Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).
Install with Tessl CLI
npx tessl i github:boisenoise/skills-collections --skill mcp-builder93
Does it follow best practices?
If you maintain this skill, you can automatically optimize it using the tessl CLI to improve its score:
npx tessl skill review --optimize ./path/to/skillEvaluation — 94%
↑ 2.23xAgent success when using this skill
Validation for skill structure
TypeScript MCP server project setup and tool registration
Server naming convention
100%
100%
Tool naming with prefix
0%
100%
Uses registerTool API
0%
100%
Zod input schemas
100%
100%
Zod strict mode
0%
100%
Tool annotations present
0%
100%
Correct SDK import path
100%
100%
Project structure
100%
100%
package.json type module
100%
100%
tsconfig strict mode
100%
100%
tsconfig module config
100%
100%
Stdio transport
100%
100%
Python MCP server with Pydantic validation and response formatting
Server naming convention
0%
100%
FastMCP initialization
100%
100%
Tool naming with prefix
0%
100%
Pydantic BaseModel inputs
25%
100%
Pydantic v2 patterns
0%
100%
ConfigDict settings
0%
100%
Dual response format
0%
100%
Pagination metadata
0%
100%
Uses httpx not requests
50%
100%
Async tool functions
0%
100%
Tool annotations
0%
100%
Shared API utility
66%
100%
Streamable HTTP transport and response truncation
Streamable HTTP transport
0%
100%
Stateless transport config
0%
100%
JSON response enabled
0%
100%
New transport per request
100%
100%
CHARACTER_LIMIT constant
80%
100%
Response truncation logic
100%
100%
Output schema defined
0%
0%
structuredContent in response
0%
100%
Actionable error messages
25%
100%
Error in result object
0%
100%
Localhost binding
0%
0%
Multi-tool code reuse and error handling patterns
Shared API client function
100%
100%
Centralized error handler
50%
100%
Consistent service prefix
0%
100%
Domain-specific naming
100%
100%
Comprehensive tool descriptions
100%
100%
API key from environment
100%
100%
API key validation on startup
100%
100%
Tool annotations on all tools
0%
100%
Shared formatting logic
0%
100%
Error in result objects
100%
100%
Organized file structure
0%
100%
MCP server evaluation creation in XML format
Valid XML structure
100%
100%
Exactly 10 questions
0%
100%
Read-only operations
100%
100%
Independent questions
87%
100%
Single verifiable answers
0%
100%
Stable answers
20%
30%
Multi-tool complexity
30%
100%
No keyword giveaways
12%
62%
Human-readable answers
0%
100%
Diverse answer types
0%
100%
Realistic questions
90%
70%
Table of Contents
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.