mcp-developer

Implement MCP server and client integrations over JSON-RPC 2.0.

1|1|Updated Feb 2, 2026
One-click install
npx skills add https://github.com/corticalstack/claude-code-flow --skill mcp-developer-corticalstack
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: mcp-developer
Source: https://github.com/corticalstack/claude-code-flow/tree/main/.claude/skills/mcp-developer
Command: npx skills add https://github.com/corticalstack/claude-code-flow --skill mcp-developer-corticalstack

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill removes guesswork when building Model Context Protocol (MCP) servers and clients that must reliably connect AI assistants to real tools and data sources.

Core Features & Use Cases

  • Protocol-ready implementations: Implement MCP’s JSON-RPC 2.0 flows (initialize handshake, tools/resources/prompts, notifications, and tasks where applicable).
  • Transport correctness: Set up stdio for local subprocess servers and Streamable HTTP for remote servers (with correct session handling).
  • Production-grade safety: Enforce schema validation (Zod/Pydantic), strict tool naming rules, secure auth expectations, and safe error handling to avoid leaking internals.

Quick Start

Use this skill when you need a TypeScript or Python MCP server that exposes validated tools and resources over Streamable HTTP for a Claude-based workflow.

Frequently Asked Questions about mcp-developer

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

FAQPage Schema
How do I build an MCP server that exposes validated tools over JSON-RPC 2.0?

To build an MCP server, you implement JSON-RPC 2.0 flows including the initialize handshake, tool/resource/prompt handling, and strict schema validation using Zod or Pydantic to ensure safe tool exposure.

What is the difference between stdio and Streamable HTTP transports in MCP?

Stdio transport connects local subprocess servers directly, while Streamable HTTP handles remote servers with correct session lifecycle management and capabilities negotiation over the network.

How do I validate tool schemas and prevent internal data leaks in MCP implementations?

Validate tool schemas using Zod or Pydantic and enforce strict tool naming rules alongside secure error wrapping to prevent leaking internal data during MCP protocol interactions.

Can I use MCP clients to connect AI assistants to external resources via Streamable HTTP?

Yes, MCP clients can connect AI assistants to remote external resources by implementing Streamable HTTP transport with proper session handling, capabilities negotiation, and secure auth practices.

What are the protocol constraints for tool annotations and naming rules in MCP?

MCP spec constraints require strict compliance with tool annotations and naming rules during capabilities negotiation and tool handling to maintain protocol correctness and production-grade safety.

Why does my MCP server fail during the initialize handshake over JSON-RPC?

MCP initialize handshake failures often stem from incorrect JSON-RPC 2.0 protocol logging, improper capabilities negotiation, or lacking strict schema validation for tool and resource definitions.