Chrome extension that shows your remaining Claude, ChatGPT and Gemini usage limits (5-hour and weekly) in the toolbar badge, a popup and an always-on-top overlay. No cookies or API keys to paste, no server.
翻訳待ち
Loading the full README.
Read the full README ↗Claude, ChatGPT and Gemini usage — how much of the 5-hour and weekly limits you have used, or what is left — in Chrome's toolbar badge, a popup and an always-on-top overlay. Free and open source.
Add to Chrome (Chrome Web Store) · Website · 한국어 · Privacy policy
Unofficial project. Not affiliated with Anthropic, OpenAI or Google.
The extension uses the normal Chrome profile that is already signed in. It does not use Playwright, automate sign-in, or copy cookies/passwords out of the browser.
| Service | How usage is read | State |
|---|---|---|
| Claude | claude.ai authenticated internal usage endpoint | working |
| ChatGPT | chatgpt.com internal usage endpoint (the one its Settings > Usage page reads) | working — confirmed in a logged-in Chrome on 2026-10-15 (an account with a 30-day limit); accounts with 5-hour/weekly limits not yet seen |
| Gemini | gemini.google.com/usage page, read from an already-open tab | working — reading confirmed in a logged-in Chrome on 2026-10-16; needs that page open |
None of these are public, stable APIs. They may change without notice.
The extension keeps the latest usage in its own storage and shows it in its popup and toolbar badge.
The floating overlay is built into the extension (click Open overlay in the popup, then Float on top, or just press Enter); it uses Chrome's Document Picture-in-Picture, so there is no separate app to install.
For each service that is turned on, the popup shows how much of each limit is used and when it resets. The Used | Left switch at the top of the popup turns every number — popup, overlay and toolbar badge — into what is left, and back; the choice is kept. The 5-hour limit is the large number; an account without a 5-hour limit (e.g. ChatGPT Pro) shows its weekly limit there instead. 0% is shown as 0%; a value that could not be read is ?.
Colors follow what is left, whichever way the numbers read: green with half or more left, orange from 20% left, red below that — so 85% used is red. The bars are as long as the numbers shown.
The plan (e.g. Max 5x, Plus) is shown next to the service name when the service reports it. Gemini does not, so it has none.
One claude.ai sign-in can belong to several organizations (a personal plan and a Team workspace, say). Each one with chat limits gets its own block, named by its plan, so you can see which has how much left. The first one is the main value (the toolbar badge and the overlay's collapsed line).
When a read fails, the last good values stay visible, dimmed, with how old they are. Hover the status label for the failure reason.
Values older than 6 minutes (three missed refreshes) are dimmed; a good read that old is marked Outdated.
Reset times come from the services as exact times. Only ChatGPT's page fallback gives relative "4시간 20분 후 초기화"-style text; the popup counts that down (in the current UI language) from when it was read.
Turning a service on asks Chrome for that site's permission. If the popup closes while Chrome asks, allowing it still turns the service on.
The UI is available in English, Korean, Spanish, Chinese (Simplified and Traditional) and Japanese. Chrome picks it from your browser language, falling back to English.
To see several Claude (or ChatGPT, Gemini) accounts at once, sign in to each in its own Chrome profile and install Leeway in each. With the Leeway helper installed, every profile's popup and overlay also show the other profiles' usage.
leeway-helper.exe from the latest release (the popup's Get the helper (Windows) link goes straight to it) and run it once. It registers itself for your Windows user (no administrator rights) and copies itself to %LOCALAPPDATA%\Leeway\helper. Each release also has leeway-helper.exe.sha256 to check the download against.Work.This is an experimental, Windows-only feature. The helper is not code-signed yet, so Windows SmartScreen may warn the first time it runs, and Windows 11 Smart App Control or a work PC's policy can block it.
This is the beginning of the README. Read the rest at the original. Read the full README ↗