Every AI API provider gives you a snapshot of current usage. None of them show you trends over time, project when you will hit your limit, or let you compare across providers.
I built onWatch to solve this. It runs in the background as a single Go binary, polls your configured providers every 60 seconds, stores everything locally in SQLite, and serves a web dashboard.
What it shows you that providers do not:
- Usage history from 1 hour to 30 days
- Live countdowns to each quota reset
- Rate projections so you know if you will run out before the reset
- All providers side by side in one view
Around 28 MB RAM, no dependencies, no telemetry, GPL-3.0. All data stays on your machine.
https://onwatch.onllm.dev https://github.com/onllm-dev/onWatch
[link] [comments]