ck:mcp-builder

Build MCP servers for LLM interaction with external services and APIs.

1|Updated Mar 6, 2026
One-click install
npx skills add https://github.com/quanganh208/cookmate --skill ck-mcp-builder-quanganh208
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ck:mcp-builder
Source: https://github.com/quanganh208/cookmate/tree/main/.opencode/skills/mcp-builder
Command: npx skills add https://github.com/quanganh208/cookmate --skill ck-mcp-builder-quanganh208

SYSTEM DOCUMENTATION & REQUIREMENTS

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

What problem does it solve?

Build production-ready MCP servers that let LLMs reliably interact with external services and APIs, removing the friction of designing agent-friendly tool workflows and transport lifecycles.

Core Features & Use Cases

  • Language-specific guides: Implementation guidance for Python (FastMCP) and Node/TypeScript MCP SDKs with Pydantic and Zod examples.
  • Tool and schema design: Patterns for designing validated, discoverable tools, pagination, response formats (markdown/json), and actionable error messages.
  • Transports & evaluation: Support for stdio, SSE, and HTTP transports plus an evaluation harness and checklist for creating stable, read-only QA evaluations.

Quick Start

Generate a FastMCP Python server scaffold that exposes validated, paginated read-only tools with Pydantic models and both markdown and JSON output formats.

Frequently Asked Questions about ck:mcp-builder

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

FAQPage Schema
How do I build an MCP server for LLM-API integration?

An MCP server exposes validated, discoverable tools that let LLMs interact with external services and APIs. This skill generates server scaffolds supporting FastMCP (Python) and Node/TypeScript SDKs with transport lifecycle management for stdio, SSE, and HTTP.

Does the MCP server SDK support input schema validation with Pydantic and Zod?

Yes, MCP server SDK implementations support input schema validation using Pydantic for Python and Zod for Node/TypeScript. This ensures LLM tool invocations receive properly validated arguments before executing external API interactions.

How do I handle response formats and CHARACTER_LIMIT when exposing tools to LLMs?

Handle response formats and CHARACTER_LIMIT by configuring the MCP server to output markdown or JSON and truncate responses appropriately. The server implementation includes actionable error messages to guide LLMs when limits are exceeded.

Can I use FastMCP to create paginated, read-only tools for external APIs?

Yes, you can use FastMCP to create paginated, read-only tools for external APIs. The skill provides implementation patterns for designing discoverable tools with Pydantic models, supporting both markdown and JSON output formats for LLM consumption.

What is the best way to evaluate MCP server stability for LLM tool workflows?

The best way to evaluate MCP server stability is using an evaluation harness and checklist to create stable, read-only QA evaluations. This ensures your validated tools and transport lifecycles perform reliably during LLM interactions.

What transports are supported for MCP server lifecycle management?

MCP server lifecycle management supports stdio, SSE, and HTTP transports. This allows flexible deployment options for exposing validated tools and paginated responses to LLMs across different runtime environments.