typescript-mcp-server-generator
Generate a complete MCP server project in TypeScript using the MCP TypeScript SDK v2 (@modelcontextprotocol/server) with tools, resources, and proper configuration
- License 许可
- MIT
- 最近更新
- 2026-11-13
- Stars
- ★ 4.1w
- Forks
- 5.4k
01 前言 Front Matter
- name
- typescript-mcp-server-generator
- description
- Generate a complete MCP server project in TypeScript using the MCP TypeScript SDK v2 (@modelcontextprotocol/server) with tools, resources, and proper configuration
03 SKILL.md 正文
Generate TypeScript MCP Server
Create a complete Model Context Protocol (MCP) server in TypeScript using the MCP TypeScript SDK v2 with the following specifications:
Requirements
- Project Structure: Create a new TypeScript/Node.js project with proper directory structure
- NPM Packages: The v1 monolithic
@modelcontextprotocol/sdkpackage is retired. Use the focused v2 packages:@modelcontextprotocol/server— server implementation (stdio transport via the@modelcontextprotocol/server/stdiosubpath)@modelcontextprotocol/node— Node HTTP transport (NodeStreamableHTTPServerTransport), or a framework adapter:@modelcontextprotocol/express,@modelcontextprotocol/hono,@modelcontextprotocol/fastify— each adapter requires its peer framework to be installed alongside it (e.g.@modelcontextprotocol/express+express)@modelcontextprotocol/core— shared protocol schemas (import*Schemaconstants from here, not fromsdk/types.js)zod@^4.2— v2 requires Zod 4.2+; do not use zod@3
- Runtime: Node.js 20+ (v2 minimum); ESM-first with
"type": "module"(a CommonJS build is also shipped, sorequire()works if needed) - Server Type: Choose between HTTP (Streamable HTTP transport) or stdio-based server. SSE and WebSocket transports were removed in v2 — do not generate them.
- Tools: Create at least one useful tool with proper schema validation
- Error Handling: Include comprehensive error handling and validation
Implementation Details
Project Setup
- Initialize with
npm initand create package.json - Install dependencies:
@modelcontextprotocol/server,zod@^4.2, and the transport package —@modelcontextprotocol/nodefor plain Node HTTP, or a framework adapter together with its peer framework (e.g.npm install @modelcontextprotocol/express express) - Configure TypeScript with ES modules:
"type": "module"in package.json - Add dev dependencies:
tsxorts-nodefor development - Create proper .gitignore file
Server Configuration
- Use
McpServerclass from@modelcontextprotocol/serverfor high-level implementation - Set server name and version
- Choose the appropriate transport:
- HTTP (Node):
NodeStreamableHTTPServerTransportfrom@modelcontextprotocol/node - HTTP (Web Standard runtimes):
WebStandardStreamableHTTPServerTransportfrom@modelcontextprotocol/server - stdio:
StdioServerTransportfrom@modelcontextprotocol/server/stdio
- HTTP (Node):
- For HTTP: prefer a framework adapter (
@modelcontextprotocol/express, etc.) with proper middleware and error handling - Note that v2 uses Web Standard
Headers/Requesttypes; read headers withctx.http?.req?.headers.get('x-custom')
Tool Implementation
- Use
registerTool()with a config object — v1 variadic.tool()signatures are gone:server.registerTool('greet', { description: 'Greet user', inputSchema: z.object({ name: z.string() }) }, async ({ name }, ctx) => { return { content: [{ type: 'text', text: `Hello, ${name}!` }] }; }); - Schemas must be full Zod objects (
z.object({...})) — raw shape objects ({ name: z.string() }) are deprecated - Provide clear
titleanddescriptionfields - Return both
contentandstructuredContentin results - The handler's second parameter is a structured
ctxobject (replaces v1extra):ctx.mcpReq.signal,ctx.mcpReq.id,ctx.mcpReq.send(...),ctx.mcpReq.notify(...) - Implement proper error handling with try-catch blocks; use the v2 error hierarchy (
ProtocolError,SdkError,SdkHttpErrorwith.status) instead of v1McpError/StreamableHTTPError - Support async operations where appropriate
Resource/Prompt Setup (Optional)
- Add resources using
registerResource()with ResourceTemplate for dynamic URIs - Add prompts using
registerPrompt()with argument schemas (same config-object style asregisterTool()) - Consider adding completion support for better UX; note the v2
completable()wrapper order:completable(z.string(), callback).optional()(optional applied outside)
Code Quality
- Use TypeScript for type safety
- Follow async/await patterns consistently
- Implement proper cleanup on transport close events
- Use environment variables for configuration
- Add inline comments for complex logic
- Structure code with clear separation of concerns
Example Tool Types to Consider
- Data processing and transformation
- External API integrations
- File system operations (read, search, analyze)
- Database queries
- Text analysis or summarization (LLM-assisted via the multi-round
input_requiredpattern) - System information retrieval
Configuration Options
-
For HTTP Servers:
- Port configuration via environment variables
- CORS setup for browser clients
- Session management (stateless vs stateful)
- DNS rebinding protection for local servers
- Strict
Content-Typehandling: v2 rejects non-application/jsonPOST bodies
-
For stdio Servers:
- Proper stdin/stdout handling
- Environment-based configuration
- Process lifecycle management
Migrating an Existing v1 Server
- Run the official codemod first:
npx @modelcontextprotocol/codemod@latest v1-to-v2 . - Then search for
@mcp-codemod-errormarkers for the parts requiring manual judgment (transport choice, header reads, error classification) - Swap
McpError + ErrorCodechecks for the new error classes; HTTP status now lives onerror.status, noterror.code Server.createMessage(),listRoots(),sendLoggingMessage()and theroots/sampling/loggingcapability fields are deprecated in v2 — avoid them in new code
Testing Guidance
- Explain how to run the server (
npm startornpx tsx server.ts) - Provide MCP Inspector command:
npx @modelcontextprotocol/inspector - For HTTP servers, include connection URL:
http://localhost:PORT/mcp - Include example tool invocations
- Add troubleshooting tips for common issues
Additional Features to Consider
- LLM-powered tools using the multi-round
input_requiredpattern (the v2 replacement for the deprecated sampling subsystem) - User input elicitation for interactive workflows
- Dynamic tool registration with enable/disable capabilities
- Notification debouncing for bulk updates
- Resource links for efficient data references
Generate a complete, production-ready MCP server with comprehensive documentation, type safety, and error handling.
04 同仓库更多 Skills
supabase-rls-policies
Write, review, and test Supabase Row Level Security (RLS) policies so that one user cannot read or change another user's rows. Use when creating or changing tables, policies, views, or security definer functions in a Supabase project, when reviewing a migration that touches RLS, or when asked to prove user isolation with pgTAP tests that sign in as two users.
skill-distribution-audit
Audit a skill collection before release or investigate installer omissions, archived/template leakage, duplicate identities, and manifest drift. Reconcile the intended public source set, distribution manifest, and actual discovery output for a specified consumer version. Includes a deterministic set-comparison helper.
logicapps-custom-code-net10-migration
Migrate Azure Logic Apps Standard local-function custom-code projects from .NET 8 to .NET 10. Use for legacy Logic Apps custom-code workspace upgrades, project/package configuration, custom-worker settings, publication, and integrated VS Code debugging. Not for ordinary .NET applications, standalone Azure Functions, Consumption workflows, or cloud deployment.
winmd-api-search
Find and explore Windows desktop APIs. Use when building features that need platform capabilities — camera, file access, notifications, UI controls, AI/ML, sensors, networking, etc. Discovers the right API for a task and retrieves full type details (methods, properties, events, enumeration values).
resemble-detect
Deepfake detection and media safety — detect AI-generated audio, images, video, and text, trace synthesis sources, apply watermarks, verify speaker identity, and analyze media intelligence using Resemble AI
quality-playbook
Run a complete quality engineering audit on any codebase. Derives behavioral requirements from the code, generates spec-traced functional tests, runs a three-pass code review with regression tests, executes a multi-model spec audit (Council of Three), and produces a consolidated bug report with TDD-verified patches. Finds the 35% of real defects that structural code review alone cannot catch. Works with any language. Trigger on 'quality playbook', 'spec audit', 'Council of Three', 'fitness-to-purpose', or 'coverage theater'.