| Silver Wizard Software

Overview

A collection of Model Context Protocol (MCP) servers that add domain-specific capabilities to Claude Code. Plug-and-play intelligence for common development tasks.

Available Servers

gcc-memory - Persistent Memory

Persistent session memory for Claude Code agents. Store context, blockers, and checkpoints that survive session restarts.

Learn more →

gpyqt-instrument - PyQt Testing

UI testing and instrumentation for PyQt applications. Connect to running apps, capture widget trees, automate interactions.

Learn more →

gmsg - Multi-Agent Messaging

Messaging infrastructure for multi-agent systems. Send messages, coordinate work, manage conferences between Claude Code instances.

Coming soon to silverwizard.dev

gmesh - Conference Mode

Conference management for coordinated multi-agent workflows. Register agents, broadcast messages, track active conferences.

Coming soon to silverwizard.dev

What is MCP?

Model Context Protocol is Anthropic’s standard for extending Claude with external tools. MCP servers:

Provide tools Claude can call directly
Run as separate processes (secure isolation)
Use standardized communication protocol
Work across all Claude interfaces (Code, API, Desktop)

Why MCP?

For Developers

No API integration - Install server, tools appear automatically
Type-safe - Tools have schemas, Claude knows how to use them
Composable - Mix and match servers for your workflow
Local-first - Servers run on your machine, your data stays private

For Tool Creators

Standard protocol - Build once, works everywhere
Clear contracts - JSON Schema for tool definitions
Easy distribution - npm/pip packages, MCP registries
Community - Growing ecosystem of compatible tools

Getting Started

1. Install an MCP Server

# Example: Install gcc-memory
npm install -g gcc-memory

# Or with pip
pip install gcc-memory

2. Add to Claude Code

# Option A: Auto-configure
claude code mcp add gcc-memory

# Option B: Manual config
# Edit ~/.claude/mcp_config.json

3. Start Using

Tools appear automatically in Claude’s toolkit. Just ask Claude to use them:

"Claude, save my current progress to memory with keywords 'last,completed'"
"Claude, connect to the PyQt app and take a screenshot"

Building Your Own MCP Server

# Example: Simple MCP server in Python
from mcp import MCPServer, Tool

server = MCPServer(name="my-server")

@server.tool()
def my_tool(param: str) -> str:
    """Description of what this tool does"""
    return f"Processed: {param}"

if __name__ == "__main__":
    server.run()

See MCP Server Development Guide for full details.

MCP Registries

Discover more MCP servers:

Use Cases

AI-Assisted Development

Add tools for: code search, refactoring, test generation, documentation, deployment.

System Integration

Connect Claude to: databases, APIs, cloud services, monitoring systems, CI/CD pipelines.

Domain Expertise

Give Claude specialized knowledge: medical terminology, legal standards, financial analysis, scientific computation.

Workflow Automation

Automate: git operations, file management, data processing, report generation, notification sending.

Architecture

┌─────────────────┐
│  Claude Code    │
│                 │
│  ┌───────────┐  │
│  │  MCP      │  │
│  │  Client   │  │
│  └─────┬─────┘  │
└────────┼────────┘
         │ JSON-RPC
    ┌────┴────┐
    │         │
┌───▼───┐ ┌──▼────┐
│ gcc-  │ │ gpyqt │
│ memory│ │ inst. │
└───────┘ └───────┘

MCP uses JSON-RPC 2.0 over stdio/HTTP for communication. Servers run as separate processes, providing isolation and language flexibility.

Best Practices

Start small - Install 1-2 servers, learn them well before adding more
Read tool docs - Understanding parameters and return values improves results
Combine servers - MCP servers compose naturally (memory + testing + messaging)
Contribute back - Found a bug? Submit a PR. Built a server? Share it.