mcp-builder

Create MCP servers in Python or TypeScript with validated tool schemas.

6|1|Updated Apr 7, 2026
One-click install
npx skills add https://github.com/kmshihab7878/claude-code-setup --skill mcp-builder-kmshihab7878
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: mcp-builder
Source: https://github.com/kmshihab7878/claude-code-setup/tree/main/skills/mcp-builder
Command: npx skills add https://github.com/kmshihab7878/claude-code-setup --skill mcp-builder-kmshihab7878

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires anthropic, mcp, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This guide removes guesswork and friction from building robust MCP (Model Context Protocol) servers so LLMs can reliably interact with external services via well-designed tools, clear schemas, and predictable transports.

Core Features & Use Cases

  • Language-specific implementation guides for TypeScript (MCP SDK) and Python (FastMCP) with project structure, schema examples, and build/run patterns.
  • Tool design and validation guidance including tool naming, Zod/Pydantic input schemas, structuredContent outputs, annotations (readOnlyHint, destructiveHint, idempotentHint), pagination, and actionable error messages.
  • Transports, testing & evaluation covering stdio and streamable HTTP transports, automated evaluation harness with Anthropic, and a reproducible evaluation XML format for verifying agent performance.

Quick Start

Create a new MCP server project following the TypeScript or Python guide, register your core tools with explicit input/output schemas, and run the provided evaluation harness to verify read-only task performance.

Frequently Asked Questions about mcp-builder

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build a production-grade MCP server in Python or TypeScript?

To build a production-grade MCP server, use FastMCP for Python or the MCP SDK for TypeScript to wrap external APIs, define tool schemas with Pydantic or Zod, and configure stdio or streamable HTTP transports.

How do I validate MCP tool inputs and design structured schemas?

Validate MCP tool inputs by defining strict schemas using Pydantic in Python or Zod in TypeScript, applying annotations like readOnlyHint or destructiveHint, and returning structuredContent outputs with pagination metadata.

What's the best way to evaluate MCP server tool performance with Anthropic?

Evaluate MCP server tool performance by running the provided automated evaluation harness with Anthropic, generating reproducible evaluation XML formats to verify agent task completion and read-only performance.

Does this MCP server guide support both stdio and streamable HTTP transports?

Yes, this MCP server guide supports configuring both stdio and streamable HTTP transports, providing language-specific implementation patterns to expose validated tools and resources over your chosen network protocol.

How should I handle security and error messages in MCP tool design?

Handle security in MCP tool design by returning actionable, security-aware error messages, applying tool annotations like idempotentHint, and ensuring strict schema validation to prevent unpredictable LLM interactions.