mcp-recall-md
Enables semantic search over local markdown note collections using vector embeddings, with real-time file watching and zero-config setup.
README
<h1 align="center">mcp-recall-md</h1>
<p align="center"> Local semantic search for your markdown notes — via <a href="https://modelcontextprotocol.io/">MCP</a>. </p>
<p align="center"> <a href="https://pypi.org/project/proj-mcp-recall-md/"><img src="https://img.shields.io/pypi/v/proj-mcp-recall-md" alt="PyPI version"></a> <a href="https://pypi.org/project/proj-mcp-recall-md/"><img src="https://img.shields.io/pypi/pyversions/proj-mcp-recall-md" alt="Python versions"></a> <a href="LICENSE"><img src="https://img.shields.io/github/license/kalikin-artem/proj-mcp-recall-md" alt="License"></a> <a href="https://ko-fi.com/kalikinartem"><img src="https://img.shields.io/badge/Ko--fi-Support%20this%20project-FF5E5B?logo=ko-fi&logoColor=white" alt="Ko-fi"></a> </p>
<br>
graph LR
A["<b>.md files</b><br><i>Obsidian · Logseq · plain md</i>"] -- "watch + index" --> B["<b>mcp-recall-md</b><br><i>local vector store</i>"]
C["<b>AI client</b><br><i>Claude · Cursor · Windsurf</i>"] -- "MCP search" --> B
style A fill:#f9f4eb,stroke:#c9a96e,color:#333
style B fill:#e8f0fe,stroke:#4a86c8,color:#333
style C fill:#e8f5e9,stroke:#5b9a5b,color:#333
"Search my notes about Kubernetes networking"
→ finds
kubernetes-networking.md(similarity: 0.53) — even though you phrased it differently than the note
<br>
- Search by meaning, not keywords — finds notes even when your wording doesn't match
- 100% offline — no API keys, no cloud, nothing leaves your machine
- Zero config — point at your folders, restart your AI client, done
- Real-time sync — file watcher picks up changes instantly, re-embeds only what changed
Quick start
Add to your MCP client config and restart:
{
"mcpServers": {
"mcp-recall-md": {
"command": "uvx",
"args": ["mcp-recall-md", "--vaults", "C:/Users/you/notes"]
}
}
}
Config file location:
.mcp.json(Claude Code) ·claude_desktop_config.json(Claude Desktop) · Cursor / Windsurf MCP settings
That's it. Your notes are searchable.
Installation
The quick start above uses uvx (recommended). Other options:
<details> <summary><b>pip</b></summary>
pip install mcp-recall-md
{
"mcpServers": {
"mcp-recall-md": {
"command": "mcp-recall-md",
"args": ["--vaults", "C:/Users/you/notes"]
}
}
}
</details>
<details> <summary><b>Standalone .exe (no Python needed)</b></summary>
- Download mcp-recall-md.exe from the latest release
- Put it somewhere permanent (e.g.
C:\Tools\mcp-recall-md\)
{
"mcpServers": {
"mcp-recall-md": {
"command": "C:/Tools/mcp-recall-md/mcp-recall-md.exe",
"args": ["--vaults", "C:/Users/you/notes"]
}
}
}
</details>
Configuration
Multiple vaults
List all folders — each is indexed independently:
"args": ["--vaults", "C:/notes/work", "C:/notes/personal", "C:/docs"]
.recallignore
Drop a .recallignore in any vault root to exclude files. Standard .gitignore syntax:
.obsidian/
_templates/
drafts/
CLI flags
| Flag | Default | Description |
|---|---|---|
--vaults |
(none) | Folders to index and watch |
--db-path |
~/.mcp-recall-md/db |
ChromaDB storage location |
--verbose |
off | Debug logging to stderr |
Tools
Your AI assistant gets these tools automatically via MCP:
| Tool | Description |
|---|---|
search |
Find notes by meaning — returns ranked results with similarity scores and file paths |
status |
Show indexed article count and watched vaults |
index |
Manually store an article (for use without --vaults) |
remove |
Delete an article from the index |
Most users only interact with search — everything else is automatic.
Troubleshooting
| Problem | Fix |
|---|---|
| Search returns nothing | Check that --vaults points to folders with .md files |
| First run is slow | Embedding model (~80 MB) downloads once on first use |
| Need to debug | Add --verbose, check ~/.mcp-recall-md/server.log |
| Force re-index | Delete ~/.mcp-recall-md/db and restart |
Logs: ~/.mcp-recall-md/server.log (5 MB max, 3 rotated backups)
Limitations
- Single-chunk embedding — large files (10k+ words) may search less precisely than shorter notes
- English-optimized — other languages work but with lower accuracy
License
MIT
<!-- mcp-name: io.github.kalikin-artem/recall-md -->
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.