list_projects

Cost: Free (0 credits).

list_projects returns the domains you're tracking, oldest first, each with its project_id, its Search Console state and its last background job — and, below them, anything you have archived. If you have neither yet, it points you to setup_project.

What it does

Reads your projects, scoped to your account, and returns them as two lists: the ones you are tracking, and the ones you archived with untrack_project.

What each tracked line tells you

Beside the domain and its project_id, every tracked line carries two facts, because a list of fifteen identical-looking domains answers neither question a customer actually has: can Search Console be read for this one, and has anything ever run against it.

Search Console is reported in three states, never as a tick:

• not connected — no Google account is linked to this project.
• connected, no property selected — an account is linked but no property has been matched to it. Nothing can be pulled yet. This state is named rather than folded into "connected" precisely because it looks like a working connection and is not.
• the property itself (for example sc-domain:example.com) — the mapping that pulls will use. If the stored credential behind it has died, (reconnect needed) is appended: the property is still right, the credential is not.

Last job names the tool that ran and the day it ran. A job is a background run — crawl_site or pull_gsc_data. Audits, keyword and backlink lookups run synchronously and are not jobs, so a project can read none yet and still have been analysed — the reply says so, under the list, rather than leaving the line to be read as "nothing has ever happened here".

If two tracked projects are mapped to the same Search Console property, the reply names that property and both projects underneath the list. Each pull fetches one set of rows and is billed once per project, so the pair costs credits twice for the same data. Nothing is deduplicated for you — which of the two to keep is a decision only you can make.

If two tracked projects are the same site — an apex and its www. twin, say — the reply names them together as well. This is a different warning and it fires independently: a pair with no Google account at all still has it, because each of the two is crawled, audited and billed on its own. New projects can no longer split this way, so a pair you see is left over from before that changed; untrack_project archives the one you do not want, and nothing it holds is deleted.

You do not always need the project_id

The project_id is what most tools use to know which of your sites they are working on, and this is where you get it. But a large part of the surface will also take a plain domain instead, passed as target — including a competitor's, which is the point: those tools answer questions about any public site, not only your own.

Pass target OR project_id, never both. These tools accept either: ai_visibility, analyze_backlinks, backlink_changes, backlink_details, compare_competitors, disavow_candidates, discover_keywords, keyword_gap, keyword_positions, link_gap, my_pages, ranked_keywords, serp_snapshot.

Everything else needs the project_id, because it reads data stored against YOUR project — your crawls, your audits, your Search Console pulls — and a domain does not identify those.

The archive

Archived projects are listed separately, most recently archived first, each with its project_id and the date it was archived. They are not counted as tracked and never appear in the tracked list — but they are visible, which is what makes them recoverable: you do not have to remember the exact domain to bring one back.

Either route restores the original project in place, on the same project_id, with its crawls, reports and Search Console connection intact: setup_project with the same domain, or track_gsc_property for its property.

Example

Ask your MCP client in plain language:

Which sites am I tracking?

or

What did I archive?

Input

No parameters.

Returns

One line per tracked project (domain, project_id, Search Console state, last job), oldest first, followed by one line saying what counts as a job; then an archive section, most recently archived first, with each archived project's domain, project_id, archive date, and how to restore it. Guidance to create your first project when you have nothing at all.