Summarize

The note limit in summarize's description is the default. MARKDOWN_VAULT_MCP_SUMMARIZE_MAX_NOTES sets it, and a running server states its own value.

summarize

Summarize Notes. Read-only. Tags: summarize.

Summarize notes or folders with a language model; returns one summary that cites its source notes by path, or one summary per note.

It covers at most 50 notes per call; split a larger folder into subfolders, found with get_toc, and combine the results. The notes are sent to a language model this server's operator chose.

Parameters

Name	Type	Default	Description
paths	array of string	required	Note paths, such as "notes/topic.md", and folders, such as "notes/project", which stand for every note under them.
focus	string \| null	null	What the summary should concentrate on, such as "extract action items"; omit for a general summary.
mode	synthesis \| per_note	"synthesis"	"synthesis" (default) for one summary across all the notes, or "per_note" for one per note.
max_notes	integer \| null	null	A smaller note limit for this call, to narrow the work; omit for the server's limit.

Returns

When the summary completes within the soft deadline, a dict with "status": "completed" plus:

• summary (str): The generated summary text.
• sources (list[dict]): The notes that were summarised, each with path and title: always populated so individual notes are attributable even when the prose does not name every one.
• mode (str): The mode used ("synthesis" or "per_note").
• truncated (bool): True when content was lost (notes omitted at the note limit, notes skipped, or content cut to fit a request budget).
• notes_included (int): Notes whose content reached the model.
• notes_omitted (int): Matched notes dropped by the note limit. When non-zero, the summary does not cover the whole selection; surface that to the reader.
• notes_limit (int): The note limit in effect for this call.
• hint (str | None): Recovery guidance when notes were omitted or skipped, one step per cause; follow it for full coverage. None when fully covered.
• skipped (list[dict]): Matched notes left out for a reason other than the note limit, each with path and reason (not_found, invalid_path, over_read_limit, unreadable).

When the work is promoted to a background job, a dict with "status": "working", a job_id string, poll_with naming get_job_result, retry_after_s (the seconds to wait before polling) and a message; call get_job_result with the job_id to fetch the result.

Outcomes and errors

• InvalidRequestError: If paths is empty, mode is invalid, max_notes is below 1, or the paths hold no note that exists and is within the read limit.
• ValueError: If every note found exists but cannot be read.
• SummarizeTimeoutError: If a backend call outran its time limit; narrow the request.
• RuntimeError: If the summarization backend call fails within the soft deadline. A backend failure that happens after promotion is reported through get_job_result instead.
• ToolError: If the index is busy or still building; retry shortly.
• IndexUnavailableError: If the index build failed or the index is broken; get_index_status reports the error.

{"paths": ["1-Projects/export"], "focus": "decisions and their rationale"}

A slow run returns {"status": "working", "job_id": "…", "poll_with": "get_job_result", "retry_after_s": 5.0, "message": "…"}; poll get_job_result with that job_id.