mcp-builder

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

28|5|Updated Jun 24, 2025
One-click install
npx skills add https://github.com/thevibeworks/claude-code-docs --skill mcp-builder-thevibeworks
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: mcp-builder
Source: https://github.com/thevibeworks/claude-code-docs/tree/main/content/github/skills/skills/mcp-builder
Command: npx skills add https://github.com/thevibeworks/claude-code-docs --skill mcp-builder-thevibeworks

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill helps you design and implement high-quality MCP (Model Context Protocol) servers so LLM agents can reliably use tools to interact with external services and complete real tasks.

Core Features & Use Cases

  • MCP server architecture guidance: Plan tool vs API coverage, naming conventions, context management patterns, and transport selection for stdio vs streamable HTTP.
  • Tool and schema design: Create clear input/output schemas using Zod (TypeScript) or Pydantic (Python), including structured output expectations and actionable error messages.
  • Quality and verification workflow: Review build/test steps (including MCP Inspector) and create evaluation questions to validate tool effectiveness with read-only operations.
  • Practical implementation paths: Follow language-specific guides for TypeScript (TypeScript SDK, WebFetch patterns, Zod schemas) or Python (FastMCP, Pydantic models) during implementation.

Quick Start

Ask Claude Code to guide you through implementing an MCP server for your external API, including designing tool schemas, annotations, and a 10-question evaluation plan.

Frequently Asked Questions about mcp-builder

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

FAQPage Schema
How do I build an MCP server in TypeScript or Python for LLM tool integration?

To build an MCP server, use the TypeScript SDK with Zod schemas or Python FastMCP with Pydantic models to define tool inputs, register tools with correct annotations, and select stdio or streamable HTTP transports for external API integration.

What's the best way to design tool input schemas for an MCP server?

Designing MCP tool schemas requires using Zod in TypeScript or Pydantic in Python to validate inputs and outputs, define structured expectations, and produce actionable error messages so LLM agents can reliably interact with external services.

How does pagination and context management work in Model Context Protocol servers?

MCP server context management involves planning tool coverage versus API surface area, applying naming conventions, and handling pagination patterns to maintain stable read-only operations when LLMs call external services through registered tools.

Can I evaluate MCP server tool effectiveness before deployment?

Evaluating MCP server tools involves creating stable read-only evaluation questions in XML format and running verification workflows like the MCP Inspector to validate that tool operations work correctly with external APIs.

Do I need schema validation libraries to build a Model Context Protocol server?

Building an MCP server requires schema validation libraries like Zod for TypeScript implementations or Pydantic for Python implementations to ensure tools accept validated inputs and return structured outputs for reliable LLM agent interactions.

Why does my MCP server return errors when LLMs call external API tools?

MCP server errors during external API calls often stem from poorly specified tool schemas, missing annotations, or unhandled pagination and context, which can be resolved by following MCP protocol documentation and framework best practices.