I Built an MCP Server That Tracks AI Quota Resets — Here's How

A programmer created a Model Context Protocol (MCP) server called Reset Beer that monitors AI quota resets for multiple products. AI subscription services like OpenAI Codex, Claude Pro, and Gemini Advanced charge users for usage, and when the quota resets, the cost can be significant. However, the quota reset information is scattered across various status pages and there is no unified API to track them.

To address this issue, the programmer built Reset Beer, a dashboard that aggregates real-time AI quota reset signals from over 15 products and sends alerts to the user. Pro users of Reset Beer can connect the service to their AI assistants and query any recent quota resets for a specific product. The MCP server's architecture is a Next.js API route that implements JSON-RPC 2.0 over HTTP, and it exposes three tools: list_recent_resets, get_product_status, and list_products.

Each Pro user is assigned a unique API key for authentication and it is stored as a SHA-256 hash in the database.

The programmer learned that implementing MCP is relatively straightforward, with a well-designed API and simple API key authentication. However, they also faced challenges with authenticating first-party tools and ensuring that the dashboard's data is actionable and up-to-date. The Reset Beer dashboard can be accessed at reset.beer, and it can be installed using npm or found on GitHub.

The service costs $9.99 as a one-time payment with a free tier available. The programmer built the service using Next.js 15, Neon Postgres, Better Auth, Resend, and Vercel. They are active on X.