NordikResearchBot: what it fetches and how to block it
If NordikResearchBot is in your server logs, this page says what it loads and how to stop it: one rule in robots.txt, or one email.
Who runs it
Nordik Audit, a code audit service operated by one owner with AI software, runs the three crawlers on this page. Each one sends a User-Agent that starts with NordikResearchBot/1.0.
Contact is by email only: outreach@getnordik.com.
What each crawler fetches
| Crawler | What it loads | Pace and limits |
|---|---|---|
| Prospect reader | Public pages of a product's site that appeared in a public launch listing: the home page and at most two more pages the home page links to (a contact or pricing page, and one legal or about page read only for the country). It notes what the pages show about the product and, where one is published, a business contact address on the site's own domain. | At least 2 seconds between two requests to one site. GET only: no request body, no cookie, no credential. No API, database, login, form, source map or guessed path. |
| Launch Risk Score scan | One page, when a person asks for it. Without proof that the person controls the site, the free scan runs a quick scan that requests only what a visitor's browser loads: the page, the scripts the page loads and the robots.txt file. It sends no well-known path and no Supabase request. The full scan runs only after the owner shows the site is theirs by placing a one-time token on it, which is read back first, or on the address a paying customer gave for an order. It loads the page, the scripts the page loads, robots.txt and a few well-known paths. Supabase checks use only the public anon key the site itself ships: row-count requests that return a number, not data, and two read-only settings requests. |
It waits at least one second between requests to the same host. A quick scan makes at most 30 requests in all, a full scan at most 60. It never logs in, writes, submits a form or reads a row of data, and it never uses a key it finds. |
| Intent watcher | Public JSON from three fixed services, never a company site: Freelancer's public project listing, Hacker News search and the Hacker News API. | At least 1.5 seconds between two requests to one host, and at most 24 requests in one run, robots.txt included. Its scheduled runs are daily. GET only: no request body, no cookie, no credential, and a redirect is refused, not followed. |
Block it with robots.txt
The robots.txt token is NordikResearchBot. These two lines keep all three crawlers off every page of a site:
User-agent: NordikResearchBot Disallow: /
- Which group is read. The group for
NordikResearchBotwhen the file has one, otherwise the*group. - A Disallow is honoured. A path the group disallows is not requested.
- A file that cannot be read stops the visit. When
robots.txtanswers with a server error or cannot be reached, nothing else is requested from that site in that run. - Two requests are outside robots.txt, and both happen only on a site owner's own request, never in a quick scan: the free scan reads back the token its owner placed, and the Supabase checks go to the Supabase project's own address, not to the site.
Opt out by email
Write to outreach@getnordik.com with the address of your site. Nordik adds the site to its opt-out list within 10 business days and replies to say so.
The prospect reader never requests a site on that list again, not even its robots.txt. An entry covers the site's www address and its subdomains.
The list covers the prospect reader and the free scan's quick scan, the one that runs without proof of control: neither requests a listed site. A scan whose owner proved control of the site, and the scan of an order, run on the site owner's own request, so the list does not cover them. The intent watcher reads no company site.
Contact
outreach@getnordik.com. Email only.
What this page does not claim
- The prospect reader and the intent watcher are not a security test of your site. They send no login, form, key or guessed path.
- The Launch Risk Score scan is a passive check. Anyone can ask for its quick scan, which requests only what the site sends to every visitor; the checks beyond that run only for the site's owner. A score of 100 means no verified finding among the checks that were completed; it is not a claim that the site is secure.
- No AI model reads your pages in the prospect reader, the intent watcher or the free scan. They are plain software.