build-mcp-server

Design and implement MCP servers for Claude with deployment and tool pattern selection.

Updated Mar 11, 2026
One-click install
npx skills add https://github.com/selfagency/agentsy --skill build-mcp-server-selfagency
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: build-mcp-server
Source: https://github.com/selfagency/agentsy/tree/main/.agents/skills/build-mcp-server
Command: npx skills add https://github.com/selfagency/agentsy --skill build-mcp-server-selfagency

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Building an MCP server for Claude fails most often at the design stage, when the server’s shape, deployment model, and tool patterns are chosen incorrectly, causing costly rewrites and poor tool selection later.

Core Features & Use Cases

  • Use-case discovery first: Interrogates the integration goals (upstream system, audience, number of actions, UI/input needs, and auth requirements) before proposing any scaffold.
  • Deployment model recommendation: Chooses between remote HTTP (default), elicitation, MCP apps (widgets), MCPB (bundled local server), or local stdio prototypes based on practical constraints.
  • Tool design patterns: Recommends either one-tool-per-action for small surfaces or a search + execute pattern for large APIs to preserve context quality.
  • Framework selection & handoff: Directs the next step toward the right implementation path using the TypeScript SDK or FastMCP, then instructs how to scaffold and verify.

Quick Start

Ask the AI: Build an MCP server that wraps my REST API and exposes tools to Claude Desktop via remote HTTP with OAuth, using search + execute if there are many endpoints.

Frequently Asked Questions about build-mcp-server

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

FAQPage Schema
How do I build an MCP server that wraps a REST API for Claude?

A search + execute pattern is recommended for large APIs to preserve context quality, while one-tool-per-action suits small surfaces. This choice prevents costly rewrites and poor tool selection during MCP server design.

When should I use remote HTTP versus local stdio for MCP server deployment?

Remote HTTP is the default for MCP server deployment, while local stdio is used for prototypes. Choose remote HTTP for broad access with OAuth, or MCPB for bundled local server distribution.

Does the MCP server use the TypeScript SDK or FastMCP for implementation?

Framework selection between the TypeScript SDK and FastMCP occurs after determining the deployment model and tool design patterns, directing the implementation path before scaffolding the MCP server.

How do I expose tools, resources, and prompts in my MCP server design?

Exposing tools, resources, and prompts requires interactive use-case interrogation to define upstream systems, audience, and auth requirements before selecting a tool schema pattern and scaffolding the server.

How do I verify an MCP server after scaffolding it for Claude integration?

Verify the MCP server using a tools/list workflow to ensure reliable Claude integration after scaffolding. This confirms that the exposed tools, resources, and prompts are correctly structured.