Your LLM just got read-write access to your Obsidian vault

What it does This is a thin but well-used MCP server that sits between an LLM client and Obsidian, turning your vault into a set of callable tools. It lets an agent list directories, read note contents, search across files, and modify notes by appending or patching content relative to headings or frontmatter. In practice, that means your assistant can summarize meeting notes, search for mentions of a technology, or write new documents directly into your knowledge base.

The interesting bit The real value is the two-way street: most integrations treat Obsidian as a read-only archive, but patch_content and append_content let the model edit your second brain. It is essentially polite glue code around the Obsidian Local REST API community plugin, but with 4,300+ stars it clearly fills a gap.

Key highlights

• Exposes seven vault operations: list, read, search, patch, append, and delete
• Enables LLM agents to both consume and update Obsidian notes
• Relies on the Obsidian Local REST API plugin for actual vault access
• README examples center on Claude Desktop workflows

Caveats

• Requires a separate community plugin (Obsidian Local REST API) to function
• Documentation and examples are heavily Claude-centric; generic MCP compatibility is assumed but not demonstrated
• Core logic is a straightforward wrapper over the REST API, so advanced Obsidian features (canvas, plugins, metadata) are not exposed

Verdict Worth a look if you want your LLM assistant to actively manage an Obsidian vault rather than just quote from it. If you do not use Obsidian or your MCP client lacks robust tool support, this will not change your life.