For site owners and developers

EchobatBot

EchobatBot is the scanner behind Echobat: it reads a local business’s public website the way AI assistants read it, so the business can see what they see.

  • What it does

    Echobat tells a local business how its website looks to AI assistants such as ChatGPT, Gemini, Perplexity, Claude, Copilot, Google’s AI answers, Siri and Alexa. EchobatBot reads robots.txt first, then your homepage and up to 14 more pages linked from it or listed in your sitemap, so at most 15 pages a scan. The result goes to the person who asked for the scan.

  • What it never does

    • Never fills in forms or logs in.
    • Never goes looking for personal data: no customer records, no accounts, nothing behind a password. It reads only the public pages anyone can open.
    • Never runs your JavaScript, solves challenges or CAPTCHAs, or pretends to be a browser.
    • Always follows robots.txt.
  • How often it visits

    • When someone asks, almost always the business itself, typing its address into our scanner.
    • Once a month for a paying customer’s own site.
    • Not twice in 24 hours for the same site.
    • One request at a time, at least 1 second apart.

The user agent

Every request EchobatBot makes carries exactly this, so you can find it in your logs or match it in a rule.

User-Agent
EchobatBot/1.0 (+https://echobat-ai.com/bot)

robots.txt

We read robots.txt before anything else and do what it says. We follow the group written for EchobatBot if there is one, and the User-agent: * group otherwise. Rules written for our old name, EchobatScanner, still count. If the file answers with a server error, we treat that as “no” and stop. A Crawl-delay is honored up to 30 seconds, and we read fewer pages rather than stay longer.

Allow EchobatBot
User-agent: EchobatBot
Allow: /
Block EchobatBot
User-agent: EchobatBot
Disallow: /

A block takes effect on our next visit. Or email support@echobat-ai.com with the domain and we will stop scanning it for everyone, including the business itself. A person reads it.

How to verify it’s really us

Anyone can type our name into a request. So EchobatBot also signs its requests with Web Bot Auth: a digital signature made with a private key only we hold. The matching public key is published on our own site, so a firewall that checks signatures can confirm that a request saying “EchobatBot” really came from us and was not changed on the way.

For developers

Requests are signed with HTTP Message Signatures (RFC 9421). A signed request carries Signature-Agent: "https://echobat-ai.com" plus Signature-Input and Signature headers. The key directory is:

Key directory
https://echobat-ai.com/.well-known/http-message-signatures-directory

Current key id: jhioZekf28kMAtZDLHSfAvj0n_u3FMGF1sHV5pE2iAw

If your server answers 429 or 503, we back off. With Retry-After we wait that long; over 30 seconds, we stop that scan.

If your firewall turns it away

Nothing to do on most sites. On Cloudflare, open Security, then WAF, then Custom rules, and create a rule with the expression (http.user_agent contains "EchobatBot") and the action Skip, placed first. Bot Fight Mode cannot be skipped by a rule, so if that is what blocks us, the choice is to switch it off or leave us out. Anywhere else, ask whoever runs your firewall to allow the user agent EchobatBot. Step-by-step versions are on About the Echobat scanner.

One more visitor you may see: to measure speed, a scan asks Google PageSpeed Insights to test your homepage. That visit comes from Google’s servers with Google’s own user agent (it contains Chrome-Lighthouse), not ours.

Site owners: if a scan caused a problem, or your logs show anything heavier than this page describes, write to support@echobat-ai.com.
Echobat, P.O. Box 24381, Winston-Salem, NC 27114