Open source · 2026
QuotaOps: one view of AI capacity and usage
An open-source, local-first dashboard for AI-account capacity, reset windows and locally observed token usage.
- Type
- Open source
- Year
- 2026
- Licence
- Apache-2.0
- Technology
- Next.js, React, TypeScript, Playwright
Overview
- QuotaOps is an open-source dashboard created by SO HOMELY. It brings AI-account capacity, reset windows and locally observed token activity into one local workspace.
- It is published under the Apache-2.0 licence and is currently a local-first MVP supporting Claude / Anthropic and OpenAI / ChatGPT Codex.
Problem
- People who work with several AI accounts have to check each provider and their local tools separately to know how much capacity is left and when their limits reset.
Technical approach
- Local-first: no hosted backend, database, account system or cloud sync. An optional collector runs on the same computer and listens only on 127.0.0.1.
- Provider-neutral quota windows with usage percentages and exact reset times. Provider adapters normalise the data before it reaches the dashboard.
- Capacity and token analytics are kept apart. Token figures come from local Claude Code and Codex session logs and are labelled "locally observed", not presented as billing data.
- Privacy by design: prompts, responses, source code, passwords and API keys are not intentionally exported or stored in dashboard data.
Solution
- A dashboard for Claude / Anthropic and OpenAI / ChatGPT Codex accounts with live reset countdowns, account states and best-capacity recommendations.
- Token analytics with input, output and cache breakdowns for today, the last 7 and 30 days, and all time.
- Reset notifications, keyboard shortcuts, optional voice input, and light and dark themes.
Outcomes
- Published as open source under the Apache-2.0 licence, with contribution, governance, privacy and security documentation.
- Continuous integration, dependency review and CodeQL analysis are set up in the public repository, and unit tests cover quota logic, provider parsers, notifications and token analytics.
- Currently an MVP: automatic collection is experimental, and there is no hosted version.