MCP server for Obsidian vaults: Obsidian CLI when the app runs, direct file access when it does not
Wellknown found it in public sources; nobody has proven control of it yet. Claiming takes one click if the repository is under your GitHub account, or a small file on your domain otherwise. Verified owners get the badge, 15-minute checks, status alerts, edits that outrank crawled data, and a ranking boost.
Agents can do it too: POST https://wellknown.network/api/v1/claims with {"agent":"obsidian-blade-mcp","method":"well_known_file"} — machine-readable steps at claim.json, guide at /docs/claim.
Everything here was measured by our prober or read from a registry. Nothing is self-reported.
Attributed to the source that supplied each field. Treated as claims, not facts.
# Obsidian Blade MCP An [MCP](https://modelcontextprotocol.io) server that gives any MCP client, such as Claude Desktop, Claude Code or a headless agent, precise tools over an [Obsidian](https://obsidian.md) vault. When Obsidian is running it uses the official Obsidian CLI and Obsidian's own index. When Obsidian is closed, or the host has no Obsidian at all, it reads and writes the vault directory directly. Same 40 tools either way, read-only unless you opt in to writes, and it never launches Obsidian for you. This project is not affiliated with, endorsed by, or sponsored by Obsidian.md or Dynalist Inc. "Obsidian" is a trademark of its respective owner. ## Contents - [Why this exists](#why-this-exists) - [Two backends, one surface](#two-backends-one-surface) - [Install](#install) - [Enabling writes](#enabling-writes) - [Safety model](#safety-model) - [Tools](#tools) - [Configuration](#configuration) - [Limits and known differences](#limits-and-known-differences) - [Troubleshooting](#troubleshooting) - [Development](#development) ## Why this exists Most Obsidian integrations need the desktop app open, a community plugin installed, or a REST bridge running. That is fine on a laptop and useless on a server, in CI, or on a machine where Obsidian is simply closed. This server works in all of those places. On a desktop with Obsidian open it defers to Obsidian's index, so search, Bases and the link graph are exactly what you see in the app. Everywhere else it falls back to a deterministic parser over the Markdown files: frontmatter, wikilinks, tags, tasks, daily notes and templates, with no embeddings, no inference and no network. It is also built for agents rather than for people typing commands. Output is compact and capped, every response says which backend answered, writes are off until an operator turns them on, destructive operations dry-run first, and every edit can be guarded by a content hash so an agent never overwrites a note a human changed underneath i…
Mapped onto the structured taxonomy from declared text and observed tool names. Confidence shown for derived entries.
Every source is kept verbatim. Field changes are logged as events.