The crawler behind the business index.
What it sends, how it reads robots.txt, how fast it asks, and how to keep your site out.
01User agents
The exact strings we send.
The crawler that builds the business index:
FrontlatchBot/0.1 (builds the Sydney agent-action index, read-only; +https://frontlatch.com/docs/agents)
The jobs that seed the index from public datasets:
FrontlatchBot/0.1 (+https://frontlatch.com/bot; index seed from OpenStreetMap) FrontlatchBot/0.1 (+https://frontlatch.com/bot; Common Crawl host graph) FrontlatchBot/0.1 (+https://frontlatch.com/bot; AU index ingest)
The scanner, when someone asks for a scan or an AI assistant reads a page through the gateway, and the browser that maps a site’s forms (it never submits them):
FrontlatchScanner/0.1 (read-only agent-readiness scan; +contact: <address>) <browser user agent> FrontlatchMapper/0.1 (renders pages, fills forms to trigger client-side validation, NEVER submits; +contact: <address>)
02robots.txt
We read robots.txt before every site.
Each agent looks for a group naming it (FrontlatchBot, FrontlatchScanner or FrontlatchMapper) and otherwise follows the * group. A group named Frontlatch applies to all three. The longest matching rule wins, and an Allow beats a Disallow of the same length. If robots.txt disallows a page, the crawler does not fetch it. Crawl-delay is honoured.
03Crawl rate
How fast we ask any one host.
- FrontlatchBot: at most one request per second to a host, or slower if your
Crawl-delaysays so, and at most 8 pages of a business’s site per visit. - FrontlatchScanner: at least half a second between requests to a host, or your
Crawl-delayif longer, and at most 15 pages per scan. - FrontlatchMapper: at least 2.5 seconds between page loads on a host.
- Every agent is read-only. None of them signs in, buys, or submits a form.
04Opting out
How to keep a site out.
To stop all three agents, add this to your site’s robots.txt:
User-agent: Frontlatch Disallow: /
To remove a listing that is already in the index, use the remove link on the business’s own page in the business index: we email a code to an address at that domain, and once it is entered the listing is hidden and never added again. Or write to [email protected] with the domain.