AI Dev Curriculum
Phased curriculum for AI-assisted development
A hands-on, phased curriculum for learning AI-assisted development by directing agents to build real software. Phases are git tags — clone, branch from phase-00, and level up through SaaS builds, quality engineering, research, protocols, skills ecosystems, and model evaluation. The open repo is the product.
What's in it
- Phases 00–06: first AI build → SaaS killers → quality → research → protocols → skills → eval lab
- Claude Code starter kit with slash skills, agents, and background skills
- Phase 04: A2A discovery, task delegation, x402 gates, AP2 mandates, protocol dashboard
- Reference implementations as git tags so each phase starts from a known-good state
- Built to run solo or with a team — clone, branch, and level up at your pace
Phases
Each phase is a git tag. Branch from the tag, build, then move on — your work stays on your branches while the reference stays clean.
Phase 00
Your first AI build
Intention → PRD → plan → build — with the starter kit pipeline.
Phase 01
SaaS killers
Real apps with auth, databases, and tests.
Phase 02
Quality engineering
Scan, fix, monitor, harden, and ship with zero-tolerance gates.
Phase 03
Research
Understand any codebase and author your own tools.
Phase 05
Skills ecosystem
Browse, filter, and preview skills across the ecosystem.
Phase 06
Model eval lab
Tournament-style model evaluation with analyst prompts and judges.
Start here
Clone the open curriculum, check out Phase 00 on your own branch, and open the phase guide before you write a line.
git clone https://github.com/GolferGeek/ai-dev-curriculum.git cd ai-dev-curriculum git checkout -b my-phase-00 phase-00