Overview
AgentPay sits in front of your API, collects payment from agents and API clients, and forwards paid requests to your backend.
How it works
Agent or clientCalls your AgentPay URL
AgentPayNegotiates and verifies payment
Your APIReceives a normal HTTP request
You configure projects, payment channels, and endpoints in the dashboard. Clients call the generated AgentPay URLs instead of your backend. A request without valid payment gets an HTTP 402 response that tells the client how to pay. A paid request is forwarded to your backend, and your response is returned to the client.
Why AgentPay
- One paid entry point. Publish payment-protected URLs without exposing your backend URL or changing your API.
- Several ways to pay. Offer stablecoin, testnet, token, and card-backed payments from the same project.
- Built for agents. A generated OpenAPI document tells agents which routes exist, what they cost, and how to pay for them.
Payment protocols
Each project can enable any combination of these channels. See Choosing a protocol for a side-by-side comparison.
| Protocol | How clients pay | Plans |
|---|---|---|
| x402 | Onchain USDC or ETH | Paid |
| x402 Testnet | Onchain, on test networks | Free and paid |
| MPP Tempo | Tokens on Tempo | Paid |
| MPP Tempo Testnet | Test tokens on Tempo testnet | Free and paid |
| MPP Stripe | Cards | Paid |
Next steps
QuickstartCreate a project and make your first paid request.How requests workWhat happens between the client, AgentPay, and your backend.Batch settlementLet clients prepay once and pay per request without an onchain transaction each time.Agent setupGive agents what they need to discover and pay for your API.