mcp-ts-template

MCP TypeScript Template 🚀

Jumpstart your Model Context Protocol (MCP) Client & Server development with this TypeScript MCP Repo Template!

This template provides a solid, beginner-friendly foundation for building robust MCP servers and clients, adhering to the MCP 2025-03-26 specification. It includes production-ready utilities, a well-structured codebase, working examples, and clear documentation to get you up and running quickly.

Whether you're creating a new MCP server to extend an AI's capabilities or integrating MCP client features into your application, this template is your starting point.

📋 Table of Contents

• ✨ Key Features
• 🌟 Projects Using This Template
• 🏁 Quick Start
• ⚙️ Configuration
• 🔩 Server Configuration (Environment Variables)
• 🏗️ Project Structure
• 🧩 Extending the MCP Server
• 🌍 More MCP Resources
• 📜 License
• 📊 Detailed Features Table

✨ Key Features

Feature Area	Description	Key Components / Location
🔌 MCP Server	Functional server with example tools (EchoTool, CatFactFetcher for async/Promise API example) and an EchoResource. Supports stdio and Streamable HTTP transports.	src/mcp-server/
💻 MCP Client	Working client aligned with MCP 2025-03-26 spec. Connects via mcp-config.json. Includes detailed comments.	src/mcp-client/
🚀 Production Utilities	Logging, Error Handling, ID Generation, Rate Limiting, Request Context tracking, Input Sanitization.	src/utils/
🔒 Type Safety/Security	Strong type checking via TypeScript & Zod validation. Built-in security utilities (sanitization, auth middleware for HTTP).	Throughout, src/utils/security/, src/mcp-server/transports/auth/
⚙️ Error Handling	Consistent error categorization (BaseErrorCode), detailed logging, centralized handling (ErrorHandler).	src/utils/internal/errorHandler.ts, src/types-global/
📚 Documentation	Comprehensive README.md, structured JSDoc comments, API references	README.md, Codebase, tsdoc.json, docs/api-references/
🤖 Agent Ready	Includes a .clinerules developer cheatsheet tailored for LLM coding agents.	.clinerules
🛠️ Utility Scripts	Scripts for cleaning builds, setting executable permissions, generating directory trees, and fetching OpenAPI specs.	scripts/
🧩 Services	Reusable modules for LLM (OpenRouter) and data storage (DuckDB) integration, with examples.	src/services/, src/storage/duckdbExample.ts

For a more granular breakdown, see the Detailed Features Table below.

🌟 Projects Using This Template

This template is already powering several MCP servers, demonstrating its flexibility and robustness:

Project	Description	Status / Notes
clinicaltrialsgov-mcp-server	An MCP (Model Context Protocol) server providing a robust, developer-friendly interface to the official ClinicalTrials.gov v2 API. Enables LLMs and AI agents to search, retrieve, and analyze clinical study data programmatically.	Actively using this template.
pubmed-mcp-server	MCP server for PubMed, enabling AI agents to search, retrieve, analyze, and visualize biomedical literature via NCBI E-utilities. Features advanced research workflow capabilities.	Actively using this template.
git-mcp-server	Provides an enterprise-ready MCP interface for Git operations. Allows LLM agents to initialize, clone, branch, commit, and manage repositories via STDIO & Streamable HTTP.	Actively using this template.
obsidian-mcp-server	MCP server for Obsidian, enabling AI agents to read, write, search, and manage notes via the Local REST API plugin.	Actively using this template.
atlas-mcp-server	Advanced task and knowledge management system with Neo4j backend, enabling structured data organization and complex querying for AI agents.	Aligned with this template (as of v2.8.8).
filesystem-mcp-server	Offers platform-agnostic file system capabilities for AI agents via MCP. Enables reading, writing, updating, and managing files/directories, featuring advanced search/replace and directory traversal.	Actively using this template.

Note: toolkit-mcp-server was initially built using an older version of this template and is pending updates to the latest structure.

You can also see my GitHub profile for additional MCP servers I've created, many of which are planned to be migrated to or built upon this template in the future.

📦 Installation

This project is a template, but can also directly be run as an MCP server. To get started, follow these steps:

• Clone the repository:

  git clone https://github.com/cyanheads/mcp-ts-template.git
  cd mcp-ts-template
  

• Install dependencies:

  npm install
  

• Build the project:

  npm run build
  # Or use 'npm run rebuild' for a clean install
  

🚀 Usage

Once built, you can run the server. If you intend to publish your project as a package, you can test the npx command by linking it locally.

Using NPX

To run the MCP server directly using npx, you can use the following command:

Now you can run it via `npx`:

```bash
# Run with default stdio transport
npx mcp-ts-template

# Run with HTTP transport
MCP_TRANSPORT_TYPE=http npx mcp-ts-template

From Source (for Development)

You can also use the npm scripts to run the server directly from your cloned repository:

• Via Stdio (Default):

  npm start
  # or 'npm run start:stdio'
  

• Via Streamable HTTP: bash npm run start:http This starts a Streamable HTTP server (default: http://127.0.0.1:3010) which uses Server-Sent Events for the server-to-client streaming component. The port, host, and allowed origins are configurable via environment variables (see Configuration).

⚙️ Configuration

🔩 Server Configuration (Environment Variables)

Configure the MCP server's behavior using these environment variables:

Variable	Description	Default
MCP_TRANSPORT_TYPE	Server transport: stdio or http.	stdio
MCP_HTTP_PORT	Port for the HTTP server (if MCP_TRANSPORT_TYPE=http).	3010
MCP_HTTP_HOST	Host address for the HTTP server (if MCP_TRANSPORT_TYPE=http).	127.0.0.1
MCP_ALLOWED_ORIGINS	Comma-separated allowed origins for CORS (if MCP_TRANSPORT_TYPE=http).	(none)
MCP_SERVER_NAME	Optional server name (used in MCP initialization).	(from package.json)
MCP_SERVER_VERSION	Optional server version (used in MCP initialization).	(from package.json)
MCP_LOG_LEVEL	Server logging level (debug, info, warning, error, etc.).	debug
LOGS_DIR	Directory for log files.	logs/ (in project root)
NODE_ENV	Runtime environment (development, production).	development
MCP_AUTH_SECRET_KEY	Required for HTTP transport. Secret key (min 32 chars) for signing/verifying auth tokens (JWT).	(none - MUST be set in production)
MCP_AUTH_MODE	Authentication mode: jwt (default) or oauth.	jwt
OAUTH_ISSUER_URL	Required for oauth mode. The issuer URL of your authorization server.	(none)
OAUTH_AUDIENCE	Required for oauth mode. The audience identifier for this MCP server.	(none)
OAUTH_JWKS_URI	Optional for oauth mode. The JWKS endpoint URL. If omitted, it's discovered from the issuer.	(none)
OPENROUTER_API_KEY	API key for OpenRouter.ai service. Optional, but service will be unconfigured without it.	(none)
LLM_DEFAULT_MODEL	Default model to use for LLM calls via OpenRouter.	google/gemini-2.5-flash-preview-05-20
LLM_DEFAULT_TEMPERATURE	Default temperature for LLM calls (0-2). Optional.	(none)

Note on HTTP Port Retries: If the MCP_HTTP_PORT is busy, the server automatically tries the next port (up to 15 times).

Security Note for HTTP Transport: When using MCP_TRANSPORT_TYPE=http, authentication is mandatory as per the MCP specification. This template supports two modes via MCP_AUTH_MODE:

• jwt (default): A simple, self-contained JWT mode ideal for development. It requires the MCP_AUTH_SECRET_KEY to be set for signing and verifying tokens.
• oauth: A production-ready OAuth 2.1 mode where the server validates Bearer tokens from an external Authorization Server. This requires OAUTH_ISSUER_URL and OAUTH_AUDIENCE to be configured.

You MUST configure one of these modes for the security mechanism to function correctly when using the HTTP transport.

🔌 Client Configuration

For detailed information on configuring the built-in MCP client, including how to set up connections to external MCP servers using mcp-config.json, please see the Client Configuration Guide.

Example: Adding to an MCP Client

To add this server to an MCP client application (like Claude Desktop), you would update the client's configuration file. This example assumes you have published your package or have used npm link to make it available locally.

{
  "mcpServers": {
    "mcp-ts-template": {
      "command": "npx",
      "args": ["mcp-ts-template"],
      "env": {
        "MCP_LOG_LEVEL": "debug",
        "MCP_TRANSPORT_TYPE": "http",
        "MCP_HTTP_PORT": "3010"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}

🏗️ Project Structure

This project follows a standard TypeScript project layout. Here's an overview of the key directories and files:

• .clinerules: Developer cheatsheet and guidelines for LLM coding agents working with this repository.
• docs/: Contains project documentation, including API references and the auto-generated tree.md file.
• scripts/: Utility scripts for development tasks like cleaning builds, generating directory trees, and fetching OpenAPI specs.
• src/: The heart of the application, containing all TypeScript source code.
  • src/config/: Handles loading and validation of environment variables and application configuration.
  • src/mcp-client/: Implements the MCP client logic for connecting to and interacting with external MCP servers. This includes client configuration, core connection management, and transport handlers.
  • src/mcp-server/: Contains the MCP server implementation provided by this template, including example tools, resources, and transport handlers (Stdio, HTTP).
  • src/services/: Provides reusable modules for integrating with external services, such as DuckDB for local data storage and OpenRouter for LLM access.
  • src/types-global/: Defines shared TypeScript interfaces and type definitions used across the project, particularly for error handling and MCP-specific types.
  • src/utils/: A collection of core utilities.
    • src/utils/internal/: Core internal utilities like the logger, error handler, and request context management.
    • src/utils/metrics/: Utilities related to metrics, such as token counting.
    • src/utils/network/: Network-related utilities, like fetch with timeout.
    • src/utils/parsing/: Utilities for parsing data, such as dates and JSON.
    • src/utils/security/: Security-focused utilities including ID generation, rate limiting, and input sanitization.
  • src/index.ts: The main entry point for the application, responsible for initializing and starting the MCP server. The MCP client is meant to be built upon, so it does not have a dedicated entry point in this template.
• package.json: Defines project metadata, dependencies, and npm scripts.
• README.md: This file, providing an overview of the project.
• tsconfig.json: TypeScript compiler options for the project.
• LICENSE: Apache 2.0 License file.

Explore the full structure yourself:

See the current file tree in docs/tree.md or generate it dynamically:

npm run tree

(This uses scripts/tree.ts to generate a current file tree, respecting .gitignore.)

🧩 Extending the MCP Server

For detailed guidance on how to add your own custom Tools and Resources to this MCP server template, including workflow examples and best practices, please see the Server Extension Guide.

🌍 Explore More MCP Resources

Looking for more examples, guides, and pre-built MCP servers? Check out the companion repository:

➡️ cyanheads/model-context-protocol-resources

This collection includes servers for Filesystem, Obsidian, Git, GitHub, Perplexity, Atlas, Ntfy, and more, along with in-depth guides based on my real-world MCP development.

📜 License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.

📊 Detailed Features Table

Category	Feature	Description	Location(s)
Core Components	MCP Server	Core server logic, tool/resource registration, transport handling. Includes Echo Tool & Resource examples.	src/mcp-server/
	MCP Client	Logic for connecting to external MCP servers (updated to MCP 2025-03-26 spec). Refactored for modularity.	src/mcp-client/ (see subdirs: core/, client-config/, transports/)
	Configuration	Environment-aware settings with Zod validation.	src/config/, src/mcp-client/client-config/configLoader.ts
	Streamable HTTP Transport	Hono-based server implementing the MCP Streamable HTTP transport with session management, CORS, and port retries.	src/mcp-server/transports/httpTransport.ts
	Stdio Transport	Handles MCP communication over standard input/output.	src/mcp-server/transports/stdioTransport.ts
Utilities (Core)	Logger	Structured, context-aware logging (files with rotation & MCP notifications).	src/utils/internal/logger.ts
	ErrorHandler	Centralized error processing, classification, and logging.	src/utils/internal/errorHandler.ts
	RequestContext	Request/operation tracking and correlation.	src/utils/internal/requestContext.ts
Utilities (Metrics)	TokenCounter	Estimates token counts using tiktoken.	src/utils/metrics/tokenCounter.ts
Utilities (Parsing)	DateParser	Parses natural language date strings using chrono-node.	src/utils/parsing/dateParser.ts
	JsonParser	Parses potentially partial JSON, handles <think> blocks.	src/utils/parsing/jsonParser.ts
Utilities (Security)	IdGenerator	Generates unique IDs (prefixed or UUIDs).	src/utils/security/idGenerator.ts
	RateLimiter	Request throttling based on keys.	src/utils/security/rateLimiter.ts
	Sanitization	Input validation/cleaning (HTML, paths, URLs, numbers, JSON) & log redaction (validator, sanitize-html).	src/utils/security/sanitization.ts
Services	DuckDB Integration	Reusable module for in-process analytical data management using DuckDB. A storage layer that runs on the same level as the application. Includes connection management, query execution, and example usage. Integrated with our utils (logger, etc.)	src/services/duck-db/, src/storage/duckdbExample.ts
	OpenRouter LLM Integration	Reusable module for interacting with various LLMs via the OpenRouter API (OpenAI SDK compatible). Integrated with our utils (logger, etc.)	src/services/llm-providers/openRouterProvider.ts
Type Safety	Global Types	Shared TypeScript definitions for consistent interfaces (Errors, MCP types).	src/types-global/
	Zod Schemas	Used for robust validation of configuration files and tool/resource inputs.	Throughout (config, mcp-client, tools, etc.)
Error Handling	Pattern-Based Classification	Automatically categorize errors based on message patterns.	src/utils/internal/errorHandler.ts
	Consistent Formatting	Standardized error responses with additional context.	src/utils/internal/errorHandler.ts
	Safe Try/Catch Patterns	Centralized error processing helpers (ErrorHandler.tryCatch).	src/utils/internal/errorHandler.ts
	Client/Transport Error Handling	Specific handlers for MCP client and transport error handling.	src/mcp-client/core/, src/mcp-client/transports/
Security	Input Validation	Using validator and zod for various data type checks.	src/utils/security/sanitization.ts, etc.
	Input Sanitization	Using sanitize-html to prevent injection attacks.	src/utils/security/sanitization.ts
	Sensitive Data Redaction	Automatic redaction in logs.	src/utils/security/sanitization.ts
	Configuration Validation	Throws a descriptive error if the primary client config (mcp-config.json) is missing, preventing fallback to a potentially insecure example file.	src/mcp-client/client-config/configLoader.ts
Scripts	Clean Script	Removes dist and logs directories (or custom targets).	scripts/clean.ts
	Make Executable Script	Sets executable permissions (chmod +x) on specified files (Unix-like only).	scripts/make-executable.ts
	Tree Script	Generates a directory structure tree, respecting .gitignore.	scripts/tree.ts
	Fetch OpenAPI Spec Script	Fetches an OpenAPI spec (YAML/JSON) from a URL with fallbacks, saves locally.	scripts/fetch-openapi-spec.ts

Built with ❤️ and the Model Context Protocol