How to study this course
🌱
Complete beginner
Never used Claude Code or Codex.
Path: T1 → T2 → T3 → T4
🚀
I already use Claude Code
You want to learn Codex and migrate projects.
Path: T3 → T4 → T5 → T6
🦜
Fluent in both
You want to stop maintaining duplicate skills.
Path: T5 → T6 (directly)
Detailed map of the learning paths
T1
Open track →
🧱 Fundamentals
Shared vocabulary. Who’s who. Why both.
- 1.1 — Claude Code vs Codex: terminology, philosophy, real differences
- 1.2 — Why use both together: complementarity and tool-agnostic
T2
Open track →
🔷 Anatomy of Claude Code
Everything that lives in CLAUDE.md e .claude/.
- 2.1 — CLAUDE.md, settings.json, hooks, slash commands
- 2.2 — Skills, sub-agents, MCP servers, and permissions
T3
Open track →
🟣 Anatomy of Codex
AGENTS.md, .codex/, .agents/ and the peculiarity of openai.yaml.
- 3.1 — AGENTS.md, config.toml, and agents in TOML
- 3.2 — Skills in Codex, openai.yaml sidecar, MCP deps
T4
Open track →
🟡 Converting your project
From AI-assisted migration to parallel maintenance.
- 4.1 — Automatic conversion prompt and post-migration checklist
T5 • FEATURED
Open track →
🦜 polyskill — cross-runtime skills
The heart of the course. One canonical source, two optimized outputs.
- 5.1 — The Agent Skills pattern, the drift problem, the adapter architecture
- 5.2 — polyskill CLI in practice: init, import, build, install, validate, reconcile
T6
Open track →
🌹 Advanced workflows
How people who use both agents orchestrate their day to day.
- 6.1 — Session handoff, two terminals, shared MCP, and governance
Prerequisites
- ✓ Comfortable with the terminal (bash, npm, git)
- ✓ Has installed at least one of the two: Claude Code (
npm i -g @anthropic-ai/claude-code) or OpenAI Codex CLI - ✓ An active account with at least one of the providers (Anthropic, OpenAI). Ideally, both.
- ✓ Editor with terminal integration (VS Code, Cursor, or similar) — optional, but recommended.