Fetch domain configs

When you generate content from a web address — either the Generate content feature or the inline Generate from URL action — the plugin fetches the page and pulls out the product description. On a site it has never seen, it falls back to a best-guess extraction. A fetch domain config removes the guesswork: it tells the fetcher exactly which part of a given website holds the description, so the text handed to the AI is clean.

You usually only need this for sites where the automatic extraction brings back menus, related products or other clutter.

How to use

Open the list

Go to Catalogues → AI Assistant.
On the Dashboard, click the Fetch domain configs quick-start tile.
Click Create config (or Edit on an existing row).

Let the AI propose the selectors

The wizard runs in three steps:

Source URL — paste a sample Example product URL from the website you want to configure, pick an AI connection, and click Analyze with AI. The AI reads the page and proposes the selectors that isolate the product description, along with its Confidence and AI reasoning.
Review & configure — check and adjust the proposal:
- Host — the website this config applies to (e.g. www.example.com).
- Path pattern (optional) — limit the config to part of the site (e.g. /product/*). Leave empty to cover every page on the host.
- Description selector — the part of the page that holds the product description.
- Product name selector (optional) — used when generating several fields at once.
- Block selectors (optional) — extra page noise to strip out before extraction (e.g. related products, reviews), one per line.
- Enabled — when off, the config is ignored and the fetcher falls back to automatic extraction.
- Advanced · Network settings (optional) — per-website overrides for the URL fetcher (response size, timeout, redirects, User-Agent, proxy). Leave any field empty to use the global default.
Extraction preview — click Test extraction to fetch the page with your selectors and confirm the right text comes back, then Save.