Monetize your APIs
for AI Agents and Bots

A low-code middleware framework that lets developers charge AI agents micropayments in crypto to access protected routes. Drop it into Express or Next.js in under 2 minutes.

Terminal

How it works

The x402 payment loop — from first request to verified access.

AI Agent / Bot

Crawling content or querying an endpoint

1GET /api/scraped-data
2402 Payment Required
3USDC Payment (on-chain)
4GET with X-Payment-Signature
5200 OK + Payload Released

Your API + Centinel

Validates on-chain signatures & manages sessions

Built for production

Everything you need for machine-to-machine micropayments, out of the box.

Replay protection

In-memory signature deduplication and on-chain transaction age checks prevent double-spending. Each signature can only be used once.

Multi-chain, multi-token

Accepts USDC, SOL, and ETH across Solana and Base L2. Agents pay with whatever token they have — you get paid either way.

Flexible billing

Choose per_request for pay-as-you-go access or per_session for timed JWT tokens. Set different prices per route.

Rate limiting

Built-in per-IP rate limiter blocks signature spamming after 5 failed attempts, protecting your RPC nodes from abuse.

Zero config

The CLI auto-detects your framework, language, and directory structure. One command scaffolds everything you need.

Edge-ready

Next.js middleware runs at the edge with Web Crypto — no cold starts, no Node.js SDK dependencies. Works on Vercel out of the box.

Who is this for?

If bots are consuming your resources for free, Centinel lets you charge them.

Publishers & Bloggers

AI crawlers like GPTBot and ClaudeBot are training on your articles, recipes, and reporting. Centinel lets you charge them per-page instead of blocking them entirely with robots.txt.

API & SaaS Providers

Instead of requiring agents to sign up, enter a credit card, and commit to subscriptions, charge them fractions of a cent per call. No accounts. No onboarding friction.

AI Compute Hosts

Running GPU models or fine-tuned nets? Charge executing agents exactly what the inference costs. Per-request pricing that matches your actual compute spend.