Feature request: a search-index enricher (domain -> indexed pages) backed by Serply

Hi @dextmorgn,

I have been reading through flowsint-enrichers and the enricher-builder skill, and the one input pivot I cannot find anywhere in the catalog is a search index. There are 48 enrichers and none of them asks a search engine what is publicly indexed about a target. The closest are website_to_crawler and website_to_links, but both start from a URL you already have, so they widen a page rather than discover one.

I would like to add that pivot. Before writing anything I want to check whether you want it at all, and which direction you want first.

What the service is

Serply is a keyed API that returns Google results as JSON. One call:

GET https://api.serply.io/v1/search/?q=site%3Aflowsint.io&num=10
X-Api-Key: <key>
Accept: application/json

The response carries a results array of {position, title, link, description} rows, plus optional blocks for answers, related searches and news when Google returns them. I ran exactly that query today: HTTP 200, 9 organic rows, each with a title, a link and a description.

1  Flowsint - Automated graph analysis & intelligence   https://www.flowsint.io/
2  Welcome to Flowsint - Documentation                  https://www.flowsint.io/docs/overview
3  Enrichers catalog - Documentation | flowsint.io      https://www.flowsint.io/docs/sources/available-enrichers

Billing is per request on a self-serve key, so operationally it is the same shape as the keyed services already in the catalog (Dehashed, Scamalytics, Whoxy, Veriphone, HudsonRock, Etherscan) rather than a new kind of dependency.

What I would build

flowsint-enrichers/src/flowsint_enrichers/domain/to_indexed_pages.py:

scan() runs site:<domain> and pages through the result set; postprocess() emits one Website node per organic row and relates it back to the input Domain. The mapping needs no new type: link lands on Website.url (the primary field), title on title, description on description.

The value is that it surfaces pages DNS, certificate transparency and crawling all miss: hosts that only appear in Google's index, stale pages that still resolve but are linked from nowhere, exposed documents, and third-party pages that mention the target.

Which pivot do you want first

I sketched three, and I would rather build the one you want than the one I happened to pick:

1. domain -> Website from site:<domain>, the one above. It slots straight into the existing domain flows.
2. individual -> Website or organization -> Website from a plain name query, for pages that mention a subject rather than pages the subject owns.
3. organization -> Website restricted to news. Google's tbm=nws passes through, so this is the same adapter with one extra parameter.

Two things I would get right in the first push

• Pagination. start is the only parameter Serply honours. page, offset and p are accepted and silently ignored, so a loop built on any of those refetches the first window forever and spends a credit per duplicate page. I verified this: start=10 returns a window with no overlap against the base page, while the others return page one verbatim.
• num is an approximate cap of about 10 per page, not an exact count, so surplus rows need trimming on our side before they become nodes.

Disclosure, and the offer

I work on Serply, so weigh the recommendation accordingly. I am proposing it because the pivot is missing, not because the catalog needs another vendor.

If you want it, I will open a PR with the enricher, tests under flowsint-enrichers/tests/enrichers/, a row in docs/sources/available-enrichers.mdx, make lint / make typecheck / uv run pytest -q green, and a screenshot of the nodes it produces on a real sketch. Provider docs: https://serply.io/docs

If a search-engine source is out of scope for the catalog, or if you would rather it arrive as a configurable template enricher than as a Python module, say which and I will follow it.