MCP server

Lists the websites this connection can read, with their id, domain, content language, Search Console connection status and the state of the latest crawl. Call this first to get the website_id that the other tools need.

This tool takes no arguments.

Lists the SEO opportunities Lastmod found on a website: what to improve, on which page, with priority, effort, a confidence score of the advice (higher is more certain; it is an estimate, not a measurement), the reason and the recommended action. Without `status` the list leaves out RESOLVED and VOIDED opportunities; pass status ALL to get every status in one list. A page holds at most 200 opportunities: when `meta.truncated` is true, call again with `cursor` set to `meta.nextCursor` until it is null. The list has no evidence; call get_opportunity for that.

Returns one opportunity in detail, including the evidence Lastmod recorded when it detected it and, when it was resolved, how and when. A site-wide opportunity lists at most 100 pages in its evidence rows; the summary row carries `pagesAffected` (all affected pages) and `truncated`. When `pagesAffected` is above the number of page rows, say so (e.g. '100 of 178 shown'); the remaining pages are not available through this tool.

Lists the tasks of a website (created from opportunities): title, reason, expected impact, priority, status, the page, the measurement state and step progress, newest first. Optionally filter by status. A page holds at most 200 tasks: when `meta.truncated` is true, call again with `cursor` set to `meta.nextCursor` until it is null.

Lists the measured results of finished tasks, newest first, cursor-paginated: the metric before and after the change, the relative change, and a confidence band of the measurement. A result is a measurement next to other changes, not proof of cause; when other changes overlapped, 'confounded' is true and the reasons are listed. Pass meta.nextCursor as cursor for the next page.

Returns the dashboard KPI groups of a website for a period: opportunity inventory, tasks completed, changes in monitoring, measured results by confidence band, and search presence. Counts of results describe measurements, not proven causes.

Returns the Search Console performance of a website over time (clicks, impressions, CTR, position) with markers for the changes made in that period. Markers show when a change was made; they do not show that the change caused the movement.

Returns how often AI engines mention a website and cite its pages over a recent window, per engine: counts of answers with an uncertainty interval and a reliability band. When there is too little data the result is INSUFFICIENT_DATA and carries no figure. An engine that could not be measured is listed under unavailableEngines; that is not a zero. The previous period is a loose observation, not a comparison. A measurement is not proof of cause.

Lists the completed measurement rounds of a website's AI visibility: when each round ran and how many answers it counted per engine. Descriptive only; it does not rank rounds or say whether visibility went up or down.

Lists the questions Lastmod asks AI engines for a website (status, source, whether the text contains the brand, theme) with the active count and the limit. Retired questions are not included. Read-only: prompts are managed in the Lastmod dashboard.

Returns the themes where AI engines answered without mentioning the website, with the kinds of sources they cited and the linked opportunity or task. Source hosts appear only when they recur across answers. A theme gap is an observation about the answers seen, not a ranking.

Returns the visits that reached a website from AI assistants, from Google Analytics 4, per source, for the last 7, 28 or 90 days (default 28). The status says whether Analytics is connected; without a connection nothing is measured, which is not zero visits.