Know before you hit the limit.
See what you’ve used. Hover to see when each allowance resets, so you can plan your next stretch of work.
FOR THE TOOLS YOU ALREADY BUILD WITH
You have better things to do than open another usage page. Super Notchy keeps the answers close, without taking over your workspace.
See what you’ve used. Hover to see when each allowance resets, so you can plan your next stretch of work.
A session can be working, finished, or waiting on you. A brief peek and optional sounds help you catch the handoff.
Pin the notch to any screen edge. Keep it visible, reveal it on hover, or let it fold away for full-screen apps.
A working copy of the notch, right here. Hover to unfold, rest on a ring for its card, open the palette, move it to another edge, or run a session to the limit.
Values illustrative. The real thing reads your own accounts, and folds away when you look elsewhere.
The ring gives you the headline. The hover panel gives you the context: limit windows, reset times, and the activity behind each account.
See current-session and weekly allowances together, with reset times and a list of your agent sessions.

Make the notch feel like it belongs. Choose your edge, size, and surface. Keep it quiet when your screen needs all the space.
Top, bottom, left, or right. Option-drag to find your spot.
Choose Liquid Glass, dark glass, or solid black.
Always visible or on hover. Sized to suit your workspace.
The notch at rest, on hover, and in a different tone; then a closer look at Appearance and Notifications.







Start with the assistants you use. Add the rest when you need them.
Install Super Notchy on macOS 15 or later.
Most providers use an existing local sign-in. Some need a separate sign-in or API key in Settings.
Arrange your accounts and choose how the notch appears. It stays close while your work stays in front.
Super Notchy reads supported provider data to show usage and activity. Switch a provider off to stop its polling and clear its saved readings. The original tool stays signed in.
Read the privacy policy ↗01 Provider connections you choose
02 Local preferences and usage cache
03 Optional website analytics
A small place for the things you keep checking. A one-time price for keeping them in view.
If Super Notchy isn’t a fit, request a full refund within 30 days of purchase.
Read the refund policy ↗Lifetime access. No recurring charge.
30-day refund policy · Applicable taxes may be added.
AI subscriptions and provider credits are separate.
Lifetime license details
It puts a compact panel on a screen edge of your Mac. Rings show supported AI assistants’ usage, and hover panels show more detail such as reset windows and session activity. It helps you see what is happening without repeatedly opening each provider’s usage page.
Claude (through Claude Code or Claude Desktop), Codex, Cursor, GitHub Copilot, Kimi, Kiro, Grok, Grok Bot, OpenCode, Command Code, Antigravity, GLM, Devin, MiniMax, DeepSeek, and Perplexity, plus Gemini token counts read from Gemini CLI, Hermes, and OpenCode. Local models are monitored in Ollama and LM Studio, and Ollama’s cloud usage can be shown too. Grok uses Grok Build, xAI’s command-line tool, with any SuperGrok account. What each one can show depends on the provider and your account.
No. Super Notchy creates its own small panel on a screen edge. You can choose the top, bottom, left, or right. The macOS app requires macOS 15 or later and supports Apple silicon and Intel.
No. The $27 purchase is for Super Notchy. Your AI subscriptions, provider credits, and API charges remain separate and are managed by those providers.
Yes. The offer is $27 USD for lifetime access, with no recurring Super Notchy subscription charge. Applicable taxes may be added at checkout. See the lifetime license details for the license scope.
Depending on the provider, Super Notchy reads local sign-in data, cached usage responses, session records, or model-server information and contacts the provider’s API. It stores preferences and last-known readings on your Mac. Some connections need an explicit sign-in or API key. The privacy policy explains these data flows and your controls.
Integrations depend on the information each provider makes available, which can change. When data cannot be refreshed, the app can show a stale reading or an authentication/error state. You may need to sign in again in the original tool. Exact real-time availability is not guaranteed.
Request a full refund within 30 calendar days of your purchase. Include your order reference and purchase email so the order can be located. Refunds go back through the original payment method. See the refund policy for the contact details and process.
For most providers, nothing new is created and nothing is typed in. Super Notchy reads the credential the tool already keeps for itself — a keychain item, a session file, a local database — read-only. It never refreshes, rewrites, or copies that credential, and it sends it to exactly one place: that provider’s own usage endpoint, the same place the tool sends it. When a token expires, the ring simply waits for the owning tool to renew it. That covers Claude Code, Codex, Cursor, GitHub Copilot, Kimi, Kiro, Grok, OpenCode, Command Code, Antigravity, GLM, and Devin. Grok Bot reads its saved sign-in too; macOS may ask once to allow access to “Grok Bot Safe Storage”.
Your Gemini API key is never opened at all: the app counts tokens from the logs Gemini CLI, Hermes, and OpenCode write, and the key stays wherever you put it. DeepSeek, Perplexity, and MiniMax have no key to borrow, so you sign in inside a window in the app; that session lives in Super Notchy’s own web store, nothing is read from Chrome or Safari, and signing out clears it. The only credentials you type in yourself are an Ollama cloud key and an LM Studio server token, and each is kept in your login keychain under a service name no other app uses.
None of it reaches us. Our licensing service receives your activation code, purchase email, and an identifier for your Mac — never a provider credential or a usage reading.
Claude has three sources, tried in order, so it can answer without interrupting you. If Claude Desktop is running, its local cache is read first — no keychain, no network. Otherwise the app asks Claude Code for its own /usage figures using the credential Claude Code already holds. Only if neither can answer does it read the OAuth token Claude Code keeps in your login keychain, which is why macOS may ask once whether to allow that; the token is read, never refreshed.
Codex is simpler: the app reads the session Codex signed in with and asks OpenAI’s usage endpoint with it. Codex keeps that session current, and Super Notchy never writes to it.
One payment. Lifetime access. 30-day refund policy.