Help
Support
Email support@mannbadal.com and you will get a reply from the person who wrote the app.
Connecting Claude
Choose Sign in with Claude. Your browser opens Anthropic's sign-in page, you approve access, and Claude shows you a code to paste back into AI Quota. AI Quota asks for profile access only, so it can read your limits but cannot send prompts.
If you already use Claude Code on a Mac, the second option reuses its existing sign-in: run the command AI Quota shows you in Terminal and paste the result. A sandboxed App Store app cannot read another app's Keychain items directly, which is why this is a copy and paste rather than an automatic import.
Connecting Codex
On a Mac, paste the contents of ~/.codex/auth.json —
the file the Codex CLI creates when you sign in with your ChatGPT
account. AI Quota reads the token and your plan tier from it.
Codex limits exist only for ChatGPT plan sign-ins. If your
auth.json contains an API key instead, there is no plan
allowance to meter and AI Quota will tell you so rather than showing
an empty gauge.
Connecting Cursor
Cursor has no personal usage API, so AI Quota reuses the session your Cursor app already holds. Sign in through AI Quota and it captures the session token.
This is the most fragile of the four connections. Signing out of Cursor, or a change to how Cursor authenticates, can require you to reconnect. When that happens the account shows a Reconnect badge and your existing figures stay visible.
Connecting GitHub Copilot
Create a classic personal access token on GitHub with
the read:user scope, then paste it into AI Quota along
with your GitHub username.
Fine-grained tokens do not work here. GitHub's billing endpoints accept classic tokens only — this is a GitHub restriction, not an AI Quota one. If a token is rejected, check that it is the classic kind before anything else.
A meter has stopped updating
AI Quota shows the last figure it managed to retrieve, with its age underneath, so a stale meter looks stale rather than wrong. Common causes:
- Rate limiting. Anthropic in particular limits how often usage can be checked. AI Quota deliberately spaces its requests out and backs off further after a refusal, so a meter can legitimately be up to an hour old. Forcing repeated refreshes makes this worse, not better.
- An expired sign-in. The account shows a Reconnect badge. Reconnecting takes a few seconds and keeps your history.
- A service changed its interface. Only GitHub publishes a documented usage endpoint; the others can change without notice. If a meter stays stale after reconnecting, please email — knowing which service and roughly when it stopped is genuinely useful.
Widgets are not refreshing
Widgets are refreshed on a schedule the system controls, and AI Quota keeps its own request budget so several widgets never multiply into several requests. Opening the app always refreshes immediately, within the limits above. If a widget looks stuck, check that the app itself shows current figures first.
Why do the analytics say "activity" rather than tokens?
Most of these services report how full a window is, not how many tokens you spent. When a real token count is unavailable, AI Quota builds the heatmap from the movement it observes and labels it as activity, because presenting that as a token total would be a fabrication. On a Mac, enabling local log reading in Settings gives true token counts, and those figures sync to your other devices.
Team and enterprise accounts
AI Quota supports personal accounts. Team, business and enterprise plans report usage through administrator APIs that a personal app cannot read.
Refunds
Purchases and refunds are handled by Apple. Request a refund through reportaproblem.apple.com.