Will your quota last until the reset?

A window is what the vendor reports

Each row is one window the vendor returns, with its own length, name and reset time. Agent HUD does not assume what your plan's limits are. Claude Code reports a five-hour window, a weekly window and, on some plans, a weekly window for a model family. Codex names its windows by their length, and other limits by their own names. Every other agent with a quota reports its own.

When two programs share one account, its windows appear once: the Codex app and the Codex CLI on one sign-in, or a ChatGPT sign-in in Pi.

Which agents report a quota

Used or left

By default a percentage is the share used. The Mac can show the share left instead, so 82% used reads as 18% left. The colours, the burn rate, the forecast and the alerts always work from the share used.

A window with no reading shows no number, never a zero. Tokens are never turned into quota, and a missing quota is never worked out from token counts.

Burn rate and the forecast

The burn rate is the share of a window used per hour, measured from the readings of its current cycle. If that pace empties the window before its reset, the row shows the time it runs out. Otherwise it shows the share you will have used by the reset.

A forecast needs at least 15 minutes of readings from the current cycle. Until then, and again after each reset, the row says it is still sampling.

For example, with 40% left and a burn rate of 10% an hour, the window lasts about four hours if the pace holds. If it resets in three hours, it lasts; if it resets in six, it runs out first. The burn rate is separate from tokens per hour and from cost.

A reset counts once a reading confirms it

When a window's reset time passes, the row keeps its last reading and says it is waiting for an update.

The reset is confirmed when a new reading moves the reset time forward, or shows the window back at zero after it had been higher. Only then does Agent HUD say the quota is back. The vendor's timing, a clock difference or a reading not yet taken can all leave a passed deadline pending.

Every allowance counts on its own

Windows are never added together into one number. A weekly Claude window and a five-hour Codex window run on different clocks, and a total of the two would not say when either one stops you.

Quota belongs to the account. Two Macs signed in to the same account read the same windows from the vendor, so nothing is counted twice.

An agent that is missing, signed out or failing never hides the others. Its row names the reason, and its last reading stays, marked as out of date. Accounts are read one at a time, and no read sends a message to a model.

Balances and earned resets

A DeepSeek balance is an amount of money. Agent HUD shows the balance, the spend per day and the day it runs out. Each currency stays as it is, and money is never turned into a percentage.

Codex's earned resets are a count, shown as Codex reports it.

When does Agent HUD warn you?

The glow is green under 70% used, yellow from 70% and red from 90%. These levels are fixed. With Agent HUD Remote, the alerts for crossing 90%, a used-up window, a confirmed reset and a new earned reset also reach iPhone and Apple Watch.

The first reading after Agent HUD starts sets a baseline without an alert, and a reading older than 30 minutes raises none. On the HUD, Agent HUD alerts you:

• once when a window is at risk, because the forecast says it runs out before its reset or because it crosses 90%;
• when a window is used up;
• when a new reading confirms a reset, naming the account's other windows that are still used up;
• when Codex reports a new earned reset.

Replies and quota alerts on your Mac