MCP server that can read rendered Dataview queries

I have many Dataview queries in my vault, often one note containing exactly one Dataview query and nothing else.

Is it possible somehow to read the rendered result of such a note using MCP?

For instance, when I use MCP in my Cline VS Code plugin, I get the following response upon the Obsidian MCP call:
“”"
The content of myobsidianvault/AI✨/SW-Engineering/Evaluation.md is a Dataview query, not a direct list of Markdown file paths as we initially assumed.

The query is:

TABLE 
  swe-agent AS "SWE Agent"
FROM "myobsidianvault/daily" 
WHERE swe-agent AND file.name >= "2025" 
SORT file.day DESC

This query, when rendered in Obsidian, would dynamically generate a list of notes from the myobsidianvault/daily/ directory that have an swe-agent property and a filename indicating they are from 2025 or later.

I cannot execute Dataview queries directly. To proceed with fetching the actual label notes, I need to know which files these are.
“”"

Two possibilities come to my mind:

  1. Use a plugin that converts your dataview queries in actual markdown, so that the MCP can read the markdown contents. For example, DataView Serializer or DataView Publisher.
  2. Use my fork of mcp-obsidian which, in combination with obsidian-local-rest-api can run dataview queries and retrieve the results.

Some notes about the 2nd approach:

  • The capability of running dataview queries through the rest api was already in the rest api, but the original implementation of mcp-obsidian did not take advantage of it, that’s why I did a fork. I also did a PR so perhaps this functionality ends up in the main mcp-obsidian repo.
  • Not all kind of dataview queries are supported, only TABLE ones. The local rest api plugin do not implement TASK, LIST or CALENDAR queries. Also, it rejects TABLE WITHOUT ID queries.
  • The LLM should be smart enough to detect that the contents of a note he retrieved via MCP contains a dataview query, and that this requires another mcp call to run that query and get the results. I guess most LLMs would be smart enough for this.