MCP Server Builder
Paste the following prompt into your AI chat to install this skill:
Please install @org-02qudk26/cn-mcp-builder according to https://skillhub.cn/install/skillhub.md
About this skill
What It Solves
MCP servers often become thin API wrappers: vague tool descriptions, unhelpful errors, and unstructured output make it hard for LLMs to call external services reliably. mcp-builder turns those development constraints into a practical workflow for exposing external APIs as composable tools.
How It Works
The guide organizes development into four stages:
- Research and planning: read the MCP specification, transports, and tool/resource definitions; compare broad API coverage with task-specific workflow tools.
- Implementation: prefers TypeScript with streaming HTTP for remote servers and Python with stdio for local servers; uses Zod/Pydantic for input schemas, defines outputSchema where possible, and adds hints such as readOnlyHint and idempotentHint.
- Review and testing: checks DRY code, error handling, and type coverage; verifies builds and tests with the MCP inspector.
- Evaluation: creates 10 independent, read-only, verifiable real-world questions to test whether an LLM can complete complex tasks.
Limits
It is a development guide rather than a replacement for SDKs, frameworks, or backend services. For unstable APIs, complex authentication, or write-heavy workflows, add business-level encapsulation and permission controls before exposing the tools through MCP.
Use Cases
- Name and paginate GitHub API tools for stable repo queries.
- Define Zod input and outputSchema for a weather MCP tool.
- Use MCP inspector to test errors and readOnlyHint behavior.
- Create 10 read-only eval questions for multi-step API tasks.
Best For
- Backend engineers building GitHub API MCP servers with controlled tool behavior.
- Full-stack engineers wrapping weather services as MCP tools.
- Application engineers reviewing local MCP server error handling.
- Algorithm engineers validating multi-step LLM queries through MCP tools.
Related Skills
Automatically indexes Gradle-cached AAR/JAR dependency classes and returns library coordinates, versions, and public APIs by fully qualified name, using only the Python standard library.
Codifies AMT and YourMT3 training conventions, script patterns, hyperparameters, precision, checkpoints, and NaN safeguards.
Retrieve relevant chunks from a customer-managed PKM dataset by dataset_id and return concise, source-annotated answers.
Convert PRDs, user stories, or functional specs into prioritized test-point checklists covering functional, business-rule, boundary, exception, and non-functional dimensions.