The connect0 fetcher (connect0-agent)
Who sends requests as connect0-agent, why your site sees them, how to tell a genuine request from an impostor, and how to limit or stop them.
Updated Oct 24, 2026
If your logs show requests with this user agent, they come from connect0:
connect0-agent/1.0 (+https://connect0.ai/bot)
connect0 runs AI agents for its customers. When one of those agents needs to read a public page — a docs page, a changelog, a price list — it asks connect0's fetcher, and the fetcher requests that one page on the agent's behalf. Each request is a person's or an agent's specific call with a specific URL. The fetcher does not crawl, does not follow the links on a page, and does not run the page's JavaScript.
What it does on your site
- One page per call. A redirect is followed (up to five), and each hop is checked again as if it were the first URL.
- robots.txt first. Before a page is requested, your
robots.txtis read (and cached for 24 hours). A path you disallow forconnect0-agent, or for*, is never requested. - HTTPS only. Plain
http://addresses are refused, and so is any private, loopback or metadata address. - Short cache. A page it fetched recently may be served from a cache
of about 15 minutes instead of being requested again. A response with
cookies or
Cache-Control: privateis never cached. - GET only. It never submits a form, logs in or writes anything.
How to tell a genuine request
Every request is signed with Web Bot Auth (HTTP Message Signatures,
RFC 9421). The Signature-Agent header points at our key directory:
https://connect0.ai/.well-known/http-message-signatures-directory
A request that carries our user agent but no valid signature from a key in that directory did not come from connect0. If your CDN supports verified bots (Cloudflare does), it can check this for you.
How to limit or stop it
Add a group for our token to your robots.txt. To stop it everywhere:
User-agent: connect0-agent
Disallow: /
To keep it out of one part of the site:
User-agent: connect0-agent
Disallow: /private/
The change applies within a day (the robots cache). The fetcher does
not read Crawl-delay yet; if our requests are a load problem for you,
write to us and we will act on it directly.
Contact
Write to hi@connect0.ai with the URL, the time
and, if you have it, the request's Signature-Agent header. We answer
every message about the fetcher.
What this does not do
- It fetches one page per request an agent makes; it does not crawl or follow links on its own.
- It honours robots.txt; Crawl-delay is not read yet.
- It runs on Cloudflare's network, so there is no fixed IP range to allow-list.
Ask Zero
Ask a question about connect0 and get an answer grounded in the docs, with links to the sources. Signed in? Zero answers with your account in mind.