I craft unique cereal names, stories, and ridiculously cute Cereal Baby images.

go-mcp
Make your own MCP servers with Golang
3 years
Works with Finder
1
Github Watches
0
Github Forks
0
Github Stars
Go-MCP
Go-MCP is a Go implementation of the Model Context Protocol (MCP). MCP is a protocol for building AI services, defining three core primitives: Prompts, Tools, and Resources.
Features
- Complete MCP protocol implementation
- Type-safe API
- Multiple transport options (HTTP, SSE)
- Unified response structure
- Pagination support
- Change notifications support
Installation
go get github.com/virgoC0der/go-mcp
Quick Start
Creating a Server
package main
import (
"context"
"log"
"github.com/virgoC0der/go-mcp"
"github.com/virgoC0der/go-mcp/internal/types"
)
// Implement MCPService interface
type MyService struct {
// ... your service implementation
}
func main() {
// Create service instance
service := &MyService{}
// Create server
server, err := mcp.NewServer(service, &types.ServerOptions{
Address: ":8080",
Type: "sse", // or "http"
})
if err != nil {
log.Fatal(err)
}
// Initialize server
ctx := context.Background()
if err := server.Initialize(ctx, nil); err != nil {
log.Fatal(err)
}
// Start server
if err := server.Start(); err != nil {
log.Fatal(err)
}
}
Creating a Client
package main
import (
"context"
"log"
"github.com/virgoC0der/go-mcp"
"github.com/virgoC0der/go-mcp/internal/types"
)
func main() {
// Create client
client, err := mcp.NewClient(&types.ClientOptions{
ServerAddress: "localhost:8080",
Type: "http", // or "sse", "websocket"
UseJSONRPC: true,
SubscribeToNotifications: true,
})
if err != nil {
log.Fatal(err)
}
// Connect to server
ctx := context.Background()
if err := client.Connect(ctx); err != nil {
log.Fatal(err)
}
defer client.Close()
// Get service interface
service := client.Service()
// Use service
result, err := service.ListPrompts(ctx, "")
if err != nil {
log.Fatal(err)
}
log.Printf("Available prompts: %v", result.Prompts)
// Get next page if available
if result.NextCursor != "" {
nextPage, err := service.ListPrompts(ctx, result.NextCursor)
if err != nil {
log.Fatal(err)
}
log.Printf("Next page prompts: %v", nextPage.Prompts)
}
}
Response Structure
JSON-RPC Responses
For JSON-RPC endpoints, responses follow the JSON-RPC 2.0 specification with a unified response handling system across all transport layers:
type JSONRPCResponse struct {
JSONRPC string `json:"jsonrpc"`
ID interface{} `json:"id"`
Result interface{} `json:"result,omitempty"`
Error *JSONRPCError `json:"error,omitempty"`
}
type JSONRPCError struct {
Code int `json:"code"`
Message string `json:"message"`
Data interface{} `json:"data,omitempty"`
}
JSON-RPC Success Response Example:
{
"jsonrpc": "2.0",
"id": 1,
"result": {
"prompts": [
{
"name": "example_prompt",
"description": "An example prompt"
}
],
"nextCursor": ""
}
}
JSON-RPC Error Response Example:
{
"jsonrpc": "2.0",
"id": 1,
"error": {
"code": -32602,
"message": "Invalid params",
"data": "Missing required parameter: name"
}
}
The library provides a unified response handling system that works across HTTP, WebSocket, and stdio transport layers, ensuring consistent error handling and response formatting.
Examples
- Echo Server - A simple echo server example
- Weather Service - A weather service example using OpenWeatherMap API
- App Launcher - A macOS application launcher example with stdio server support
API Documentation
Server Interface
Servers must implement the types.Server
interface:
type Server interface {
MCPService
Initialize(ctx context.Context, options any) error
Start() error
Shutdown(ctx context.Context) error
}
The MCPService
interface defines the core functionality:
type MCPService interface {
// ListPrompts returns a list of available prompts with pagination support
ListPrompts(ctx context.Context, cursor string) (*PromptListResult, error)
// GetPrompt retrieves a specific prompt by name with optional arguments
GetPrompt(ctx context.Context, name string, args map[string]any) (*PromptResult, error)
// ListTools returns a list of available tools with pagination support
ListTools(ctx context.Context, cursor string) (*ToolListResult, error)
// CallTool invokes a specific tool by name with arguments
CallTool(ctx context.Context, name string, args map[string]any) (*CallToolResult, error)
// ListResources returns a list of available resources with pagination support
ListResources(ctx context.Context, cursor string) (*ResourceListResult, error)
// ReadResource reads the content of a specific resource
ReadResource(ctx context.Context, uri string) (*ResourceContent, error)
// ListResourceTemplates returns a list of available resource templates
ListResourceTemplates(ctx context.Context) ([]ResourceTemplate, error)
// SubscribeToResource subscribes to changes on a specific resource
SubscribeToResource(ctx context.Context, uri string) error
}
Client Interface
Clients access services through the types.Client
interface:
type Client interface {
// Connect establishes a connection to the server
Connect(ctx context.Context) error
// Close terminates the connection
Close() error
// Service returns the underlying MCPService interface
Service() MCPService
}
Contributing
Contributions are welcome! Please follow these steps:
- Fork the repository
- Create your feature branch (
git checkout -b feature/amazing-feature
) - Commit your changes (
git commit -m 'Add some amazing feature'
) - Push to the branch (
git push origin feature/amazing-feature
) - Open a Pull Request
License
This project is licensed under the MIT License - see the LICENSE file for details
Updated to MCP Specification 2025-03-26
This library has been updated to support the Model Context Protocol (MCP) 2025-03-26 specification. Major updates include:
New Features
- Complete JSON-RPC Support: Implemented JSON-RPC 2.0 API endpoints compliant with the latest MCP specification
- Enhanced Multimodal Content: Support for text, image, audio, and embedded resource content transmission
- Pagination Support: All list APIs now support cursor-based pagination
- Resource Templates: Support for parameterized resource URI templates
- Resource Subscriptions: Support for client subscriptions to resource change notifications
- Rich Server Capabilities: More granular server capability declarations
- Unified Response Handling: Standardized response handling across different transport layers
Backward Compatibility
- Preserved original REST API endpoints to ensure compatibility with older clients
- Notification system supports both new and old notification formats
Example Usage
// Create server
server, err := mcp.NewServer(service, &types.ServerOptions{
Address: ":8080",
Capabilities: &types.ServerCapabilities{
Prompts: &types.PromptCapabilities{
ListChanged: true,
},
Resources: &types.ResourceCapabilities{
ListChanged: true,
Subscribe: true,
Templates: true,
},
},
})
// Create client
client, err := mcp.NewClient(&types.ClientOptions{
ServerAddress: "localhost:8080",
Type: "http",
UseJSONRPC: true,
SubscribeToNotifications: true,
})
// Get prompts list (with pagination support)
result, err := client.Service().ListPrompts(ctx, "")
// Get next page
nextPage, err := client.Service().ListPrompts(ctx, result.NextCursor)
// Subscribe to resource changes
err := client.Service().SubscribeToResource(ctx, "file:///example.txt")
相关推荐
Converts Figma frames into front-end code for various mobile frameworks.
I find academic articles and books for research and literature reviews.
Confidential guide on numerology and astrology, based of GG33 Public information
Embark on a thrilling diplomatic quest across a galaxy on the brink of war. Navigate complex politics and alien cultures to forge peace and avert catastrophe in this immersive interstellar adventure.
Advanced software engineer GPT that excels through nailing the basics.
💬 MaxKB is an open-source AI assistant for enterprise. It seamlessly integrates RAG pipelines, supports robust workflows, and provides MCP tool-use capabilities.
Micropython I2C-based manipulation of the MCP series GPIO expander, derived from Adafruit_MCP230xx
MCP server to provide Figma layout information to AI coding agents like Cursor
The all-in-one Desktop & Docker AI application with built-in RAG, AI agents, No-code agent builder, MCP compatibility, and more.
Reviews

user_FzfKwhWe
I've been using go-mcp for a while, and it has significantly improved my workflow. The project is well-documented, and the creator, virgoC0der, has done an excellent job with the implementation. It's fast, reliable, and easy to integrate into existing systems. Highly recommend checking it out on GitHub!