A methodology for building production-ready apps with AI assistance.
A structured approach to AI-assisted development that prevents the common failure pattern: spending thousands on tokens and ending up with broken code.
Core ideas:
- Scope an MVP before building everything
- Document first (AI has no memory—your docs are its memory)
- Work in focused tasks, one per session
- Check quality as you go (confidence scoring)
- Audit between phases
- Read the Introduction
- Set up your tools
- Have the brainstorming session
- Create your five docs
- Start building with the Cline workflow
docs/
├── introduction.md # What and why
├── part-1/ # Foundation
│ ├── philosophy.md # The mindset
│ └── tool-selection.md # Cline vs Claude Code
├── part-2/ # Pre-Development
│ ├── brainstorming.md # Scoping your MVP
│ └── documentation-architecture.md
├── part-3/ # Execution
│ ├── cline-workflow.md # Plan → Act → Verify
│ ├── task-patterns.md # Documenting tasks
│ └── confidence-scoring.md
├── part-4/ # Quality
├── part-5/ # Advanced
└── part-6/ # Resources & templates
See WRITING_GUIDE.md for documentation standards. Key rules:
- 1,500 word max per chapter
- Templates under 30 lines
- Say it once, don't repeat
- Write like you're talking to a smart friend
This methodology built RISE — an Electron desktop app — in 4 weeks for ~$400 in tokens. Production-ready, documented, maintainable.
Documentation rewrite in progress. Core chapters complete, advanced topics coming.