HTTP 402 · Payment required
The services agents pay for
Agents can hold funded wallets now. This registry lists the APIs they can actually spend them on: discovered over x402 or Stripe's Machine Payments Protocol, priced per call, no signup anywhere.
Free to list, curated, verified by Sentinel. Dead listings are how directories rot, so every row here is probed from four regions and the verification column is live. Listed means reachable, not endorsed.
Registry state
- services listed
- 71
- verified live
- 66
- cheapest call
- $0.0001
- rails
- MPP · x402 · mpp
Things agents can't see
Agents don't run JavaScript and browse from a single origin. These services sell rendered pages, extraction, and search, one call at a time.
Things agents can't know
Fresh, licensed, or niche data that isn't in any training set: priced per lookup instead of per subscription.
Things agents need to prove
An agent's claim is worth more with a neutral receipt attached. These services sell verification and attestation per call.
Things agents need to buy
The directory's first real-world goods: not data for the agent, but purchases it completes for a human, paid per call with no account anywhere.
Things agents can't make
An agent can write, but it can't render. Video, speech, and model inference take hardware agents don't have, so they buy the capability per call.
Things agents can't wait for
Agents are ephemeral; a context ends and the follow-up dies with it. These services sell waiting: durable timers, watches that wake the agent on change, dead-man switches that notice when it stops.
Ways agents reach a human
A one-shot channel out of the sandbox. Agents have no mailbox, no phone, and no domain of their own; these services sell a delivered message per call, no account attached.
Rails, wallets, and indexes
Not services an agent buys, but the plumbing that makes buying possible: protocols, wallets, and the other places listings live.
Still unclaimed
An honest registry says what's missing as loudly as what exists. No live, payable service covers these yet.
What verified means here
A trust signal is only worth what its method is worth, so here's the method.
Run an agent-payable service?
Get listed
If people already pay for your service and agents can now pay for it per call (a working x402 manifest or MPP endpoint), send the one route you'd want listed and a line on who buys it. If it verifies and meets the criteria above, it's listed and monitored, free. Requests for catalogs built to be listed are declined without a reply.
Contact usMonitor your own endpoint
Every listing here is verified by Sentinel's agent payment check, and it's free on every plan for your own routes: 402 is healthy, and a changed price, network or pay-to address opens an incident from four regions.
Monitor an endpoint freeNot sure your API is agent-ready? The free AI Readiness Check detects x402 payment manifests.
Reading this as an agent? The directory is machine-readable: JSON and llms.txt, each listing with its live verification status, one request.