Kernel

Remote MCP server that connects AI assistants to Kernel tools, browser automation, and docs via secure HTTP or stdio transports.
  • javascript

21

GitHub Stars

javascript

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

This Kernel MCP Server lets AI assistants securely access Kernel platform tools and browser automation capabilities. It provides a remote, authenticated MCP endpoint you can connect to from compatible clients to deploy Kernel apps, run headless browsers, search docs, and automate web actions without hosting your own server.

How to use

You connect to the Kernel MCP Server from your MCP-compatible client to perform actions across Browser Automation, App Management, and Documentation & Search. Use the HTTP endpoint for remote, hosted access or use the stdio transport for clients that do not support remote MCP.

Key capabilities you can leverage include launching and managing headless browser sessions, running Playwright code against live browsers, deploying Kernel apps in the cloud, retrieving deployment status and logs, and searching Kernel documentation to inject context into your conversations. You can also record video replays of automated browser actions for review.

To get started, choose one of the supported transports, connect your client, and authorize access to your Kernel resources via the authentication flow provided by the server.

How to install

Prerequisites you need on your machine or environment:

  • Node.js runtime (recommended)
  • bun (optional, used for local dev)
  • Git (for cloning if you import from source)

Run the server locally to test and develop against your own environment. Use the following commands to install dependencies and start the server:

cp .env .env.local
bun install
bun run dev

Connect in your client

The server is hosted at a remote MCP URL you can connect to from any MCP-compatible client. You can use either the HTTP transport or the stdio transport depending on what your client supports.

Supported clients and setup flows are described below for popular tools. Follow the steps to add Kernel as an MCP server, then authorize access when prompted.

Claude

Pro, Max, Team & Enterprise (Claude.ai and Claude Desktop) integrate Kernel via a custom connector.

Steps to connect:

  1. In Claude, go to Settings → Connectors → Add custom connector.

  2. Set Integration name to Kernel and Integration URL to https://mcp.onkernel.com/mcp, then Add.

  3. In Settings → Connectors, click Connect next to Kernel to launch OAuth and approve.

  4. In chat, click Search and tools and enable Kernel tools if needed.

Cursor

Cursor can automatically configure Kernel when available and provide a one-click setup experience.

Automatic setup link and manual steps are provided in the client to add Kernel as an MCP server.

Goose

Goose supports Kernel via a custom extension that runs MCP remotely.

Automatic setup and manual extension configuration are described in Goose’s extension management UI.

Visual Studio Code

Add Kernel as an MCP server in VS Code by configuring the HTTP MCP URL.

Steps:

{
  "mcpServers": {
    "kernel": {
      "url": "https://mcp.onkernel.com/mcp",
      "type": "http"
    }
  }
}

Windsurf

Windsurf can load MCP servers from raw config settings.

{
  "mcpServers": {
    "kernel": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.onkernel.com/mcp"]
    }
  }
}

Zed

Zed uses a similar custom server entry for MCP connections.

{
  "context_servers": {
    "kernel": {
      "source": "custom",
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://mcp.onkernel.com/mcp"]
    }
  }
}

Smithery

You can connect directly to the remote MCP URL or use Smithery as a proxy.

Use Smithery URL in any MCP client by copying the connection URL and adding it as a server.

Others

Many MCP-capable tools accept a standard transport using the following command pattern.

{
  "kernel": {
    "command": "npx",
    "args": ["-y", "mcp-remote", "https://mcp.onkernel.com/mcp"]
  }
}

Available tools

create_browser

Launch a new browser session with options such as headless mode, timeouts, and user profile.

get_browser

Retrieve information about an active browser session.

list_browsers

List all currently active browser sessions.

delete_browser

Terminate a browser session and clean up resources.

execute_playwright_code

Run Playwright/TypeScript code in a fresh browser session with automatic video replay and cleanup.

setup_profile

Create or update browser profiles with guided setup.

list_profiles

List all browser profiles available in Kernel.

delete_profile

Delete a browser profile permanently.

list_apps

List deployed Kernel apps in your organization, with optional filtering.

invoke_action

Execute actions within Kernel apps.

get_deployment

Get deployment status and logs for a specific app.

list_deployments

List all deployments with optional filtering.

get_invocation

Retrieve details about a specific action invocation.

search_docs

Search Kernel platform documentation and guides.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational