Odaq Bot

Odaq is an AI marketing platform operated by Digital Peax OÜ. Our members ask Odaq to look at websites, mostly their own. This page explains what Odaq reads, how it introduces itself, which rules it follows, and how you can allow, slow down, or block it.

1. How Odaq introduces itself

Odaq's readers that name themselves send one of these two user agents:

These named readers sign their requests using Web Bot Auth, so a website can verify that the request came from Odaq. Our public key directory publishes the verification keys. Signing may be temporarily disabled during maintenance. A user agent alone is not proof of identity.

Other readers look at a member's own website the way a visitor's browser would, so they see what visitors see: UX audits, screenshots, brand scans, and reading a member's products and home page. Those use one ordinary browser identity (Chrome on a desktop, Safari on a phone) and do not carry the Odaq name. So does the browser Odaq opens for a page that stays empty until JavaScript runs. These visitor-view requests and reads made through external services do not carry Odaq signatures.

2. Rules Odaq follows

• Finding pages. Odaq-Bot reads your robots.txt and sitemaps and, when the sitemap does not list enough pages, follows links on a member's website. While it follows links, it does not follow links into paths your robots.txt disallows for Odaq or for all crawlers, and it waits the Crawl-delay you set between requests. A member cannot switch this off. Some requests are not covered: Odaq still requests your home page to start, it opens a link that redirects into a disallowed path before it can check it, and it records pages your sitemap lists. When the sitemap has no language (hreflang) links, Odaq also opens up to 50 listed pages, ten at a time with a short fixed pause.
• Research for articles. Scribe reads a page that a web search returned only if the site's robots.txt allows Odaq to read it. If the page redirects, Scribe checks robots.txt for the address it lands on too and discards what it read if that is not allowed, though the redirected request has already been made. It treats a robots.txt rule that shuts out well-known AI agents (such as GPTBot or ClaudeBot) as a no, and it skips pages marked noai or carrying a text-and-data-mining reservation.
• Pages a web search found. When an Odaq agent opens a page that a web search returned, for example while researching competitors, it reads that page only if your robots.txt allows it for Odaq: a rule for Odaq, for Odaq-Bot, or for all crawlers. A page your rules close is not requested at all, and if your robots.txt cannot be reached Odaq reads nothing from the site. Unlike Scribe, these reads do not treat a rule that shuts out other AI agents, or a noai marker, as a no. Odaq checks the address the search returned, not where it may redirect. In the SEO agent's chat, a page on another site is checked the same way unless the member typed that site's address or it is the member's own website.
• Pages a member asks about. When a member asks an Odaq agent to look at a page, for example by pasting its address, Odaq reads that one page as Odaq-Bot on the member's behalf. These reads do not check robots.txt.
• Links a member pastes. When a member gives Scribe a link, Scribe reads that one page as the member's browser would.
• A member's own website. When a member asks Odaq to check their own website, for example the structured-data check, Odaq reads the pages the member points it to, on the member's behalf. robots.txt rules guide which pages Odaq discovers, but they do not stop a page the member asked it to check. If a site refuses Odaq, Odaq may try to read that page through a service it uses for the purpose (see our Sub-processors).

3. Allow Odaq

Bot protection and firewalls often block crawlers they do not know. Prefer a rule that verifies Odaq's Web Bot Auth signature against our public key directory. Signing does not automatically put Odaq on a provider's verified-bot list; provider registration is a separate step. If signature verification is not available, a narrowly scoped allow rule for a user agent containing Odaq is a fallback. Anyone can copy a user agent, so keep the rule as narrow as you can and allow only the protection that blocks Odaq. Odaq members can find step-by-step instructions in the Help Center under "Let Odaq read your website".

Screenshots, UX audits, brand scans, and product reads use an ordinary browser identity, so they cannot be allowed by name. If your protection blocks them, write to us at info@odaq.ai.

4. Slow down or block Odaq

To ask Odaq to wait between requests while it follows links, add this to your robots.txt:

User-agent: Odaq
Crawl-delay: 10

To ask Odaq not to follow links into your site, and to stop its article research and its agents from reading pages a web search found on it:

User-agent: Odaq
Disallow: /

These rules do not stop the requests listed in section 2 as not covered, a check a member runs on their own website, or a page a member asks an agent to look at. If you want any of these to stop, write to us. Blocking the user agents in section 1 in your firewall may not be enough, because a refused page of a member's own site can be read through a service Odaq uses.

5. Contact

Questions, or a request about how Odaq reads your site? Write to us and tell us the website. Related documents: our Acceptable Use Policy and Privacy Policy.