Webix MCP Server

AI coding assistants generate code from their training data, which means they can miss recent API changes, new widgets, or updated configuration options. The Webix MCP Server fixes this by giving your AI tool direct access to up-to-date Webix documentation using RAG (retrieval-augmented generation).

This guide describes how the Webix MCP Server works, and how to connect it to your AI coding tool.

The MCP server URL is: https://docs.webix.com/mcp

When you ask a Webix question, the MCP server searches its documentation index and returns relevant content to your AI tool as context. The result is code that reflects the actual current API.

The server exposes four primitive types described below. If you're using a supported AI tool (Claude Code, Cursor, Gemini CLI), you won't interact with these primitives directly. The primitives matter if you're building your own MCP host.

• Tools — search and inference. Search runs in two modes: sparse (keyword matching) or dense (semantic similarity). Inference answers questions grounded in verified docs and code snippets.
• Resources — two documents per domain: full documentation in LLM-friendly format, and a condensed version you can inject into the LLM context. The additional Runbook resource defines the agentic workflow for your AI assistant.
• Prompts — predefined use cases for demonstration, debugging, and migration. Runbook is also available here with a prompt interface.
• MCP sampling — a primitive that lets the inference tool use your own trusted LLM at response-generation time instead of the server's default.

The server works with any AI tool that supports the MCP protocol. Select yours below.

Claude Code

Connect via the CLI:

claude mcp add --transport http webix-mcp https://docs.webix.com/mcp

Or add it manually to ~/.claude.json:

{
  "mcpServers": {
    "webix-mcp": {
      "type": "http",
      "url": "https://docs.webix.com/mcp"
    }
  }
}

See the full guidance in the Claude Code documentation.

Cursor

1. Open settings (Cmd+Shift+J on Mac, Ctrl+Shift+J on Windows/Linux)
2. Run View: Open MCP Settings
3. Click Add Custom MCP
4. Paste this configuration:

{
  "mcpServers": {
    "webix-mcp": {
      "url": "https://docs.webix.com/mcp"
    }
  }
}

Check the complete guide in the Cursor documentation.

Antigravity CLI (former Gemini CLI)

Define a server inside mcp_config.json as follows:

• for global servers: ~/.gemini/config/mcp_config.json

• at workspace: .agents/mcp_config.json

After that, add the server configuration as shown below:

{
  "mcpServers": {
    "webix-mcp": {
      "serverUrl": "https://docs.webix.com/mcp"
    }
  }
}

and run the agy command in the terminal.

Learn the details on migrating Gemini CLI to Antigravity CLI.

The more specific your prompt is, the better the MCP server can retrieve the right documentation. Name the widget, describe what you want to achieve, and include any relevant constraints.

Here are examples that tend to work well:

• "How to create a simple layout with two buttons using Webix?"
• "Is there a way to build a nested Webix menu?"
• "How to bind a datatable with form inputs using Webix?"
• "Provide initialization and customization guidance on the Webix sidebar"
• "How to create a Webix bar chart with 2 series?"

The Webix MCP Server runs in the cloud and has no access to your environment. No personal information is stored. Requests may be logged for debugging and service improvement.