Back to MCP directory
publicPublicdnsLocal runtime

mcp-everything

一个实现MCP协议所有功能的测试服务器,包括提示、工具、资源和采样,旨在测试MCP客户端而非实际应用。

article

README

MCP Everything

Note: This project was extracted from https://github.com/modelcontextprotocol/servers/tree/main/src/everything to create a standalone implementation.

This MCP server project demonstrates various features of the Model Context Protocol (MCP). It includes server implementations in TypeScript and Python, serving as test servers for MCP client builders. Both implementations aim for functional parity, showcasing capabilities like prompts, tools, resources, sampling, logging, and more.

Common Features

Tools

  1. echo

    • Simple tool to echo back input messages
    • Input:
      • message (string): Message to echo back
    • Returns: Text content with echoed message
  2. add

    • Adds two numbers together
    • Inputs:
      • a (number): First number
      • b (number): Second number
    • Returns: Text result of the addition
  3. longRunningOperation

    • Demonstrates progress notifications for long operations
    • Inputs:
      • duration (number, default: 10): Duration in seconds
      • steps (number, default: 5): Number of progress steps
    • Returns: Completion message with duration and steps
    • Sends progress notifications during execution
  4. sampleLLM

    • Demonstrates LLM sampling capability using MCP sampling feature
    • Inputs:
      • prompt (string): The prompt to send to the LLM
      • maxTokens (number, default: 100): Maximum tokens to generate
    • Returns: Generated LLM response
  5. getTinyImage

    • Returns a small test image
    • No inputs required
    • Returns: Base64 encoded PNG image data
  6. printEnv

    • Prints all environment variables
    • Useful for debugging MCP server configuration
    • No inputs required
    • Returns: JSON string of all environment variables
  7. annotatedMessage

    • Demonstrates how annotations can be used to provide metadata about content
    • Inputs:
      • messageType (enum: "error" | "success" | "debug"): Type of message to demonstrate different annotation patterns
      • includeImage (boolean, default: false): Whether to include an example image
    • Returns: Content with varying annotations

Resources

The server provides 100 test resources in two formats:

  • Even numbered resources:

    • Plaintext format
    • URI pattern: test://static/resource/{even_number}
    • Content: Simple text description
  • Odd numbered resources:

    • Binary blob format
    • URI pattern: test://static/resource/{odd_number}
    • Content: Base64 encoded binary data

Resource features:

  • Supports pagination (10 items per page)
  • Allows subscribing to resource updates
  • Demonstrates resource templates
  • Auto-updates subscribed resources every 5 seconds

Prompts

  1. simple_prompt

    • Basic prompt without arguments
    • Returns: Single message exchange
  2. complex_prompt

    • Advanced prompt demonstrating argument handling
    • Required arguments:
      • temperature (number): Temperature setting
    • Optional arguments:
      • style (string): Output style preference
    • Returns: Multi-turn conversation with images

Logging

The server sends random-leveled log messages every 15 seconds to demonstrate the logging capabilities of MCP.

TypeScript Implementation

Location: typescript/

Installation

Local Development

# Clone the repository (if not already done)
# git clone https://github.com/modelcontextprotocol/mcp-everything.git
# cd mcp-everything

# Navigate to the TypeScript directory from the repository root
cd typescript

# Install dependencies
npm install

# Build the project
npm run build

# Start the server
npm start

Global Installation

# Navigate to the TypeScript directory:
# cd path/to/mcp-everything/typescript
# Then install globally from the local package:
npm install -g .

# Run the server 
# (The command name will depend on the 'bin' field in typescript/package.json, 
#  e.g., 'mcp-everything-ts' or 'mcp-everything' if modified)
# Example:
# mcp-everything-ts

Note: Global installation functionality and the exact command depend on the bin configuration within typescript/package.json.

Docker

# Build the Docker image from the repository root
# (Assumes Dockerfile is updated to handle APP_DIR build argument or typescript context)
docker build -t mcp-everything-ts -f Dockerfile . --build-arg APP_DIR=typescript

# Run the container
docker run -it mcp-everything-ts

Usage with MCP Clients (e.g., Claude Desktop)

Add to your client's MCP server configuration. Paths might need adjustment.

If using npx with a published package (e.g., mcp-everything-ts):

{
  "mcpServers": {
    "everything-ts-npx": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-everything-ts" 
        // Replace 'mcp-everything-ts' with the actual package name if different
      ]
    }
  }
}

If running from a local build (ensure cwd is the typescript directory):

{
  "mcpServers": {
    "everything-ts-local": {
      "command": "npm",
      "args": [
        "start"
      ],
      // Ensure 'cwd' points to the 'typescript' directory of this project.
      "cwd": "path/to/mcp-everything/typescript" 
    }
  }
}

Replace "path/to/mcp-everything/typescript" with the correct path.


Python Implementation

Location: python/

Installation & Setup

  1. From the repository root, navigate to the Python directory:
    cd python
    
  2. (Recommended) Create and activate a virtual environment:
    python -m venv venv
    source venv/bin/activate  # On Windows: venv\Scripts\activate
    
  3. Install dependencies:
    pip install -r requirements.txt
    

Running the Server

From within the python directory (and with the virtual environment activated if used):

python mcp_server.py

The server listens for MCP messages over stdio.

Usage with MCP Clients (e.g., Claude Desktop)

Example configuration:

{
  "mcpServers": {
    "everything-py": {
      "command": "python", // Or "path/to/python/venv/bin/python"
      "args": ["mcp_server.py"],
      "cwd": "path/to/mcp-everything/python" // Ensure this is the correct path
    }
  }
}

Replace "path/to/mcp-everything/python" with the correct path.

Testing the Server via GitHub Actions

Note: Implementation is in progress. The server is not yet fully functional.

  • Test 1
  • Test 2
help

Runtime guide

cloud

Hosted runtime

Hosted servers run from a provider-managed environment. You usually connect the MCP client to the hosted endpoint or follow the provider's authorization flow, without keeping a local process alive

  1. Open provider connection page
  2. Authorize or copy endpoint
  3. Connect from your MCP client
terminal

Local runtime / other methods

Local servers run on your own machine or infrastructure. You normally copy the server_config into your MCP client, install the required package, and provide env variables from env_schema when needed

  1. Copy server_config
  2. Install required package
  3. Fill env variables and restart client