Back
Blog
A guide on building a development workflow that's faster and more reliable. Covers plan-based development, TypeScript, Turborepo, testing LLM outputs, and juggling this across packages.
Start with Part 1 →How we solved AI laziness with a self-correcting loop that actually finishes tasks. Covers task breakdowns, prompt engineering, safety features, and real-world implementation.
Start with Part 1 →- Claude Code Series · Part 9 CLAUDE.md Files That Actually Work
Documentation that bridges human knowledge and AI understanding
- Claude Code Series · Part 8 Testing LLM Outputs Systematically
Evalite and the new paradigm of AI unit tests
- Claude Code Series · Part 7 The Virtuous Cycle: Putting It All Together
Lessons learned from six months of plan-based AI development
- Claude Code Series · Part 6 From Chatbot to Development Partner
Custom commands and skills that make Claude Code an integrated workflow
- Claude Code Series · Part 5 Fast Tests Beat Complete Tests
Why prioritizing fast feedback loops over comprehensive coverage wins
- Claude Code Series · Part 4 TypeScript Strict Mode as Your First Defense
Why strict mode accelerates development instead of slowing it down
- Claude Code Series · Part 3 Taming 60+ Packages with Turborepo
How Turborepo orchestrates complexity and makes large monorepos manageable
- Claude Code Series · Part 2 Write the Plan First, Then the Code
How planning documents accelerate development and reduce rework
- Claude Code Series · Part 1 AI-Assisted Dev That Actually Works
How plan-based development transformed our 60+ package monorepo workflow
- Why a monorepo?
but that sounds complicated