mcp-coded-tools

Generate discoverable code from MCP servers for AI agent tool usage.

Why?

The Model Context Protocol (MCP) lets AI agents connect to external tools and data. However, as Anthropic's engineering post explains, loading hundreds or thousands of tool definitions directly into an agent's context window is inefficient:

• Tool definitions consume excessive tokens (100K+ tokens for large tool sets)
• Intermediate results flow through the context window repeatedly
• Agents become slower and more expensive at scale

The solution: Present MCP servers as code APIs that agents can discover and use through a filesystem interface. This approach reduces token usage by up to 98.7% while enabling more powerful agent workflows.

The problem: Manually writing wrapper code for each MCP tool is tedious.

mcp-coded-tools: Automatically generate discoverable Python code from any MCP server.

Installation

pip install mcp-coded-tools

Quick Start

CLI Usage

# Generate code from popular MCP servers
mcp-coded-tools generate \
  --command "npx -y @modelcontextprotocol/server-github" \
  --output ./servers \
  --server-name github

# Generate from multiple servers for complete workflows
mcp-coded-tools generate \
  --command "npx -y @modelcontextprotocol/server-github" \
  --command "npx -y @modelcontextprotocol/server-postgres" \
  --command "python slack_mcp_server.py" \
  --output ./servers

# Watch mode for development (auto-regenerate on changes)
mcp-coded-tools generate \
  --command "python ./my_mcp_server.py" \
  --output ./tools \
  --watch

💡 See POPULAR_MCP_SERVERS.md for 50+ popular servers and real-world use cases!

💡 See WATCH_MODE.md for auto-regeneration during development!

Python API

import asyncio
from mcp-coded-tools import MCPCodeGenerator

async def main():
    generator = MCPCodeGenerator()

    # Connect to MCP server and generate code
    await generator.connect_and_scan([
        "npx", "-y", "@modelcontextprotocol/server-gdrive"
    ])

    generator.generate_code(
        output_dir="./servers",
        server_name="google_drive"
    )

    print("✓ Generated discoverable code!")

asyncio.run(main())

What Gets Generated

servers/
├── google_drive/
│   ├── __init__.py
│   ├── get_document.py
│   ├── list_files.py
│   └── ...
├── salesforce/
│   ├── __init__.py
│   ├── update_record.py
│   └── ...
└── _client.py

Each tool becomes a typed Python function:

# servers/google_drive/get_document.py
from typing import Optional, Dict, Any
from .._client import call_mcp_tool

async def get_document(
    document_id: str,
    fields: Optional[str] = None
) -> Dict[str, Any]:
    """
    Retrieves a document from Google Drive
    """
    return await call_mcp_tool(
        'gdrive_getDocument',
        {'documentId': document_id, 'fields': fields}
    )

Agent Usage

Agents discover tools by exploring the filesystem:

# Agent lists available servers
import os
servers = os.listdir('./servers')
# ['google_drive', 'salesforce', ...]

# Agent reads a specific tool
with open('./servers/google_drive/get_document.py') as f:
    tool_def = f.read()
    # Understands parameters, types, description

# Agent writes code to use tools
import servers.google_drive as gdrive
import servers.salesforce as sf

async d

... [View full README on GitHub](https://github.com/bluman1/mcp-coded-tools#readme)