mcp-server-starter
Minimal typed scaffold for a local Model Context Protocol stdio server, providing echo and server_status tools.
README
TypeScript MCP Server Starter
Minimal typed scaffold for a local Model Context Protocol stdio server. It ships two small read-only tools and enough tests, CI, and smoke verification to start replacing example behavior with real domain logic.
Included:
- Strict TypeScript build on Node.js 20 or newer.
- MCP SDK server with structured input and output schemas.
- Two read-only example tools:
echoandserver_status. - Node built-in unit tests and an in-memory MCP integration test.
- Local launcher for actual
mcp-smokestartup and tool discovery. - GitHub Actions checks for typecheck, tests, build, and smoke verification.
Run With npx
After npm publication, run the pinned starter server without cloning:
npx -y ryanngit-mcp-server-starter@0.1.0
The process uses stdio and waits silently for MCP JSON-RPC input. Logs belong on stderr so stdout remains protocol-only.
For source development after GitHub publication:
git clone https://github.com/ryanngit/mcp-server-starter.git
cd mcp-server-starter
npm ci
npm run verify
npm run build
node .\dist\src\index.js
Claude Desktop
Add this entry to Claude Desktop configuration. It uses the future pinned npm release and requires no local source path.
{
"mcpServers": {
"mcp-server-starter": {
"command": "npx",
"args": ["-y", "ryanngit-mcp-server-starter@0.1.0"]
}
}
}
Configuration locations:
- Windows:
%APPDATA%\Claude\claude_desktop_config.json - macOS:
~/Library/Application Support/Claude/claude_desktop_config.json
Save the file and restart Claude Desktop. On Windows, if Claude cannot resolve
npx, use "command": "cmd" and prepend "/c", "npx" to args.
Example Tools
Echo validated input:
{
"name": "echo",
"arguments": { "message": "hello MCP" },
"result": { "message": "hello MCP", "characterCount": 9 }
}
Read server identity:
{
"name": "server_status",
"arguments": {},
"result": {
"status": "ok",
"server": "mcp-server-starter",
"version": "0.1.0"
}
}
Project Layout
src/index.ts stdio entrypoint
src/config.ts shared server name and version
src/server.ts MCP registration
src/tools.ts schemas and typed handlers
scripts/run-mcp-smoke.ts local smoke launcher
test/ unit and MCP integration tests
.github/workflows/ci.yml continuous integration
Development
npm run typecheck
npm test
npm run verify
npm audit --audit-level=high
npm pack --dry-run --json
Package metadata exposes dist/src/server.js and the executable
mcp-server-starter bin. Package tests guard repository and license metadata
against unresolved publication markers.
Run mcp-smoke
Python 3.11 or newer is needed only for this optional check.
git clone --branch v0.1.0 --depth 1 https://github.com/ryanngit/mcp-smoke.git .mcp-smoke
$env:MCP_SMOKE_PATH = (Resolve-Path .\.mcp-smoke\mcp_smoke.py)
npm run smoke
macOS or Linux:
git clone --branch v0.1.0 --depth 1 https://github.com/ryanngit/mcp-smoke.git .mcp-smoke
MCP_SMOKE_PATH="$PWD/.mcp-smoke/mcp_smoke.py" npm run smoke
The launcher starts the built server, initializes MCP, calls tools/list, and
requires at least one tool. It does not invoke tools or certify domain behavior.
Server and report identity derive from SERVER_NAME and SERVER_VERSION in
src/config.ts.
Optional variables:
| Variable | Purpose |
|---|---|
MCP_SMOKE_PATH |
Required path to mcp_smoke.py. |
MCP_SMOKE_OUT |
Report directory; defaults to mcp-smoke-out. |
PYTHON |
Python executable; defaults to python on Windows and python3 elsewhere. |
Customize
- Change package name, description, repository URLs, and bin name in
package.json. - Change server name and version once in
src/config.ts; MCP metadata, status output, error prefix, and smoke identity derive from that config. - Replace example schemas and handlers in
src/tools.ts. - Register tools in
src/server.ts; keep stdout protocol-only. - Set license year and holder for the resulting project.
- Run
npm run verify,npm run smoke, andnpm pack --dry-run --json.
Input schemas are trust boundaries. Keep limits and validation for external
values. Return isError: true for expected tool failures instead of printing
errors to stdout.
Limitations
- This starter provides no authentication, persistence, prompts, resources, HTTP transport, deployment configuration, or production observability.
- Example tools are intentionally local and stateless.
mcp-smokeverifies startup and tool discovery only.- Renaming requires synchronized package metadata and
src/config.tschanges.
Support
Search or open a reproducible report in GitHub Issues. Include starter version, Node.js version, platform, sanitized config, and error text. Do not include credentials or private conversation content.
License
MIT. Copyright (c) 2026 ryanngit.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
E2B
Using MCP to run code via e2b.