Log inSubmit server
Vault Cortex logo

Vault Cortex

by aliasunder

Connects AI agents to an Obsidian vault for hybrid search, memory, tasks, file access, link analysis, and write-back.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install Vault Cortex

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "aliasunder-vault-cortex": {
      "command": "npx",
      "args": [
        "-y",
        "vault-cortex@latest"
      ],
      "env": {
        "MCP_AUTH_TOKEN": "YOUR_VALUE_HERE",
        "VAULT_PATH": "YOUR_VALUE_HERE"
      }
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Values are written into the snippet above so you can paste it as is. They never leave your browser.

Install not yet sandbox-tested— we haven't run this command through our automated sandbox yet. Not a red flag; we're working through the catalog.

Overview

aliasunder/vault-cortex connects MCP-compatible AI agents directly to an Obsidian vault or folder of Markdown files. It combines FTS5 keyword search with vector similarity and cross-encoder reranking, while also exposing tools for structured memory, tasks, files, links, and write operations. Docker runs the server without requiring Obsidian, and remote deployments can use Obsidian Sync and OAuth 2.1 for access from phones or other MCP clients. Reach for it when an agent needs to retrieve, reason over, and update a personal knowledge base.

Use cases

  • Search personal notes using exact terms and semantic similarity
  • Recall dated decisions and evolving preferences from stored memory
  • Triage and update tasks across Kanban-style lanes
  • Inspect note links and identify orphaned content
  • Read and update a remote vault from a phone or MCP client

Key features

  • Hybrid FTS5 and vector search with reranking
  • Read/write access to Markdown and other vault files
  • Append-only structured memory with topic history
  • Task queries, updates, Kanban lanes, and recurring tasks
  • Backlink, outgoing-link, and orphan detection
  • OAuth 2.1 support for remote deployments

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked 2d ago

Recent health check succeeded.

8/8 checks healthy over the last 79d

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 11d ago via OSV.dev · vault-cortex@latest (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars32Stargazers on the source repository.
npm downloads718Package downloads in the last 30 days.
Last commit3d agoMost recent push to the default branch.
Directory activity8 viewsConfig copies, upvotes, and views on AllMCPs.

How Vault Cortex works

What aliasunder/vault-cortex does

aliasunder/vault-cortex exposes an Obsidian vault to AI agents through the Model Context Protocol. It works directly with files on disk, so the Obsidian desktop application does not need to be open and no Obsidian plugin or separate bridge is required. A vault can contain standard Markdown notes as well as Obsidian-native structures such as frontmatter, wikilinks, tags, headings, and daily notes.

The server supports both retrieval and write-back. Agents can search notes, read and update content, manage tasks, store dated memory entries, inspect relationships between notes, and work with non-Markdown files. Built-in prompts support workflows such as vault health checks, memory review, and daily reconciliation using current vault data.

How it works

Search combines FTS5 keyword matching with vector semantic similarity through reciprocal-rank fusion. A cross-encoder reranker then refines results for queries where intent matters. This lets exact terms and specialist vocabulary remain searchable while also finding notes whose wording differs from the query.

Memory entries are dated and append-only. Topic recall can return both the latest position on a subject and the dated history behind it, allowing an agent to follow how a viewpoint changed over time. Task handling recognizes both Tasks-plugin emoji syntax and Dataview inline fields. Recurring tasks produce their next occurrence when completed.

The server runs in Docker and reads the vault folder directly. For remote use, the vault can reside on a server and stay synchronized through Obsidian Sync. OAuth 2.1 protects remote access, while local setup uses an MCP authentication token.

Setup and configuration

The quickest local setup uses Node.js 22.12 or newer for the command-line setup tool and Docker, or a compatible runtime such as OrbStack, Colima, or Podman. An existing Obsidian vault is suitable, and a folder containing Markdown files can also be used.

Run the setup wizard with:

Terminal
npx vault-cortex@latest init

The wizard asks for the vault path, creates authentication and configuration files, starts the container, and displays connection details for the MCP client. The CLI also provides commands for configuration, upgrades, starting, restarting, viewing logs, and stopping the service.

Manual Docker Compose setup is available without Node.js. Configure MCP_AUTH_TOKEN and VAULT_PATH in the environment file, then start the service with docker compose up. Local deployments can be used from the same machine; remote deployments support access from a phone, claude.ai, or another remote MCP client. One-click deployment options handle HTTPS, restarts, generated tokens, and persistent storage, while self-hosted HTTPS requires separate configuration.

Tools and capabilities

  • Search vault content with keyword, semantic, and reranked hybrid retrieval.
  • Read and write Markdown notes and other files in the vault.
  • Process images, PDFs, canvases, and data files in supported readable forms.
  • Create and update dated, append-only memory entries and recall topic history.
  • Query, complete, reprioritize, and move tasks between Kanban lanes.
  • Handle recurring tasks by creating their next occurrence after completion.
  • Inspect backlinks, outgoing links, and orphaned notes.
  • Use guided prompts for vault health, memory review, and daily reconciliation.
  • Work with frontmatter, wikilinks, tags, headings, and daily-note conventions.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Vault Cortex

aliasunder/vault-cortex is an MCP server that connects AI agents to an Obsidian vault or Markdown folder. Its main tools cover hybrid keyword and semantic search, note and file read/write access, structured memory, task management, and link-graph inspection. It runs without Obsidian and supports protected remote access.

Install it locally by running `npx vault-cortex@latest init` with Docker and Node.js 22.12 or newer. The setup wizard asks for the vault path, creates authentication and configuration files, starts the container, and prints MCP connection details. Docker Compose can also be configured manually without Node.js.

No, aliasunder/vault-cortex reads and writes vault files directly, so Obsidian does not need to be open. It works with `.md` files on disk and also understands Obsidian conventions such as frontmatter, wikilinks, tags, headings, and daily notes.

Yes, remote deployments can expose a vault to phones, claude.ai, and other remote MCP clients. Obsidian Sync can keep the server-side vault current, while OAuth 2.1 protects access. One-click deployments provide HTTPS and generated tokens; self-hosted deployments require you to arrange HTTPS.

aliasunder/vault-cortex parses task formats from the Obsidian Tasks plugin and Dataview inline fields. Agents can filter tasks by status, date, or priority, then complete, reprioritize, or move them between lanes. Completing a recurring task creates its next occurrence.

Related MCP servers

All in Knowledge & Memory All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewVault Cortex AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/aliasunder-vault-cortex?style=directory)](https://allmcps.com/mcp/aliasunder-vault-cortex)
HTML Embed
<a href="https://allmcps.com/mcp/aliasunder-vault-cortex"><img src="https://allmcps.com/api/badge/aliasunder-vault-cortex?style=directory" alt="Vault Cortex on AllMCPs" /></a>