Open source · MIT · self-hosted

All your AI usage limits
in one place

Track the limits of every AI account you own — Claude, ChatGPT, Cursor, Copilot and more, including several accounts of the same service — with live reset timers, low-limit alerts and real usage stats. Your tokens never leave your own device.

Claude ChatGPT Gemini Cursor Copilot Perplexity Mistral + any via cURL

Built for people with more than one AI account

Every other tracker assumes one account per service. This one doesn't — two Claude accounts, two ChatGPT accounts, all on one screen.

👥

Multiple accounts, isolated

Add any number of accounts across any service. Each keeps its own credentials, token refresh and history — no mixing, no logging in and out.

⏱️

Live reset timers

See exactly when each 5-hour and weekly window resets, counting down in real time. Never get surprised by a wall mid-task again.

📊

Honest usage stats

Day streaks, limit hits, time in the red zone, burn-rate forecast and busiest hours — all from real measurements. No invented message counts.

🔔

Low-limit alerts

Get notified at 20%, 10% and 0% remaining, on restore, and when a session needs re-login. Optional Telegram push from the server.

🔒

Private by design

Tokens live only in your device's Keychain/Keystore or your own encrypted server. Requests go straight to the AI providers — no third-party backend.

📱

Install anywhere

A mobile app, a self-hosted web panel you add to your home screen like a native app, and a Chrome extension for a glance from the toolbar.

Supported services

Ten services out of the box — and a Custom option that adds any other by pasting a cURL, with the usage fields detected automatically.

C
Claude5h · weekly · Opus/Sonnet · spend
C
ChatGPTCodex 5h & weekly · credits
G
Geminisession & weekly
C
Cursorpremium (fast) requests
C
GitHub Copilotpremium requests / month
P
PerplexityPro searches
G
Grokrequest window
M
MistralLe Chat
D
DeepSeekaccount limits
P
Poecompute points
+
Any other servicepaste a cURL — fields auto-detected

Three ways to use it

Same engine, your choice of surface.

📱

Mobile app

Your daily driver. Expo / React Native, unlimited isolated accounts.

iOS & Android
🖥️

Web panel

Self-hosted PWA with 24/7 background polling. Install to your phone's home screen.

Node 22, zero deps
🧩

Chrome extension

A glance from the toolbar using your browser's own session.

One account per profile

Self-host in two minutes

One Node file, no runtime dependencies. Runs on any small VPS.

# clone & build
git clone https://github.com/Rauan228/AI-Limits-Tracker.git
cd AI-Limits-Tracker && npm install
npm run build:server && npm run build:web

# configure & run
cp server/.env.example server/.env   # set a password
WEB_DIR=web/dist DATA_DIR=./data node server/dist/server.mjs
# → open http://localhost:8795 on any device

Full deploy guide →

Frequently asked questions

Which AI services does it support?

Claude, ChatGPT, Gemini, Cursor, GitHub Copilot, Perplexity, Grok, Mistral, DeepSeek and Poe out of the box — plus a Custom option that adds any other service: paste a cURL of the request that returns your usage and the app auto-detects the percentage used and the reset time.

Can I track multiple Claude (or Cursor, or ChatGPT) accounts at once?

Yes — that's the whole point. Add any number of accounts across any service. Each is fully isolated with its own credentials, token refresh and history, so two accounts of the same service show two different numbers side by side.

How do I see all my AI limits in one place?

Install the mobile app, the self-hosted web panel (which you can add to your phone's home screen), or the Chrome extension. It shows each service's windows — Claude's 5-hour and weekly, ChatGPT's Codex limits, Cursor's premium requests, and so on — on a single screen with live reset timers.

Is it safe? Where are my credentials stored?

They never leave your own device or server. On mobile they're in the iOS Keychain / Android Keystore; on the server they're AES-256-GCM encrypted. Requests go straight to Anthropic, OpenAI and Google — there's no third-party backend. The tool only reads your usage, it never generates anything, so it adds no load to your quota.

Is it really free and open source?

Yes. MIT-licensed, free, and the full source is on GitHub. You self-host it yourself and own all your data.

Does it show how many messages or tokens I've used?

No AI subscription exposes a message or token counter, so the app never invents one. It reads the percentage of each rate-limit window the service itself reports, and derives streaks, forecasts and charts from how that percentage changes over time.

Stop logging in and out just to check

Put every AI limit you care about on one screen. Free, open source, yours to self-host.

★ Get it on GitHub