AI workers for Google Search Console

What is an AI worker for Google Search Console?

An AI worker for Google Search Console is an AI agent scoped to one reporting job on your verified properties, such as tracking which queries gained or lost position or noticing a page whose impressions collapsed. It reads performance data through your connected Google account. This is a reporting surface, so a worker here observes and explains rather than changing anything about your site.

What these workers do

Report the week that just happened

Compares this period against the last and names the queries and pages that moved, in both directions, rather than reporting a total.

Catch a page losing ground early

Watches for pages whose impressions or position fall away and raises them while the cause is still recent enough to identify.

Surface queries you rank for by accident

Finds the terms bringing traffic that nobody planned for, which is usually where the next piece of content should go.

Track a launch after it ships

Follows how new pages are picked up over the weeks after publication and reports whether they are gaining or flat.

How an AI worker runs in Google Search Console

Reading, not acting

Movement matters more than totals

Reports that name pages and queries

Google Search Console questions, answered.

Can an AI worker submit pages for indexing?

No. This connection is for reporting only, so workers read performance data and cannot request indexing, submit URLs, or otherwise act on your site. Anything that changes what Google crawls remains a deliberate human action taken elsewhere.

Which properties can it read?

The verified properties your connected Google account has access to, narrowed to whichever ones you scope the worker to. A property left out of scope is not readable by it, so an agency or in-house team holding several properties can give each one its own worker without any of them seeing the others.

How current is the data it reports?

It reads what Search Console has published at the time it runs, which lags live traffic by a couple of days as Google's own reporting does. A weekly worker is comparing settled periods, so the lag does not distort the comparison.

Will it change anything about my site?

No. There is no write path on this connection at all, by design. A worker here observes, compares, and reports, which makes it one of the safest connections to grant and one of the easiest to leave running unattended, since the worst case of a mistaken run is a report you disagree with.

What AI workers can do in Google Search Console

The tools a worker can be granted in Google Search Console once you connect it. A kit grants only the ones its job needs, and you can narrow them further on the worker.

List Properties

List the Search Console properties the connected Google account can access. Each carries its `siteUrl` (use as siteUrl for the other tools) and `permissionLevel`. An empty `sites` list with a non-null `accessInfo` means a token rule limited the results, not that no properties exist; render `accessInfo` verbatim.

Query Search Analytics

Read search-performance rows (clicks, impressions, ctr, position) for one property over a date range, grouped by the requested dimensions. `keys` in each row is positional and matches the `dimensions` order. `ctr` is a ratio (0.0304 = 3.04%). Dates are in the property's Search Console time zone and lag 2-3 days. If the token is scoped to specific URL paths, include `page` in dimensions or the response is empty with an `accessInfo` saying so; render `accessInfo` verbatim.

List Sitemaps

List the sitemaps submitted for a property, each with its `path` (use as feedPath for search_console_get_sitemap), submission/download times, pending state, warning/error counts, and per-type submitted/indexed totals.

Get Sitemap

Get one sitemap's submission and processing status by its full feed URL. Returns null with a 404 when the feed path is not found for the property; confirm the path against search_console_list_sitemaps.

Inspect URL

Read Google's current index status for one URL within a property: the headline `verdict` (PASS/PARTIAL/FAIL/NEUTRAL), `coverageState`, crawl and canonical detail, and mobile-usability and rich-results verdicts. This reports status only; it cannot request or trigger (re)indexing. The inspected URL must belong to the named property, otherwise the result is empty with an `accessInfo`; render `accessInfo` verbatim.