Execution harness · your own ChatGPT subscription
Run your chats on OpenAI Codex
Pick OpenAI Codex as the harness in any thread and the round runs on the Codex CLI under your ChatGPT subscription or OpenAI API key — the mesh tools, nodes and context you already have, driven by a different engine.
What installing does
Getting the plugin gives you access to it, and your harness picker offers the harness from the package — nothing is copied into your home (installations that still copy apps place it at {you}/Harness/Codex).
It never appears for users who haven't opted in, because it needs a personal ChatGPT subscription
or OpenAI API key: there is no shared key and nothing is billed to the mesh.
- Get the plugin — the harness appears in your thread's harness picker.
- Pick it in a thread, then run
/loginto connect your ChatGPT account (a device code you confirm in the browser) or paste an OpenAI API key. - Chat as usual — same threads, same tools, same mesh content.
Uninstalling removes only a legacy home copy; it does not revoke your access, so the picker keeps offering the harness. Switch a thread to another harness to stop using it — no history is lost.
Good to know
- It is a harness, not an agent: it changes what executes the round, not which agent you talk to.
- Each thread is one Codex thread: every round resumes it, so the CLI keeps the thread's context and tool results itself. Command runs, MCP tool calls, file changes and web searches show up in the thread as tool calls; the model's reasoning summaries stream alongside the answer.
- How the harness drives the CLI: @@("AppServerProtocol").
- Nothing is shared. Your subscription, your login, your picker.