Discover and implement secured MCP servers with ease
MCPlug eliminates security vulnerabilities, reduces complex coding, and helps you quickly find the right MCP servers for your LLM applications.
const tools = await mcplug({ token, id });
Ship your AI apps faster with integrations you can trust
Connecting your AI apps to MCP servers means navigating security risks, complex integration, and finding quality servers. These challenges slow your launch and expose you to risk. MCPlug solves all three problems.
Comprehensive Security Protection
Automatic scanning protects your apps from malicious tool descriptions, rug pulls, and credential theft.
Single Implementation
Replace hundreds of lines of custom integration code with a single function call – add new MCP tools in seconds
Unified Control Panel
Manage all your MCP connections through one interface – with logs, usage tracking, and version control.
Curated Marketplace
Find specialized MCP tools from our catalog of pre-vetted, ready-to-use servers.
From complex manual integration to one simple line
90%
Less code
85%
Faster implementation
40%
Faster time-to-market
Without MCPlug
Manual implementation of the tools
const weatherTool = tool({
description: 'Get the weather in a location',
parameters: z.object({
location: z.string().describe('The location to get the weather for'),
}),
execute: async ({ location }) => ({
// ... your custom implementation
}),
});
const exaSearchTool = tool({
description: 'Search the web for specific information',
parameters: z.object({
query: z.string().describe('The query to search for'),
}),
execute: async ({ query }) => ({
// ... your custom implementation
}),
});
// ... any other tools definitions your app need
const { text } = await generateText({
model: openai("gpt-4o"),
system: "You are a helpful assistant.",
messages: [{ role: "user", content: "What's the weather in Tokyo?" }],
tools:[weatherTool, exaSearchTool, ...otherToolsYouHaveToMaintain]
});
With MCPlug
One line of codeconst { text } = await generateText({
model: openai("gpt-4o"),
system: "You are a helpful assistant.",
messages: [{ role: "user", content: "What's the weather in Tokyo?" }],
// Secure, simple implementation with MCPlug
tools: await mcplug({ token, id })
});What you get:
Security
Easier Implementation
Unlimited Tools
For AI clients
Use MCPlug in your favorite AI client
Be it Claude, Cursor, Windsurf, you name it, MCPlug is compatible with all of them.
Set up your json file once and manage all your MCP servers and credentials on our dashboard. Replace hundreds of lines of json configuration with a single MCP server.
{
"mcpServers": {
"mcplug": {
"command": "npx",
"args": ["-y", "@mcplug/client@latest", "<YOUR_PLUG_ID>"],
"env": {
"MCPLUG_TOKEN": "<YOUR_TOKEN>"
}
}
}
}// Initialize your server with a single command npx mcplug init my-awesome-server
// Simplified developer experience
export default {
tools: {
"get-weather": tool("Get the weather in a given city")
.input(
z.object({
city: z.string()
})
)
.handle(async ({ input }) => {
return {
city: input.city,
temp: 20,
unit: "C",
condition: "sunny"
};
})
}
}; // When you're ready, publish your server npx mcplug publish
For MCP Builders
Publish Your Own Server
Want to publish your own MCP server to the MCPlug repository?
- Deploy in seconds, no infra needed
- Simple and easy to use SDK
- Hosted on our infra (Cloudflare)
- Monetize per request (Coming soon)
Security
Protect your applications from MCP risks
Recent security research from Invariant Labs has uncovered vulnerabilities in the MCP ecosystem. MCPlug offers protection through simple implementation, letting you focus on building features, not fixing security holes.
Tool Poisoning & Shadowing Protection
Sleep better knowing your AI models cannot be tricked into stealing sensitive data through hidden instructions in tool descriptions.
Without protection: Attackers can inject malicious instructions to steal SSH keys, config files, and user data.
Rug Pull Prevention
Never worry about trusted tools turning malicious after deployment with automatic version pinning that prevents server-side changes.
Without protection: MCP servers can change tool descriptions after approval, compromising previously trusted tools.
Secure Credential Handling
Keep your API keys and sensitive data completely isolated from potential exploits with sandboxed environments and credential protection.
Without protection: Credentials can be exposed to LLMs and potentially exfiltrated through manipulated responses.
All tools on our platform are publicly accessible and can be reviewed by the community. We encourage users to report any suspicious activity to maintain the highest security standards.
Pricing
Start for free. Grow without limits.
Begin with our free tier and scale up as your needs grow. Simple, transparent pricing for developers of all sizes.
Development
$0Start integrating secure MCP servers today
No credit card required to start
Perfect for personnal useage or for validating concepts, simplified integration, and access to quality tools.
- Access all MCP servers
- Publish your MCP servers
- 2 plugs
- Rate limit: 10 requests / minute
- Community support
Production
$9.99 /month
1,000,000 requests included
+10$ / 1M additional requests
Pro features in detail, everything in Free plus:
No rate limits
Build without restrictions, perfect for production apps
Private deployments
Host internal tools securely
Advanced analytics
In-depth insights on usage and performance
Priority support
Get help when you need it most
Need custom solutions for enterprise? Contact us for enterprise plans.
Documentation
01 • For Providers
@mcplug/server
Deploy your own MCP servers and join publish it on the MCPlug repository.
Documentation02 • For Builders
@mcplug/client
Connect to any remote MCP server with just a few lines of code.
DocumentationBuilding the open infrastructure for MCP servers
The JavaScript world has npm - a trusted repository where developers can instantly add powerful capabilities to their applications. When building AI applications with LLMs, I found myself repeatedly asking: "Why doesn't this exist for AI tools and MCP servers?"
Every new tool I wanted to add to my LLM app requires hundreds of lines of custom integration code, careful security reviews, and manual credential management.
MCPlug was born to solve these implementation headaches and create a secure ecosystem developers can trust. By providing standardized implementations through MCP, automated security screening, and a verified repository of quality tools, we're building the infrastructure that lets you focus on creating amazing AI experiences, not wrestling with integrations.
Arnaud, Founder
FAQ
Everything you need to know about MCPlug and how it can enhance your AI applications.
Still have questions? Contact our team