惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

J
Java Code Geeks
Engineering at Meta
Engineering at Meta
GbyAI
GbyAI
MongoDB | Blog
MongoDB | Blog
Blog — PlanetScale
Blog — PlanetScale
腾讯CDC
U
Unit 42
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
Apple Machine Learning Research
Apple Machine Learning Research
M
MIT News - Artificial intelligence
人人都是产品经理
人人都是产品经理
Hugging Face - Blog
Hugging Face - Blog
MyScale Blog
MyScale Blog
小众软件
小众软件
博客园 - 三生石上(FineUI控件)
N
Netflix TechBlog - Medium
阮一峰的网络日志
阮一峰的网络日志
博客园 - Franky
Recent Announcements
Recent Announcements
A
About on SuperTechFans
Stack Overflow Blog
Stack Overflow Blog
The GitHub Blog
The GitHub Blog
D
Docker
H
Hackread – Cybersecurity News, Data Breaches, AI and More

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
A small open-source library for scoped, budgeted, time-bo...
Kat Laszlo · 2026-06-19 · via DEV Community

Kat Laszlo

When I led self-serve at a usage-based data company, one of the most common feature requests was credit limits per API Key. * People wanted to hand a key to a script, a teammate, or now an AI agent, and know it couldn't run up the whole bill. We get the same request at Tanso.

Account-level and user-level limits exist — That's what enterprise quota systems are for. But they're heavy. For a startup there wasn't a simple drop-in. So I wrote one.

agentkey does four things:

  • Cap what a key can spend: A budget per key, per day or month
  • Scope what it can do: Least privilege per key
  • Set when it expires: Short-lived by default if you want
  • Record which human authorized it: Delegation you can audit

The gap

AI agents made this urgent. An agent spends on its own — a loop or a bad prompt can burn a month's budget before anyone looks at a dashboard. And here's the part most tools miss: scoped keys tell you what an agent can do, not how much it can spend. LLM gateways cap spend. Identity platforms scope keys. Neither does both at the key level. agentkey does.

How it works

It's not a new auth system. It adds a few columns to your existing Postgres keys table and gives you a small API.

npm install @katrinalaszlo/agentkey

Create a key that's scoped, budgeted, and expiring:

import { AgentKey } from '@katrinalaszlo/agentkey';

const ak = new AgentKey({ pool }); // your pg Pool

const key = await ak.create({
  accountId: 'acct_123',
  scopes: ['proxy.chat'],
  budgetCents: 5000,        // $50 cap
  budgetPeriod: 'month',
  expiresIn: '7d',
  delegatedBy: 'user_456',  // the human who authorized this agent
});

Validate on each request, and track spend after a call:

const result = await ak.validate(key.key);
// { valid: true, scopes: ['proxy.chat'], budgetRemainingCents: 5000, ... }

await ak.trackUsage(key.key, { costCents: 15 }); // after an LLM call

Budget enforcement is atomic, so concurrent agent calls can't race past the cap — which matters, because agents fire requests in parallel. There's also Express middleware if you want it:

app.post('/api/proxy', agentKeyMiddleware(ak, { scope: 'proxy.chat' }), handler);

What it's not

It's small and focused, extracted from a real production key system, MIT-licensed. It isn't trying to be Clerk or Auth0. If you already have a keys table and you want per-key spend caps without building a quota system, it's a few columns and a function call.


npm: @katrinalaszlo/agentkey · GitHub: katrinalaszlo/agentkey