Hey Product Hunt — I’m John, building Caplyr. (Cap Layer) I ran into the same issue over and over while shipping AI agents for my toolkit. I’d push something live, forget to cap usage properly, and then spend the next day worrying if a prompt loop was quietly burning money in the background. One test run actually did — not catastrophic, but enough to make it a real problem. I went looking for something simple that could just enforce a budget and stop calls when needed. Everything I found was either a full proxy setup or some heavy enterprise platform. Way overkill for what I needed. So I built Caplyr. It’s basically a thin wrapper around your existing client: ``` npm install caplyr const client = protect(new Anthropic(), { budget: { monthly: 500 } }) ``` That’s it — it just sits there and steps in when needed. What it does right now: * Budget guardrails → set daily/monthly limits, calls stop when you hit them * Auto downgrade → if you’re getting close, it switches to cheaper models * Kill switch → one click to stop all AI calls across your app Works with OpenAI, Anthropic, Groq, and Together AI. No proxy, no infra changes. I’m giving the first 5 users the Growth tier ($99/mo) free for life — 3 spots left. Would genuinely love to hear how others are handling this. Are you just monitoring costs, or actually putting hard limits in place?

Caplyr

Stop runaway AI API bills with 2 lines of code

Stop runaway AI API bills with 2 lines of code