Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@MCP Server Boilerplate show me how to add a new tool to my custom MCP server"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
A starter template for building MCP (Model Context Protocol) servers. This boilerplate provides a clean foundation for creating your own MCP server that can integrate with Claude Desktop, Cursor, Claude Code, Gemini, and other MCP-compatible AI assistants.
This boilerplate helps you quickly start building:
Custom tools for AI assistants
Resource providers for dynamic content
Prompt templates for common operations
Integration points for external APIs and services
Related MCP server: MCP Server Boilerplate
Two example tools: "hello-world" and "get-mcp-docs"
TypeScript support with ES2022 target and ES modules
Multi-client installation scripts (Claude Desktop, Cursor, Claude Code, Gemini, etc.)
Automatic npm publishing workflow
Environment variable support via .env.local
Clean project structure with Zod validation
This MCP server template provides:
A basic server setup using the MCP SDK
Example tool implementation
Build and installation scripts
TypeScript configuration for development
The included example demonstrates how to create a simple tool that takes a name parameter and returns a greeting.
You can use this MCP server directly without cloning:
# Run the server directly with npx
npx @r-mcp/boilerplate# Clone the boilerplate
git clone <your-repo-url>
cd mcp-server-boilerplate
# Install dependencies
pnpm install
# Build the project
pnpm run build
# Start the server
pnpm startThis boilerplate includes convenient installation scripts for different MCP clients:
# Install to all MCP clients (Claude Desktop, Cursor, Claude Code, Gemini, MCP)
pnpm run install-server
# Install to specific clients
pnpm run install-desktop # Claude Desktop
pnpm run install-cursor # Cursor IDE
pnpm run install-code # Claude Code CLI
pnpm run install-code-library # Claude Code Library (~/.claude/mcp-library/)
pnpm run install-mcp # Local .mcp.json for development
# You can also combine multiple targets:
node scripts/update-config.js cursor code desktopThese scripts will:
Build the project automatically (TypeScript compilation + chmod permissions)
Configure clients to use npx @r-mcp/<directory-name>@latest (auto-updating)
Only the local .mcp.json uses the development version (node dist/index.js)
Include environment variables from .env.local if present
To publish your customized MCP server:
# Build, commit, and publish to npm in one command
pnpm run releaseThis script (scripts/build-and-publish.js) will:
Commit any pending changes first
Update package name to @r-mcp/<directory-name>
Update bin name to match directory
Increment patch version automatically
Build the TypeScript project
Commit version bump to git
Push to remote repository
Publish to npm with public access
The installation scripts automatically configure your MCP clients. For reference, here's what gets added:
{
"mcpServers": {
"boilerplate": {
"command": "npx",
"args": ["-y", "@r-mcp/boilerplate@latest"],
"env": {
// Environment variables from .env.local are included here
}
}
}
}.mcp.json):{
"mcpServers": {
"boilerplate": {
"command": "node",
"args": ["/absolute/path/to/dist/index.js"],
"env": {
// Environment variables from .env.local are included here
}
}
}
}After running installation scripts, restart your MCP client to connect to the server.
Tools are functions that the AI assistant can call. Here's the basic structure:
server.tool(
"tool-name",
"Description of what the tool does",
{
// Zod schema for parameters
param1: z.string().describe("Description of parameter"),
param2: z.number().optional().describe("Optional parameter"),
},
async ({ param1, param2 }) => {
// Your tool logic here
return {
content: [
{
type: "text",
text: "Your response",
},
],
};
}
);Resources provide dynamic content that the AI can access:
server.resource(
"resource://example/{id}",
"Description of the resource",
async (uri) => {
// Extract parameters from URI
const id = uri.path.split("/").pop();
return {
contents: [
{
uri,
mimeType: "text/plain",
text: `Content for ${id}`,
},
],
};
}
);Prompts are reusable templates:
server.prompt(
"prompt-name",
"Description of the prompt",
{
// Parameters for the prompt
topic: z.string().describe("The topic to discuss"),
},
async ({ topic }) => {
return {
description: `A prompt about ${topic}`,
messages: [
{
role: "user",
content: {
type: "text",
text: `Please help me with ${topic}`,
},
},
],
};
}
);├── src/
│ └── index.ts # Main MCP server implementation
├── scripts/
│ ├── update-config.js # Multi-client configuration installer
│ └── build-and-publish.js # Automated npm publishing workflow
├── dist/ # Compiled JavaScript (generated)
├── package.json # Project configuration
├── tsconfig.json # TypeScript configuration
├── CLAUDE.md # Claude Code instructions
├── .env.local # Environment variables (optional)
└── README.md # This fileMake changes to src/index.ts
Run pnpm run build to compile TypeScript
Test your server with pnpm start
Use pnpm run install-mcp for local testing
Restart your MCP client to load changes
Test your changes locally
Run pnpm run release to publish to npm
Clients using npx @r-mcp/<your-package>@latest auto-update
No client reconfiguration needed
Create a .env.local file for environment-specific configuration:
# .env.local
API_KEY=your-api-key
DATABASE_URL=your-database-urlThese variables are automatically included in MCP server configurations during installation.
Fork or clone this boilerplate
Customize the server name and tools in src/index.ts
Add your own tools, resources, and prompts
Configure environment variables in .env.local
Run pnpm run release to publish your server
Install to clients with pnpm run install-server
MIT
Make an MCP server
| Name | Required | Description | Default |
|---|---|---|---|
| name | Yes | The name of the MCP server |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden of behavioral disclosure but offers minimal information. 'Make an MCP server' suggests a creation/write operation but doesn't specify what gets created (files, configuration, documentation), whether it requires specific permissions, what the output format is, or any side effects. The description doesn't address error conditions, rate limits, or authentication requirements.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
While the description is technically concise (three words), it's under-specified rather than efficiently informative. The single phrase 'Make an MCP server' doesn't provide enough context to be genuinely helpful. A truly concise description would front-load essential information about the tool's purpose and output.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a tool with no annotations, no output schema, and a description that provides minimal context, this is severely incomplete. The agent cannot determine what the tool actually does, what it returns, when to use it, or how it behaves. The description fails to compensate for the lack of structured metadata, leaving critical gaps in understanding.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The schema has 100% description coverage with a single parameter 'name' documented as 'The name of the MCP server'. The description doesn't add any additional semantic context about this parameter beyond what the schema already provides. Since schema coverage is high, the baseline score of 3 is appropriate - the description neither enhances nor detracts from parameter understanding.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description 'Make an MCP server' is a tautology that essentially restates the tool name 'get-mcp-docs' in different words. It doesn't specify what action is actually performed (e.g., retrieve documentation, create server files, generate configuration) or what resource is being manipulated. The description fails to distinguish this tool from its sibling 'hello-world' or explain what 'making' an MCP server entails.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description provides no guidance on when to use this tool versus alternatives. There's no mention of prerequisites, appropriate contexts, or comparison with the sibling 'hello-world' tool. The agent receives no information about whether this is for development setup, documentation retrieval, or configuration generation.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Say hello to the user
| Name | Required | Description | Default |
|---|---|---|---|
| name | Yes | The name of the user |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden of behavioral disclosure. 'Say hello to the user' implies a read-only output operation, but doesn't specify whether this creates any side effects, requires authentication, has rate limits, or what the actual output format is. It's minimal behavioral information for a tool that presumably just returns a greeting.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is extremely concise at just four words, front-loaded with the core action. There's zero wasted language or redundancy. For a simple greeting tool, this brevity is appropriate and efficient.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple tool with one parameter and no output schema, the description is minimally complete. It tells what the tool does at a high level but lacks details about the return value format or any behavioral constraints. Without annotations and with no output schema, the agent would need to infer the response structure from the tool name and description alone.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The schema has 100% description coverage, with the single parameter 'name' clearly documented as 'The name of the user'. The description doesn't add any additional parameter information beyond what the schema provides, but with only one well-documented parameter and no complex semantics needed, this is adequate. The baseline would be 3 for high schema coverage, but the simplicity of the single parameter justifies a 4.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description 'Say hello to the user' states a clear action (say hello) and target (user), but it's vague about what this actually does - is it a greeting message, a notification, or something else? It doesn't distinguish from the sibling tool 'get-mcp-docs' which is completely different in function. The purpose is understandable but lacks specificity about the output format or mechanism.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No guidance is provided about when to use this tool versus alternatives. The description doesn't mention any context for usage, prerequisites, or exclusions. With only one sibling tool that serves a completely different purpose (document retrieval), there's no explicit comparison or guidance about choosing between them.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v1.0.0get-mcp-docshello-worldScored across 2 tools
The two tools have completely distinct purposes: 'get-mcp-docs' is for creating MCP servers, while 'hello-world' is for greeting users. There is no overlap or ambiguity between these functions, making tool selection straightforward.
The naming is inconsistent: 'get-mcp-docs' uses kebab-case with a verb-object structure, while 'hello-world' uses kebab-case but is a noun-phrase without a clear verb. This mixed convention lacks a predictable pattern, though both use kebab-case.
With only 2 tools, the server feels under-scoped for a 'boilerplate' purpose, which typically implies a foundational set of utilities. This minimal count may not provide sufficient coverage for common MCP development tasks, making it too thin for its apparent scope.
Given the server name 'MCP Server Boilerplate', there are significant gaps in coverage for typical boilerplate functions, such as setup, configuration, testing, or deployment tools. The two tools provided (documentation and greeting) do not form a complete surface for MCP server development, leading to potential agent failures in broader tasks.
A comprehensive Model Context Protocol (MCP) server that enables AI assistants to interact with yo…
Nifty's MCP server — exposes tasks, projects, messages, and files as tools for AI agents.
MCP server for building and testing AI agents with multi-model experimentation and insights.
A simple Typescript MCP server built using the official MCP Typescript SDK and smithery/cli. This…
