MCP Connectors: Integrate SaaS Tools into AI Apps with 35+ Pre-built Ones, Support Local Dev

MCP Connectors

This is an open - source project that includes over 35 pre - built MCP connectors for integrating SaaS tools into AI applications, supporting local development and type - safe configuration.

Developer tools Artificial intelligence chatbots #MCP Connectors #SaaS Integration #AI Tools #Open - source Project .TypeScript

rating : 2.5 points

downloads : 3.2K

update time : 2025-12-12

Open Site

What are Disco MCP Connectors?

Disco MCP Connectors is a collection of open - source connectors based on the Model Context Protocol (MCP), specifically designed to provide standardized connections between AI applications and various SaaS tools. It allows developers to easily integrate more than 35 popular tools such as Asana, GitHub, Slack, and Notion into AI applications without having to write complex API integration code from scratch.

How to use Disco MCP Connectors?

There are two main ways to use them: 1) As an end - user, you can use these connectors on the disco.dev platform without any configuration; 2) As a developer, you can install the NPM package and use these pre - built connectors in your own MCP server. The project provides full TypeScript type support and a local development environment.

Applicable scenarios

Suitable for developers who need to connect AI assistants (such as Claude, Cursor, etc.) to enterprise SaaS tools, AI application developers who want to quickly build tool integration functions, and teams that need to centrally manage multiple tool API calls.

Main Features

Abundant pre - built connectors

Provides more than 35 production - ready SaaS tool connectors, including popular tools such as Asana, GitHub, Slack, Notion, Jira, Linear, Todoist, Google Drive, and Supabase.

Type - safe development experience

Built on TypeScript and Zod, it provides complete type definitions and validation to ensure type safety during development and runtime data validation.

Cross - platform runtime support

Supports running in various runtime environments such as Bun, Node.js, and Cloudflare Workers, providing flexible deployment options.

Friendly for local development

Provides an auto - reload function, supports rapid iterative development, and does not require complex server configuration.

Optimized for AI assistants

Specifically designed for AI coding assistants (such as Claude, Cursor, etc.), it provides clear type hints and configuration patterns to facilitate AI understanding and generation of correct code.

Zero - dependency server runtime

It does not depend on a specific server runtime or transport protocol, providing pure connector functions for easy integration into existing systems.

Advantages

Ready to use: Over 35 pre - built connectors, no need to develop from scratch

Type - safe: Full TypeScript type support, reducing runtime errors

Easy to integrate: Can be used with a simple NPM package installation

Cross - platform: Supports multiple JavaScript runtime environments

Community - driven: An open - source project with continuous updates and maintenance

AI - friendly: API design optimized specifically for AI assistants

Limitations

Requires basic programming knowledge: Although pre - built connectors are provided, certain development skills are still required for integration

Depends on external tools: Requires corresponding SaaS tool accounts and API credentials

Learning curve: Needs to understand the basic concepts of the MCP protocol

Customization limitations: Additional development may be required for highly customized integration requirements

How to Use

Environment Preparation

First, you need to install the Bun runtime environment, which is the recommended development environment for the project.

Clone the Project

Clone the project code to your local development environment.

Install Dependencies

Use Bun to install all the dependency packages required by the project.

Build the Project

Compile the TypeScript code to generate runnable JavaScript files.

Start the Test Server

Start a test connector server that does not require credentials to verify the successful installation.

Use Real Connectors

When using specific SaaS tool connectors, you need to provide the corresponding API credentials.

Usage Examples

AI Assistant Manages GitHub Tasks

Create, view, and update GitHub Issues directly through the Claude AI assistant without leaving the chat interface.

Intelligent Meeting Minutes Arrangement

Automatically organize Slack discussion content into Notion documents to form structured meeting minutes.

Cross - tool Task Synchronization

Tasks created in Asana are automatically synchronized to Linear and Todoist to keep task states consistent across multiple platforms.

Intelligent Document Retrieval

Retrieve relevant documents from Google Drive and Supabase databases to provide context information for AI.

Frequently Asked Questions

Do I need programming experience to use this project?

Is this project free?

Which SaaS tools are supported?

Can I use it in a production environment?

How to add a new connector?

Do I need to manage API credentials myself?

Which AI assistants are supported?

What about the performance?

Related Resources

Official Documentation

Detailed project documentation, including connector writing guides and local running instructions

disco.dev Platform

An online platform to use these connectors without configuration

GitHub Repository

Project source code and issue tracking

NPM Package - Connectors

Page of the @stackone/mcp - connectors NPM package

NPM Package - Configuration Types

Page of the @stackone/mcp - config - types NPM package

StackOne Official Website

Official website of the project development team

Model Context Protocol

Official documentation of the MCP protocol

Apache 2.0 License

Open - source license used by the project

🚀 MCP Connectors

Pre-built MCP connectors for popular SaaS tools, powering disco.dev

[![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](https://opensource.org/licenses/Apache-2.0) [![npm version](https://badge.fury.io/js/@stackone%2Fmcp-connectors.svg)](https://badge.fury.io/js/@stackone%2Fmcp-connectors) [![npm version](https://badge.fury.io/js/@stackone%2Fmcp-config-types.svg)](https://badge.fury.io/js/@stackone%2Fmcp-config-types)

✨ Features

This monorepo contains two packages built with TypeScript and Zod:

@stackone/mcp-connectors

35+ Production-Ready Connectors: These connectors are ready to use in production environments.
Typesafe Wrapper: It provides a typesafe wrapper around MCP capabilities.
Local Development with Auto-Reload: Facilitates local development with auto - reload functionality.
No Dependencies on Server Runtime or Transport: Can run without depending on specific server runtime or transport.
Cross - Platform Compatibility: Runs on Bun, Node, and Cloudflare Workers.

@stackone/mcp-config-types

Zod Schemas: Offers Zod schemas for connector credentials and setup fields.
TypeScript Types: Provides TypeScript types for all connector configurations.
Designed for AI Coding Agents: Suitable for AI coding agents like Claude, Cursor, etc.
Type - Safe Development: Enables type - safe connector development.

🚀 Quick Start

For End Users

The MCP connectors in this repo power disco.dev, which is the easiest way to connect your favourite tools to your favourite AI applications using the Model Context Protocol.

Go to disco.dev to get started with zero setup.

For Developers

Use these packages to build your own MCP servers:

npm install @stackone/mcp-connectors @stackone/mcp-config-types

The connectors can be imported and used in your own MCP server implementations, and the config types provide TypeScript definitions for all connector configurations.

📦 Installation

If you want to run the connectors locally or contribute to the project:

Clone the repository:

git clone https://github.com/stackone-ai/mcp-connectors.git
cd mcp-connectors

Install Bun (if you don't have it already):

curl -fsSL https://bun.sh/install | bash

Install dependencies:
```
bun install
```
Build the project:
```
bun run build
```
Run the tests:
```
bun test
```
Check out the documentation:
- See the directory for detailed guides.
- Start with Running Locally for setup instructions.
- Read Writing Connectors to create your own.

💻 Usage Examples

Start a server from a connector

No credentials needed:

# Start a test server in the background
bun run server -- -- --connector test
bun run server -- -- --connector documentation

# Start with credentials (credentials object schema may be different for each connector)
bun run server -- -- --connector github --credentials '{"token":"ghp_xxx"}'

Server runs at http://localhost:3000/mcp

Package structure

packages/mcp-connectors/: The main connectors package.
packages/mcp-config-types/: Shared configuration types.

📚 Documentation

Writing Connectors: How to write a connector.
Running Locally: Detailed setup and configuration.

📄 License

Apache 2.0 - see LICENSE

Inspiration

"Everything is just fetch wrapped in a function, delivered in pure typescript."

The typing in this package is heavily inspired by the AI SDK by Vercel, and the development process has been influenced by the work of Sam at Alchemy.

Built by StackOne • Powers disco.dev