Searches pages by keyword (full-text search).

It is a lexical search (PostgreSQL FTS): use meaningful words. Empty-meaning terms ("a", "de", "the") are stopwords and return no results.

Fetches a page: metadata and content.

Returns the page's metadata plus the content field with the full text. Internally it combines /pages/info with /pages/export.

Creates a new page in a space, with or without content.

If you provide content, the page is uploaded through the import endpoint (/pages/import), which is the only REST way that persists the page body in Docmost. The title is taken from the first heading; if you also provide title, that one is used.

Renames a page.

To change the content, use update_page_content.

Writes into the body of an already-created page.

Docmost's REST API cannot do this — the body lives in the Yjs collaboration server — so this tool opens the collaboration WebSocket, edits the document and waits for it to be persisted (~13 s).

Unlike /pages/update, which accepts an operation field and ignores it, mode genuinely works here:

• "replace" (default): the page ends up with exactly this Markdown.

• "append": keeps what is there and adds this at the end.

• "prepend": keeps what is there and adds this at the start.

The Markdown is converted with Docmost's own converter, so it supports its whole schema: headings, bold, italics, code, lists, tables, blockquotes, code blocks, etc.

Requires the yjs extra (uv sync --extra yjs).

Moves a page to the trash (recoverable with restore_page).

Restores a page that is in the trash.

Moves or reorders a page. Say where it should go; the position is computed.

Three ways to place a page, in order of preference:

• Nest it: pass parent_page_id and the page becomes a child of that page (appended at the end of its children by default).

• Reorder it: pass after or before with the id of a sibling to sit next to.

• Neither: the page is appended at the end of its current parent.

You never have to invent a position. Docmost orders pages with fractional indexing — a base62 string it compares lexicographically, which must be 5-12 characters — and that is fiddly to get right by hand. This tool reads the neighbouring positions and generates a valid one in between.

Lists recently updated pages.

Start here. Lists everything in the workspace: every space and every page.

Use this to answer questions like "what do I have in Docmost?", to take an inventory, or to find a page by title before reading it with get_page. Unlike list_child_pages, it walks the whole page tree, so nested pages are included.

Each page carries id, slug_id, title, parent_page_id and depth, which is enough to rebuild the tree or to list everything flat.

With an activity level above "none" each page also carries updated_at and updated_by, and the result includes recently_updated: the most recently touched pages, newest first. Use it to decide what is worth reading first instead of walking the inventory blindly.

It also returns a summary with counts per space, tree shape (roots, containers, leaves, max depth), date ranges and recency buckets, pages never edited after creation, top editors and any orphaned pages. Every figure comes from data already gathered, so it costs nothing extra; summary.date_coverage states how much of the workspace the date-based numbers actually cover.

Lists only the direct children of a space or a page.

This is one level, not the whole tree. To enumerate everything in the workspace, use get_workspace_overview instead.

Returns the ancestor path (breadcrumbs) of a page.

Lists the version history of a page.

Lists the spaces the user has access to.

Gets the details of a space.

Creates a new space.

Gets the comments of a page.

Adds a comment to a page.

Updates the content of a comment.

Returns the authenticated user and workspace information.

Lists the members of the workspace.