Track Codex usage before you run out.

Codex-style usage limits are awkward because they show up exactly when you are trying to stay focused. You might be reviewing a change, asking for a refactor, testing an idea, or debugging a failure. The usage limit is real, but it is usually not the thing you want to think about in the middle of the work.

A raw usage number helps a little. It tells you what has happened so far. It does not tell you whether the current burn rate is sustainable for the rest of the window. AgentPace is designed for that second question. It turns Codex usage into a visible pace graph so you can see whether you are ahead of or behind the ideal line.

Why burn-rate pacing is more useful than raw totals

Usage totals are static. They are useful at the end of a window, but less useful when you are deciding what to do next. If you have used a meaningful share of the window early, that might be fine if the reset is close. It might be a problem if the reset is far away. Without a time reference, the same number can mean different things.

AgentPace gives that number a time reference. It compares actual usage to an ideal pace line across the active window. The app does not need to predict your whole day or decide how you should work. It shows the direction of travel. If the actual line is moving ahead of the ideal pace, your current workflow is spending faster than the window can comfortably support.

A Codex usage meter that lives where you work

The menu bar is the right place for this kind of utility because the information is ambient. You do not need a cloud dashboard, a big report, or a second screen. You need a quick answer while your editor is still the main context. AgentPace opens as a compact popover, shows the current usage state, and gets out of the way.

That workflow matters for Codex because the usage question often comes up between actions. Should you keep iterating with the agent? Should you save usage for the next task? Should you take a break until reset? AgentPace cannot make those tradeoffs for you, but it can make the budget visible enough that the choice is not blind.

It also keeps usage tracking from becoming its own ritual. You do not need to write down numbers, maintain a spreadsheet, or remember when the current window started. The app keeps the current state close enough to check without interrupting the work.

Actual versus ideal pace

The chart in AgentPace is intentionally direct. Actual usage shows how the current window has been consumed. Ideal pace shows what the line would look like if usage were spread evenly until reset. The gap between those two lines is the useful part. It tells you whether the current Codex usage pattern is conservative, roughly on pace, or likely to run hot.

This is especially useful across shorter reset windows, where a few intense hours can change the whole budget. It is also useful across longer windows because a limit can feel far away until it suddenly is not. A pace meter makes the drift visible earlier.

Local-first by default

AgentPace is a local macOS utility. It has no account, backend dashboard, telemetry, ads, or analytics. Usage snapshots and refresh errors stay on your Mac. That is a deliberate product boundary. A usage tracker for coding agents should not become another service that watches how you work.

The paid unlock is separate from usage tracking. AgentPace has a 7-day free trial and a $4.99 lifetime license key, but the product experience stays small: a local menu bar app that helps you understand usage pace.