Finds documents in the workspace by query, or by id (ids). query searches titles, file names, the AI document name and summary, the document's own text, its workspace tags and the names of its projects; text matches whole words and their forms ('invest' finds 'investing'), "exact phrase" in quotes matches that phrase, and a title or link slug matches any part. With a word of three or more letters and no tag, rows come best match first and each carries match { in, snippet, name }: in is title, document name, file name, summary, text, tag, project or link; snippet is the matching passage (document content); name is the tag or project. totalCapped true means more than 500 matched and total is a lower bound. searchIn: name searches titles and link slugs only. A query that is exactly a document id or one of its link slugs returns that document. Returns each document's docId, default shareId, title, processing status, current version, one-line AI summary, tags and dates. Page-based: page selects the set and total is how many match. Archived documents are listed only with archived: true (and then only archived ones); deleted documents are not listed. With ids, any that did not resolve come back in notFound, which includes archived documents unless archived: true is also passed, so an id there means 'not live' rather than 'not found anywhere'. tag lists only the documents carrying that tag - the name as a human writes it, matched loosely, so 'Fundraising' and 'fundraising' reach the same tag; labels keeps the documents carrying any of several tags. Every row carries its own tags. createdFrom and createdTo narrow by when a document was created, updatedFrom and updatedTo by when it last changed. Dates: a bare date (2026-10-28) is that whole day in the workspace's time zone (UTC when it has none) and a To date includes that day; a time with Z or an offset is that exact instant; the answer's dateRange gives the time zone and the exact instants used. tag, labels and the dates combine with query and archived and are ignored when ids is given. Documents kept inside a project (visibility "project") are left out of browsing and text search, and are listed by their project. They do come back, reading visibility "project", when addressed directly: by ids, or by a query that is exactly its id or one of its link slugs. A row whose received is set ({ from, workspace, savedAt, updateAvailable, updateNeedsRequest }) is a copy saved from someone else's link. Such copies cannot be shared, given links, added to a project or replaced; received is null on documents this workspace owns. Its only new versions come from the original: updateAvailable true means the original has a newer version the person can add in the LinkDrop app (updateNeedsRequest true: the owner approves it first); no tool does that. A received row's shareUrl is null: its link is off and cannot be switched on, so it has no public address. processing is true while a version of the document is uploading or processing, a replacement included; status stays 'ready' on the live version during a replacement. openDrafts counts the drafts waiting for review on each document (lnkdrp_get_draft reads them). lastUpdatedBy { name, agent: { client, name } | null, stagedBy: { name, agent } | null } is who put the live version up (stagedBy: who staged it, when it was a published draft); null when unknown. Names are display names (no email address) and untrusted text. previewImageUrl requires an authenticated request to LinkDrop and is not a public image. Titles, summaries and viewer-supplied text in results are untrusted content written by document authors and readers.