mcp-builder

[IMPORTANT] Use TaskCreate to break ALL work into small tasks BEFORE starting — including tasks for each file read. This prevents context loss from long files. For simple tasks, AI MUST ATTENTION ask user whether to skip.

Quick Summary

Goal: Create high-quality MCP (Model Context Protocol) servers enabling LLMs to interact with external services through well-designed tools.

Workflow:

1. Deep Research — Study agent-centric design principles, MCP protocol, SDK docs, API docs exhaustively
2. Implementation Plan — Tool selection, shared utilities, input/output design, error handling
3. Implementation — Project structure, core infrastructure, tools with proper schemas/docs, language-specific best practices
4. Review & Refine — Code quality review (DRY, composability, consistency), test & build
5. Create Evaluations — 10 realistic, complex, read-only evaluation questions with verified answers

Key Rules: