About ZeroBot
Zero is a search engine for machine-payable services. ZeroBot is how we check that the services in that index are still there and still work, so agents aren't sent to dead or broken endpoints. If it reached you, this page explains exactly what it did and how to stop it.
Mozilla/5.0 (compatible; ZeroBot/1.0; +https://zero.xyz/bot)
- Discovery. Reads public service directories and manifests (
/.well-known/mpp, OpenAPI documents) that services publish about themselves. - Liveness probes. A small, periodic request to an indexed endpoint to see whether it still answers and still returns its payment challenge. Most of our traffic is this, and most of it is answered with an HTTP 402 — no work is done on your side.
- Quality checks. For some endpoints we make a real, paid call so we can report to agents whether the service actually works. These are ordinary paying requests and settle like any other.
Match on the ZeroBot token in the User-Agent. Every automated request we make carries it, and it's the only identifier we control end to end.
Please don't match on IP. Our egress addresses sit in a shared Render (Oregon) range that we don't own: they change without notice, and most of the surrounding block belongs to other tenants. An allowlist built on them will silently start blocking us — or start trusting someone else. If your infrastructure genuinely requires an IP allowlist, email us and we'll give you the current set and tell you when it changes — the address is at the bottom of this page.
Traffic from an agent that is using Zero to pay you is not ZeroBot. Those requests belong to that user and identify themselves as zero-sdk/<version> or zero-cli/<version>. Rate-limiting them is rate-limiting a paying customer.
- Requests are spread across hosts rather than issued back-to-back against one provider.
- Every request is bounded by a short timeout, retries are backed off, and response bodies are read only up to a small cap.
- How often we probe an endpoint scales with how much it's used: at most every few minutes for the busiest, at most hourly for the rest, plus a rolling sweep across the wider catalog every six hours. Never continuously.
- We fetch documents a service publishes about itself. We do not crawl sites for content, and we don't try to work around blocks.
Want us to slow down, stop probing a specific endpoint, or remove your service from the index entirely? Email support@zeroclick.ai with the host and we'll action it — no justification needed.
Blocking or rate-limiting the User-Agent works too, and we won't route around it. Note that it also removes your service from what agents searching Zero can discover and buy.