MCP HTTP Client Server

A Model Context Protocol (MCP) server for making HTTP requests, GraphQL queries, and raw TCP/Telnet connections from AI assistants.

Installation

Choose your preferred installation method:

Option 1: npm (Recommended for easy setup)

npm install -g @mcp/http-client

Option 2: Homebrew (macOS/Linux)

brew tap ferPrieto/mcp-http-client
brew install mcp-http-client

Option 3: GitHub Packages (for Kotlin/JVM projects)

Add the repository and dependency to your build.gradle.kts:

repositories {
    maven {
        url = uri("https://maven.pkg.github.com/ferPrieto/MCP-Http-Client")
        credentials {
            username = project.findProperty("gpr.user") as String? ?: System.getenv("GITHUB_ACTOR")
            password = project.findProperty("gpr.key") as String? ?: System.getenv("GITHUB_TOKEN")
        }
    }
}

dependencies {
    implementation("ferprieto.mcp:httpclient:1.0.0")
}

Note: GitHub Packages requires authentication. Generate a Personal Access Token with read:packages scope.

Option 4: Build from Source

git clone https://github.com/ferPrieto/MCP-Http-Client.git
cd MCP-Http-Client
./gradlew clean build

This generates build/libs/mcp-http-client-all.jar.

Configuration

Add to your MCP client configuration file:

• Cursor: ~/.cursor/mcp.json
• Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json

If installed via npm:

{
  "mcpServers": {
    "http-client": {
      "command": "npx",
      "args": ["@mcp/http-client"]
    }
  }
}

If installed via Homebrew:

{
  "mcpServers": {
    "http-client": {
      "command": "mcp-http-client"
    }
  }
}

If using JAR directly:

{
  "mcpServers": {
    "http-client": {
      "command": "java",
      "args": ["-jar", "/path/to/mcp-http-client-all.jar"]
    }
  }
}

Available Functionalities

1. HTTP/HTTPS Requests

Make HTTP requests with any method (GET, POST, PUT, DELETE, PATCH, HEAD, OPTIONS).

Examples:

Simple GET request:

"Get data from https://api.example.com/posts/1"

GET with query parameters:

"GET https://api.github.com/search/repositories with params q=kotlin and sort=stars"

POST with JSON body:

"POST to https://api.example.com/users with body {name: 'John', email: '[email protected]'}"

POST with headers and body:

"Send POST request to https://api.example.com/users with header Authorization: Bearer token123 and JSON body {name: 'Alice'}"

PUT request:

"Update user at https://api.example.com/users/42 with PUT method and body {status: 'active'}"

DELETE request:

"Delete resource at https://api.example.com/users/42 with Authorization header Bearer token123"

2. GraphQL Queries

Execute GraphQL queries with variables and custom headers.

Examples:

Simple GraphQL query:

"Query GraphQL at https://api.example.com/graphql: { user(id: 1) { name email } }"

GraphQL with variables:

"GraphQL query to https://api.example.com/graphql with query: { user(id: $userId) { name } } and variable userId=123"

GraphQL with operation name:

"Execute GraphQL query at https://api.example.com/graphql with query: { users { name } } and operationName=GetUsers"

GraphQL with authentication:

"Query GraphQL at https://api.example.com/graphql with header Authorization: Bearer token and query: { me { name email } }"

3. TCP/Telnet Connections

Establish raw TCP socket connections for network testing.

Examples:

Basic TCP connection:

"Connect via telnet to 192.168.1.1 port 8080"

TCP with custom timeout:

"Test TCP connection to localhost:3000 with timeout 10 seconds"

TCP with message:

"Connect to example.com port 80 and send message: GET / HTTP/1.1"

TCP connection test:

"Check if port 8080 is open on localhost using telnet"

Performance Caching

Implements intelligent LRU (Least Recently Used) cache with TTL expiration that automatically caches GET requests, delivering 13.7x speedup (92% faster) on repeated requests. Inspired by Bruno API client, the cache is thread-safe, limits memory usage to 100 entries, and expires data after 5 minutes to balance performance with data freshness.

Benchmark Results (10 identical GET requests):

• Without Cache: 1537ms (153ms per request)
• With Cache: 112ms (11ms per request)
• Speedup: 13.7x faster with 9/10 requests served from cache

License

MIT License