Let Codex implement while you handle planning
One person used ChatGPT for planning and Codex for execution, with a small port helper to keep parallel local runs from colliding.
- What they did
- The commenter split the work between two agents: ChatGPT shaped the approach, kept the versioning rules straight, and drafted the next instruction, while Codex handled the coding. They moved text back and forth by hand in long, extended-thinking conversations. In a separate setup, they also used a tiny tool called portbroker so an agent could find an open port, record it, and place it into $PORT automatically.
- How it went
- They said this reduced prompt drift and made local multi-agent work feel steadier and easier to predict. The tradeoff was the manual copy-paste loop, along with slowdown as threads grew long enough to turn to “sludge,” sometimes forcing a fresh session.
- Tips
- This works best when one agent owns the plan and another owns the edits, with a human checking the handoff. If you keep a port-reserving helper in the loop, make sure it only sets up the environment and pauses before anything destructive or irreversible.
Try it yourself with Codex
I’ve drafted a plan for [task] in [repo path or project name]: [paste the plan or link to it]. Follow it with minimal back-and-forth, and if you need to run any server, test, or local process that depends on a port, first reserve a free one, log it, and place it into [env var or config field] so parallel runs do not conflict. Work through the implementation, then give me a concise summary of the changes and the checks you completed. Before you push, delete, submit, or make any other irreversible change, stop and ask for my approval.
Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.
Get 3 agent uses like this every morning
With the day's top AI news. Free, and one click to unsubscribe.
Related agent uses
Review a pull request with staged sub-agents
Use parallel checks and a validation pass to review a branch or PR before you decide what to fix.
Use ChatGPT for planning and Codex for local execution
They set up a handoff where ChatGPT designs the work and checks it, while Codex makes the changes and runs commands on the local machine.
Use an agent to patch a stubborn app bug
They pointed a coding agent at two annoying software behaviors and had it debug one and add a self-improvement feature to the other.
Run agent coding jobs in a sealed local box
A developer used Codex to run an almost fully self-contained build-and-deploy workflow on a locked-down spare machine.
Create and launch a BaZi reading agent
A developer used coding agents to build a BaZi reader that calculates charts, writes interpretations, and handles language detection and USDT payments.
Monitor account caps and move chats across tools
teivz built a local app that watches Codex and Claude Code account usage and shifts a chat to another account when one reaches its cap.