Run multiple Codex agents at once without losing track

Install Codex and sign in first. Hercus runs the Codex you're signed in to, so your ChatGPT plan works as it is.

1. 1

   Open each repo as a project

   Add every repository you work in. Each one becomes its own project, so its Codex sessions never mix with another repo's.

   You'll see: Your repos listed in the Projects panel.

   In the docs: Add a project

2. 2

   Make Codex your default agent

   Open Settings, go to Agents and select Set as default for Codex. New tasks then start on Codex without you choosing it each time.

   You'll see: Codex marked as your default agent.

   In the docs: Choose an agent for a task

3. 3

   Give each feature a workspace

   Inside the project, create a workspace for each feature. Choose a worktree when a feature needs its own branch and files, so its Codex sessions don't touch another feature's checkout.

   You'll see: Each feature under its project, on its own branch if you chose a worktree.

   In the docs: Give a change its own branch

4. 4

   Start one Codex task per job

   Add a task to the workspace and give it one clear job. Hercus follows each task through hooks it adds to Codex's settings, so every task shows Running, Needs input or Completed without you opening it.

   You'll see: Your Codex tasks on the board, each with its status.

   In the docs: Hooks

5. 5

   Let a task split the work

   Ask a Codex task to split a job, for example: "Split this into three Codex subtasks, one for each endpoint, and review their changes when they finish." It starts the subtasks, sends their instructions and reads their results, and Hercus tells it when each one finishes.

   You'll see: The subtasks listed under the task that started them.

   In the docs: Ask for subtasks

6. 6

   Keep them moving

   Queue your next instruction while Codex is busy, and turn on auto-send to send it as soon as the task needs input. The status bar shows how much of your plan's 5-hour and weekly limit you've used, and Keep computer awake while agents are running stops your Mac sleeping mid-task.

   You'll see: Your next instruction waiting under the task.

   In the docs: Queue follow-up instructions