Attach your Codex plan

CodeAF talks to the Codex service with your OpenAI account, through OpenAI's own Codex sign-in, so a Codex plan you already pay for can run conversations and crew seats. There is no API key to create and nothing to copy from another tool.

Try it in CodeAF

What your Codex plan powers

The plan's models appear in the picker as codex/gpt-5.5 and its siblings, each with a 272k token window, and the conversation you connected from moves onto codex/gpt-5.5. In /crew the providers row gains codex sub, and a worker, planner or checker seat on a GPT-5.5 model can route through the plan. A seat on a plan costs nothing extra in crew pricing, so to the crew picker it reads as the cheapest choice for that model.

Plan calls carry no dollar price. Tokens are still counted, but spend pages leave those lines blank rather than invent a cost. When the plan runs out, the turn stops with codex reached your chatgpt plan's usage limit · it resets on its own. The "chatgpt plan" in that line is the same Codex plan you signed in with.

Connect Codex from a terminal

codeaf connect codex does the same from a shell. On a machine with no browser, add --no-browser: it prints the link and the exact ssh -L 1455:localhost:1455 line to open it from your laptop. codeaf connect lists every model service and whether it is connected. To detach, press Enter twice on the Codex row in /connect, or run codeaf disconnect codex. Other plans, such as the Z.ai and Qwen coding plans or Kimi Code, connect with a key from the same list.

Things to build

• A cheap planner, a strong checker: put the planner seat on your plan and keep a paid model as the checker. Choose the model
• A server that uses your plan: sign in once on a headless box over the ssh tunnel, then run long work there.
• A team on the plan: a team whose workers sit on the plan's model, with a daily cap guarding everything else they spend. Budgets at every level

Go further