What is a Model Context Protocol (MCP) server?

MCP servers are a standardized way to connect AI assistants (like Claude, ChatGPT, or custom agents) to external data sources and tools. They expose "tools" that LLMs can call to perform actions like searching databases, calling APIs, or reading files. According to the MCP specification, servers provide: (1) Tools - functions the LLM can execute, (2) Resources - data the LLM can read, and (3) Prompts - pre-defined templates. This creates a clean separation between the AI model and external services.

Should I use FastMCP (Python) or the TypeScript SDK?

Use FastMCP if you're building in Python - it has a simpler decorator-based API and handles validation automatically. Use the TypeScript SDK if you need Node.js integration or prefer TypeScript's type system. FastMCP is generally faster to prototype with, while the TypeScript SDK gives you more control over server behavior. Both are production-ready and support the full MCP specification.

How do I handle authentication in MCP servers?

Best practice is to store credentials in environment variables (e.g., GITHUB_TOKEN) and load them when the server starts. Never hardcode credentials in your server code. For user-specific auth (OAuth), you can prompt users to configure credentials in their MCP settings file (~/.config/claude/claude_desktop_config.json for Claude Code). Validate authentication before executing any tools and return clear error messages if auth fails.

Can MCP servers be used with multiple AI assistants?

Yes! MCP is a standard protocol supported by Claude Code, Cursor, Cline, Zed, and 11+ frameworks. Once you build an MCP server, it works across all compatible tools. Users install it with npx skills install [your-server] or add it to their MCP config file. This means you only need to build your server once.

How do I test my MCP server?

Test MCP servers by: (1) Unit testing individual tools with mock data, (2) Integration testing with the real API in a sandbox environment, (3) Manual testing in Claude Code or Cursor by installing your server and trying different prompts. The MCP Inspector tool (from the official SDK) lets you test tool calls interactively before connecting to a client.

textcodingintermediate

MCP SERVER DEVELOPMENT PROMPT▌

Create a Model Context Protocol (MCP) server with FastMCP or TypeScript SDK following best practices

Recommended Model: Claude Opus 4.7

Est. Tokens: 1,200

Prompt Generator

Generate Your Prompt

Fill in the fields below to generate an optimized prompt based on best practices

Service/API Name *

What external service or API will this MCP server interact with?

Primary Use Case *

What will users accomplish with this MCP server?

Implementation Language *

Python (FastMCP) or TypeScript (MCP SDK)?

Required Features *

What tools/capabilities must this server provide?

Examples

GitHub MCP Server

Input Variables

serviceName:GitHub

useCase:Search, create, and manage GitHub issues and pull requests from AI agents

language:python

requiredFeatures:- Search issues - Create issue - Update issue status - List pull requests

requiresAuth:

authMethod:token

Frequently Asked Questions

Best Practices

Critical Best Practices

Use Tool Preambles

critical

Provide clear progress updates before calling tools so users can follow the agent's work. This improves transparency and user experience during long operations.

Example:

"Before calling search tools, explain: "I'm going to search our knowledge base for recent updates on this topic...""

Source: openai

Define Persistence Behavior

critical

Specify when the agent should continue working vs. asking for clarification. Set clear stop conditions to avoid premature termination or endless loops.

Example:

"Only stop when: (1) Task is fully complete, (2) User explicitly asks to stop, or (3) You encounter an unrecoverable error."

Source: openai

Define Error Handling Strategy

critical

Specify how agents should handle errors, missing data, and edge cases. Should they retry, escalate, or proceed with uncertainty?

Source: custom

Set Safety Boundaries

critical

Define which actions require confirmation (destructive operations, external API calls, file deletions) vs. which can be done autonomously.

Example:

"Ask before: deleting files, making API calls that cost money, or modifying production databases."

Source: claude

Recommended Best Practices

Enable Parallel Tool Execution

recommended

Allow agents to call multiple independent tools simultaneously for faster execution. Critical for research and multi-step workflows.

Example:

"When gathering context, run file reads and searches in parallel rather than sequentially."

Source: claude

Related Templates

AI Agent Architecture Design Prompt

text / coding

Keywords

mcp server developmentmodel context protocolfastmcpmcp typescript sdkclaude mcpai agent toolsmcp server tutorial