Memo, plan, brief, table, or review note
Start in ChatGPT or Claude chat
You use multiple AI tools. This track shows where each tool should start and when to hand work to the next one. Chat tools shape the thinking. Codex and Claude Code change files and check results. The handoff names what is true, what is decided, and what is still unknown.
Memo, plan, brief, table, or review note
Start in ChatGPT or Claude chat
Changed files, tests, build, or PR
Start in Codex or Claude Code
A creative asset that will publish elsewhere
Start in chat, then hand off only if code is needed
Both thinking work and repo work
Start in chat. Write a packet. Then move to code.
Levels 2-3 | 3 lessons
Start in the tool that owns the finished surface.
This module teaches the first split: chat work, repo work, context transfer, and the moment a handoff is required. Users finish with a packet that keeps sources, decisions, unknowns, and exclusions separate.
Levels 3-4 | 3 paid lessons
Make tool boundaries explicit before actions run.
This module keeps code-agent setup safe. You learn where repo instructions live, when a tool needs permission, and when a connection belongs. The goal is a small rule set that says what can run, what must ask, and what is blocked.
Level 5 | 3 paid lessons
Use the second tool without letting it take over.
This module teaches creative-to-code handoff, second-agent review, and the supported Claude Code to Codex import path. It avoids unsupported plugin claims and ends with verification.
Use chat tools for thinking, files, analysis, and creative artifacts. Use code agents for changed files, commands, tests, diffs, and rendered proof. When ownership changes, write the packet before the next tool starts.