mcpField
O
Omnidrive Mcp

Omnidrive Mcp

Local
github.com/sp-22Files & Storage7Updated 35w ago

About Omnidrive Mcp

OmniDrive is a desktop app for selectively sharing local folders with AI agents over MCP. It combines per-folder read/write permissions, path isolation, and a live activity log, without uploading files to OmniDrive servers.

Key features

  1. 01Select and share folders through a graphical interface
  2. 02Set read-only or read/write access per folder
  3. 03Restrict access to approved local paths
  4. 04View file reads, searches, and edits live
  5. 05Filter files using .mcpignore

Use cases

  1. 01Analyze local research collections and PDFs
  2. 02Summarize spreadsheets and invoices or prepare reports
  3. 03Search Obsidian or Markdown notes
  4. 04Let a coding agent inspect and edit a repository

MCP support

OmniDrive uses MCP to expose approved local folders to compatible MCP clients. Connections use a local process pipe or a local SSE bridge, and support file access according to configured permissions.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

No install method listed yet

We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.

Repository

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

OmniDrive

The secure, local-first gateway that bridges your personal files to any AI Agent.

OmniDrive is a "Plug & Play" desktop application that allows you to selectively share local folders with AI agents like Claude Desktop, Cursor, ChatGPT, and Gemini. Built on the Model Context Protocol (MCP), it gives your AI assistants the "eyes" they need to see your documents, spreadsheets, and codebase—while keeping you in absolute control.


Why OmniDrive?

Most AI agents are currently "context blind"—they don't know what's on your machine unless you manually upload files. Existing solutions are either complex terminal tools or insecure "all-access" scripts.

OmniDrive solves this by providing:

  • Zero Configuration: A clean, beautiful UI replaces complex JSON edits. Just drag, drop, and share.
  • Security-First (Zero Trust): AI access is mathematically restricted to approved paths. Path traversal (../) is impossible.
  • Granular Permissions: Set folders to Read-Only (safe browsing) or Read/Write (autonomous editing) on the fly.
  • Live Activity Streaming: See every file read, every search, and every edit the AI performs in real-time.

Key Capabilities

  • Universal Compatibility: Works with any LLM client that supports the MCP standard.
  • Plug & Play: Add folders in seconds. No terminal commands, no environment variables.
  • Total Transparency: A dedicated Live Activity Log window shows you exactly what the AI is "thinking" about your files.
  • Privacy by Design: No files are uploaded to OmniDrive servers. Everything happens over a local process pipe or a secure local SSE bridge.
  • Smart Filtering: Respects .mcpignore files to keep noise like node_modules out of the AI's view.

Use Cases (Beyond Coding)

While OmniDrive is a must-have for developers, it is designed for everyone:

Research & Analysis

Share your local library of PDFs and research notes. Ask your AI to synthesize trends across dozens of documents without ever hitting "upload."

Administrative Automation

Give the AI access to your spreadsheets and invoices. Ask it to generate summaries, identify anomalies, or prepare monthly reports.

Personal Knowledge Base

Bridge your Obsidian or Markdown notes to their AI agent. Talk to your "Second Brain" to find connections between your thoughts in real-time.

Advanced Software Engineering

Enable the AI to proactively fix bugs, refactor architecture, and verify changes across your entire repository with full Read/Write power.


Security Architecture

OmniDrive is built with a "Privacy First" philosophy:

  • Canonical Path Validation: Every request is verified against a secure whitelist of user-approved absolute paths.
  • Sandbox Isolation: Even if an LLM is "hallucinating" or being malicious, it cannot escape the folders you have explicitly shared.
  • Paired Web Connections: SSE (for browser agents) is locked to 127.0.0.1 and requires a secure handshake to prevent unauthorized web access.

Download (End Users)

If you just want to use OmniDrive, download the latest installer from:

Available assets depend on platform and include installers such as .dmg (macOS), .msi/.exe (Windows), and Linux bundles.

macOS "damaged and can't be opened" fix

If macOS blocks OmniDrive after downloading from Releases, run:

xattr -dr com.apple.quarantine /Applications/OmniDrive.app
xattr -dr com.apple.provenance /Applications/OmniDrive.app
open /Applications/OmniDrive.app

If needed, right-click OmniDrive.app and choose Open once.


Build From Source (Developers)

Prerequisites
Setup
  1. Clone the repository:

    git clone https://github.com/sp-22/omnidrive-mcp.git
    cd omnidrive-mcp
    
  2. Install dependencies:

    npm ci
    
  3. Run in development:

    npm run tauri dev
    
  4. Build production bundles:

    npm run tauri build
    

Build artifacts are generated under src-tauri/target/release/bundle/.


Contributing

  1. Fork this repo
  2. Create a branch
  3. Commit and push your changes
  4. Open a Pull Request

To download source without Git, use GitHub's Code -> Download ZIP on the repository page.


Release Process (Maintainers)

This repository includes a GitHub Actions workflow that builds and publishes release assets for macOS, Windows, and Linux whenever a version tag is pushed.

  1. Update versions:

    • package.json
    • src-tauri/Cargo.toml
    • src-tauri/tauri.conf.json
  2. Commit and push:

    git add .
    git commit -m "chore: release v0.x.x"
    git push
    
  3. Create and push a tag:

    git tag v0.x.x
    git push origin v0.x.x
    
  4. The workflow will publish binaries to:


Tech Stack

  • Frontend: React, TypeScript, Vanilla CSS (Premium Glassmorphism)
  • Engine: Rust (High-performance native sidecar)
  • Framework: Tauri (Secure system bridging)
  • Protocol: RMCP (Rust implementation of Model Context Protocol)

License

OmniDrive is released under the MIT License. See LICENSE for details.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.