Summary
This is your blueprint for building Model Context Protocol servers that let LLMs talk to external APIs. It walks you through four phases: research (including when to favor comprehensive API coverage over workflow shortcuts), implementation in TypeScript or Python, testing with MCP Inspector, and creating evaluations with 10 complex questions to verify your server actually works. The guide pushes TypeScript with streamable HTTP for remote servers, emphasizes actionable error messages, and includes specifics like using Zod for schemas and structuredContent for responses. If you're wiring up a service to work with Claude or other LLMs through MCP, this gives you the opinionated process and technical details to do it right the first time.
Install to Claude Code
npx -y skills add warpdotdev/oz-skills --skill mcp-builder --agent claude-codeInstalls into .claude/skills of the current project.
Keep your Mac awake while Claude Code and 40+ AI agents run. Sleeps when they're idle.
One time payment $9 →
Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →
Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →
On Capafy, your Skill runs online 24/7 as an agent product, and you get paid every time someone uses it.
Start earning →
Files
SKILL.md
Select a file.
Featured
Keep your Mac awake while Claude Code and 40+ AI agents run. Sleeps when they're idle.
One time payment $9 →Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →On Capafy, your Skill runs online 24/7 as an agent product, and you get paid every time someone uses it.
Start earning →
