Every fix arrives written out in full.
You do not need to know what a canonical tag is. You need to decide whether to apply a change, which means the change has to be in front of you, in full, before you decide.
What a queue item contains
Not a description of a change. The change: what is removed, what is added, with line numbers and context. Which check it satisfies. Which pages it applies to. Whether a machine wrote it, drafted it, or declined to.
Who is allowed to fix what
Auto
Findings with one correct answer: a missing description, an absent canonical, a malformed sitemap entry, alt text where the surrounding content makes it obvious. The engine writes it.
You still review it. Auto describes who wrote it, not who applies it.
Assist
The shape of the fix is clear but the content is a judgement: shortening a title without losing what it meant, merging two posts that compete for one query.
The draft arrives with reasoning attached. Editing it before you apply it is the expected path.
Human
Whether a page should exist, whether a claim is true, whether an author is an expert.
The engine reports the finding and stops rather than guessing convincingly.
How the queue is ordered
Severity times pages affected. A low-severity finding across 200 pages outranks a high-severity one on a page nobody visits, because fixing it is worth more.
The two review points
You review the ranked plan before anything is written. You mark the fixes ready before the finished list is handed to you. Both review points are always there. Crawld writes; you decide and you apply. Because you apply each change yourself, undoing one is the same small edit in reverse.
How changes reach your site today
Written out, for you to apply. That is true on every platform right now. Connected delivery modes are planned, not built, and each integration page marks its own status. Integrations →