

3. Quality-First Approach

Multiple review passes ensure:

  • Technical accuracy and current best practices
  • Working code examples tested across versions
  • Clear learning progression with proper scaffolding
  • Publisher compliance and formatting
  • Pedagogically sound instruction

Four-Phase Approach

Phase 1: Planning (Web UI - Gemini/ChatGPT)

Agents: Instructional Designer

Activities:

  • Design book outline with learning path
  • Define book-level and chapter-level learning objectives
  • Map prerequisites and dependencies
  • Structure parts and chapters
  • Plan code repository

Outputs:

  • Complete book outline
  • Learning objectives matrix
  • Chapter dependency map

Phase 2: Development (IDE - Cursor/VS Code/Claude Code)

Agents: Tutorial Architect, Code Curator

Activities:

  • Create detailed chapter outlines
  • Write chapter content with tutorials
  • Develop code examples
  • Test code across versions/platforms
  • Create exercises and challenges

Outputs:

  • Chapter drafts
  • Working code examples
  • Exercise sets
  • Test results

Phase 3: Review (IDE or Web UI)

Agents: Technical Reviewer, Technical Editor (Sprint 2)

Activities:

  • Technical accuracy verification
  • Code quality review
  • Editorial pass for clarity
  • Consistency checking
  • Publisher guideline compliance

Outputs:

  • Technical review reports
  • Edited chapters
  • Code improvements

Phase 4: Publishing (IDE)

Agents: Book Publisher (Sprint 2)

Activities:

  • Format for target publisher
  • Package submission materials
  • Create index and glossary
  • Final quality assurance

Outputs:

  • Publisher-ready manuscript
  • Submission package
  • Companion code repository

Agent Specializations Summary

Instructional Designer 🎓

  • Creates book and chapter outlines
  • Defines learning objectives using Bloom's Taxonomy
  • Designs learning paths with proper scaffolding
  • Maps prerequisites and dependencies
  • Ensures pedagogical soundness

Tutorial Architect 📝

  • Designs hands-on tutorials
  • Creates step-by-step instructions
  • Develops exercises and challenges
  • Ensures reproducibility
  • Adds troubleshooting guidance

Code Curator 💻

  • Develops working code examples
  • Tests code across versions and platforms
  • Manages version compatibility
  • Ensures code quality and best practices
  • Creates automated test suites

Best Practices

Learning Progression

  • Start simple, add complexity gradually
  • Introduce concepts before using them
  • Provide practice before advancing
  • Use Bloom's Taxonomy progression (Remember→Understand→Apply→Analyze→Evaluate→Create)
  • Validate prerequisites are clear

Code Examples

  • Every example must be tested and working
  • Follow language-specific style guides
  • Include inline comments explaining WHY, not WHAT
  • Document setup and dependencies precisely
  • Test across specified versions and platforms
  • Provide troubleshooting for common issues

Tutorial Design

  • Use clear, actionable steps
  • Document expected results at each stage
  • Provide hands-on practice opportunities
  • Include troubleshooting guidance
  • Ensure reproducibility

Chapter Structure

  • Introduction with real-world motivation
  • Learning objectives stated upfront
  • Concepts explained before application
  • Tutorials reinforce concepts
  • Exercises provide practice
  • Summary recaps key points

Quality Assurance

  • Use checklists to validate quality
  • Test all code examples before publishing
  • Verify prerequisites are explicit
  • Ensure learning objectives are measurable
  • Check alignment with publisher guidelines

Publisher-Specific Considerations

PacktPub

  • Hands-on, project-based approach
  • Practical tutorials throughout
  • Clear learning outcomes per chapter
  • Code-heavy with examples

O'Reilly

  • Learning path structure
  • Exercises after each concept
  • Real-world examples
  • Theory balanced with practice

Manning

  • Deep tutorial style
  • Progressive build approach
  • Iterative improvements
  • Comprehensive coverage

Self-Publishing

  • Flexible structure
  • Follow general best practices
  • Consider target platform (Leanpub, KDP, etc.)
  • Maintain high quality standards

Bloom's Taxonomy Reference

Use action verbs appropriate to learning level:

  • Remember: Define, List, Name, Identify, Describe
  • Understand: Explain, Summarize, Interpret, Compare
  • Apply: Implement, Execute, Use, Build, Demonstrate
  • Analyze: Analyze, Debug, Troubleshoot, Examine
  • Evaluate: Evaluate, Assess, Critique, Optimize
  • Create: Design, Develop, Architect, Construct

Version Management

For technical books:

  • Specify exact versions in prerequisites (e.g., "Python 3.11+")
  • Test code on all supported versions
  • Document version-specific behaviors
  • Create version compatibility matrix
  • Plan for updates when new versions release

Brownfield Support

BMad Technical Writing fully supports updating existing books:

  • Add new chapters to existing content
  • Update code examples for new framework versions
  • Refresh outdated examples
  • Incorporate technical reviewer feedback
  • Maintain consistency with existing content
  • Update for new publisher requirements

Success Metrics

A successful technical book should:

  • Have clear, measurable learning objectives
  • Include working code examples (100% tested)
  • Provide hands-on tutorials and exercises
  • Follow proper learning progression
  • Meet publisher guidelines
  • Enable readers to achieve stated objectives