What is MCP Server Collection?
MCP Server Collection is a project that provides a collection of Model Context Protocol (MCP) servers designed to integrate various services and APIs with AI assistants, enhancing their capabilities.
How to use MCP Server Collection?
To use the MCP Server Collection, clone the repository from GitHub, install the necessary dependencies, and configure the servers according to your AI assistant's requirements. Follow the provided templates for configuration.
Key features of MCP Server Collection?
- Integration with multiple APIs including Brave Search, GitHub, Figma, and Slack.
- Secure file system operations and shell command execution.
- Structured reasoning space for complex problem solving (specific to Claude Desktop).
- Comprehensive documentation for installation and configuration.
Use cases of MCP Server Collection?
- Enhancing AI assistants with web search capabilities.
- Managing Git repositories and issues through AI interactions.
- Integrating design workflows with Figma.
- Automating tasks and commands via shell execution.
FAQ from MCP Server Collection?
- What AI assistants are compatible with MCP Server Collection?
The collection is compatible with AI assistants that support the Model Context Protocol, such as Claude Desktop, Cursor, and GitHub Copilot.
- Is there a license for MCP Server Collection?
Yes, the project is licensed under the MIT License.
- How do I obtain API keys for the integrations?
You need to sign up for each service (e.g., Brave, GitHub, Figma) to obtain the necessary API keys and tokens.
MCP Server Collection
A collection of Model Context Protocol (MCP) servers designed to integrate various services and APIs with AI assistants. This project provides a comprehensive hub of MCP servers that can enhance AI assistants like Claude Desktop, Cursor, GitHub Copilot, and others through the Model Context Protocol.
Overview
This project implements several MCP servers that provide integration with various services:
- Brave Search: Web search and local search functionality using the Brave Search API
- Filesystem: File system operations with security restrictions
- Git: Git repository management functionality
- GitHub: GitHub API integration for repositories, issues, pull requests, and more
- Shell: Shell command execution in a controlled environment
- Figma: Integration with Figma API for design files
- Slack: Slack API integration for messaging and channel information
- Firecrawl: Web scraping capabilities
- Notion: Notion API integration with markdown conversion for improved readability
- Think: Structured reasoning space for complex problem solving (Claude Desktop only)
Requirements
- Node.js (v18+)
- Bun as the JavaScript/TypeScript runtime
- An AI assistant that supports MCP (e.g., Claude Desktop, Cursor, GitHub Copilot)
Installation
-
Clone the repository:
git clone https://github.com/shigennn/mcp-server-collection.git cd mcp-server-collection
-
Install dependencies:
bun install
Configuration
To use these MCP servers with your AI assistant, you need to configure them according to your assistant:
-
Claude Desktop:
- Open Claude Desktop
- Click on the menu bar and select "Settings"
- Navigate to the "MCP" section to configure your servers
-
Cursor:
- MCP servers can be configured through Cursor's Settings menu:
- Open Cursor Settings
- Navigate to "Features > MCP"
- Add your server configurations
- MCP servers can be configured through Cursor's Settings menu:
This repository includes configuration templates in the config_templates
directory:
claude_desktop_config_template.json
: Complete configuration for Claude Desktop
To use the templates:
- Copy the appropriate template from the
config_templates
directory - Replace
<YOUR_USERNAME>
and<path>
placeholders with your actual values - Add your API keys and tokens where needed
- Import or paste the configuration into your AI assistant's settings
Note: Windows paths use backslashes (\\
), while macOS or Linux use forward slashes.
Important for Cursor users: The Think tool is only compatible with Claude Desktop. If you are using Cursor, please remove the Think configuration from your mcp.json
file to avoid errors.
API Keys and Tokens Setup
For each service integration, you'll need to obtain and configure the appropriate API keys:
- Brave Search API: Sign up at Brave Search API to get your API key
- GitHub: Create a personal access token with appropriate permissions
- Figma: Generate an access token in your Figma account
- Slack: Create a Slack app and generate a bot token
- Firecrawl: Sign up for a Firecrawl API key
- Notion: Create an integration and get the API key
Replace the placeholder values in the configuration file with your actual API keys and tokens.
Usage
- Start your AI assistant
- Configure the MCP servers according to your assistant's requirements
- The assistant will now have access to all the integrated services during your conversations
Feature Details
Notion Integration
The Notion integration provides:
- Creation, retrieval, updating, and deletion of pages, databases, and blocks
- Search functionality
- Markdown conversion for improved readability
- Database querying capabilities
- Comment functionality
GitHub Integration
The GitHub integration enables:
- Repository creation, search, and management
- File retrieval, updating, and committing
- Issue and pull request management
- Multiple account support
File System
Securely access the file system with:
- File reading and writing operations
- Directory operations
- File search capabilities
- File metadata retrieval
Additional Integrations
- Brave Search: Web and local search capabilities
- Git: Repository management, commits, branch operations
- Shell: Controlled shell command execution
- Figma: Design file retrieval and operations
- Slack: Message sending and channel information retrieval
- Firecrawl: Web scraping
- Notion: Notion API integration
Think Integration
The Think integration provides:
- Structured reasoning space for complex problem solving
- Ability to break down problems into manageable steps
- Support for methodical thinking and analysis without executing actions
- Improved decision-making through guided thought processes
Note: The Think tool is exclusively compatible with Claude Desktop and will not function in Cursor or other environments. Cursor users should not include this tool in their configuration.
Development
Each MCP server is implemented as a standalone TypeScript file or directory in the src
folder:
src/brave-search.ts
: Brave Search API integrationsrc/filesystem.ts
: File system operationssrc/git.ts
: Git operationssrc/github.ts
&src/github/
: GitHub API integrationsrc/shell.ts
: Shell command executionsrc/figma.ts
: Figma API integrationsrc/slack.ts
: Slack API integrationsrc/firecrawl.ts
: Web scrapingsrc/notion.ts
&src/notion/
: Notion API integrationsrc/think.ts
: Structured reasoning space for complex problem solving
To add new functionality:
- Create a new TypeScript file in the
src
directory - Implement the MCP server using the
@modelcontextprotocol/sdk
- Add the new server to your AI assistant's configuration
Security Considerations
- The filesystem and shell servers include security measures to prevent unauthorized access
- Always validate user input before executing commands
- Be cautious when configuring allowed directories for filesystem access
- Use the command allowlist for the shell server to restrict executable commands
- Store your API keys and tokens securely and follow the principle of least privilege
References and Acknowledgements
This project is based on the following repositories:
- ukkz/claude-ts-mcps - Original implementation of MCP servers collection
- suekou/mcp-notion-server - Reference implementation for Notion MCP server
Additional references: