xk6-mcp
About
xk6-mcp is a k6 extension for testing Model Context Protocol (MCP) servers. It lets developers load-test MCP servers using k6, providing client support for stdio, SSE, and Streamable HTTP transports, built-in pagination, and automatic RED-style metrics.
Details
- Author
- dgzlopes
- GitHub stars
- 21
- Downloads
- 212
- Categories
- Other
Jump to
- Supports stdio, SSE, and Streamable HTTP transports
- Automatic pagination via *All methods (e.g., listAllTools())
- Manual pagination for fine-grained control
- Built-in RED metrics (request duration, count, errors)
- Tagged metrics per MCP method (e.g., GetPrompt, ListTools)
- Enables connection, tool, resource, and prompt operations
Install xk6, then build a k6 binary with the extension: xk6 build --with github.com/dgzlopes/xk6-mcp. Import the module in your script with import mcp from 'k6/x/mcp'; and run the built binary: ./k6 run script.js.
xk6-mcp
A k6 extension for testing Model Context Protocol (MCP) servers.
> [!CAUTION]
> This extension is experimental and not officially supported by Grafana Labs.
Installation
1. First, ensure you have xk6 installed:
go install go.k6.io/xk6/cmd/xk6@latest
2. Build a k6 binary with the xk6-mcp extension:
xk6 build --with github.com/dgzlopes/xk6-mcp
3. Import the mcp module in your script, at the top of your test script:
import mcp from 'k6/x/mcp';
4. The built binary will be in your current directory. You can move it to your PATH or use it directly:
./k6 run script.js
Example
> ⚠️ This example depends on mcp-example-server
> You can download the latest version from the releases page.
import mcp from 'k6/x/mcp';
export default function () {
// Initialize MCP Client with stdio transport
const client = new mcp.StdioClient({
path: './mcp-example-server',
});
// Check connection to MCP server
console.log('MCP server running:', client.ping());
// List all available tools
console.log('Tools available:');
const tools = client.listAllTools().tools;
tools.forEach(tool => console.log( - ${tool.name}));
// List all available resources
console.log('Resources available:');
const resources = client.listAllResources().resources;
resources.forEach(resource => console.log( - ${resource.uri}));
// List all available prompts
console.log('Prompts available:');
const prompts = client.listAllPrompts().prompts;
prompts.forEach(prompt => console.log( - ${prompt.name}));
// Call a sample tool
const toolResult = client.callTool({ name: 'greet', arguments: { name: 'Grafana k6' } });
console.log(Greet tool response: "${toolResult.content[0].text}");
// Read a sample resource
const resourceContent = client.readResource({ uri: 'embedded:info' });
console.log(Resource content: ${resourceContent.contents[0].text});
// Get a sample prompt
const prompt = client.getPrompt({ name: 'greet' });
console.log(Prompt: ${prompt.messages[0].content.text});
}
F.A.Q.
What about non-stdio transports?
Both SSE and Streamable HTTP are supported too!
You can use them like this:
// SSE
const client = new mcp.SSEClient({
base_url: 'http://localhost:3002',
});
// Streamable HTTP
const client = new mcp.StreamableHTTPClient({
base_url: 'http://localhost:3001',
});
What about pagination?
The extension offers two ways to list resources, tools, and prompts:
// With All: Handles pagination automatically
const allTools = client.listAllTools();
const allResources = client.listAllResources();
const allPrompts = client.listAllPrompts();
// Without All: Requires manual pagination
const first = client.listTools()
const second = client.listTools({ cursor: first.next_cursor });
What about metrics?
The extension automatically tracks RED-style metrics for every MCP operation:
- mcp_request_duration (trend): Duration of each MCP request (in milliseconds).
- mcp_request_count (counter): Number of MCP requests made.
- mcp_request_errors (counter): Number of failed MCP requests.
Each metric is tagged wit:
- method: The MCP method called (e.g., GetPrompt, ListTools).
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.



