List every notebook (思源笔记本) in the workspace with its ID, name and open state. Call this first when you need a notebook ID for other tools.

Full-text search across the workspace and return matching blocks with their IDs and document paths. Use this to locate content before reading or editing it. Supports SiYuan search syntax, e.g. foo bar (AND), "exact phrase", foo OR bar, -exclude, * wildcard.

List documents in the workspace or in one notebook, most recently updated first. Useful for browsing what exists before searching.

Read a document as Markdown. Accepts a document ID or a human-readable path such as /Notes/My doc. This is the primary way to read note content.

Read one block by ID. kramdown returns the source including block attributes (IDs, made useful for locating nested blocks to edit); children lists direct child blocks.

Append Markdown as new blocks. With parent_id the content is appended inside that block (use a document ID to append to a document); with previous_id it is inserted directly after that block. Provide exactly one of them.

Replace the content of an existing block with new Markdown. The block type may change (e.g. a paragraph into a heading) unless lock_type is true. Get the ID from get_block or search_notes first.

Delete a block (and its children) by ID. Deleting a document block removes the whole document. Use get_document or search_notes first to confirm the target.

Create a document from Markdown. path is the human-readable path, e.g. /Notes/Idea; missing parent documents are created automatically. An existing path is not overwritten — a new document is created alongside it.

Change the title of a document. Accepts a document ID or a path.

Delete a document. Accepts a document ID or a path. The document is moved to the workspace trash (数据历史) and can be recovered from SiYuan.

Move a document under a different parent document, or into a different notebook. to may be a parent document ID/path, or a notebook ID/name to move it to that notebook's root.

Run a read-only SQL query against the SiYuan index to answer questions the other tools cannot, e.g. counting blocks or filtering by tag/attribute. Main table: blocks (id, parent_id, root_id, box, path, hpath, type, subtype, content, tag, ial, created, updated). Only SELECT/WITH/EXPLAIN are allowed; results are capped at 64 rows by the kernel.