agent-ready-webmcp

To expose website actions to AI agents, you use the WebMCP API to register tools with JSON Schema inputs and executable callbacks during page rendering. This enables compatible agents to perform site actions like search and navigation instead of only reading content.

WebMCP tool registration uses navigator.modelContext.registerTool to define website capabilities with names, descriptions, JSON Schemas, and execute callbacks. This mechanism allows browser-based AI agents to discover and use site actions during page rendering.

You manage the lifecycle of registered WebMCP tools by using an AbortController to unregister them when they are no longer needed. This ensures that browser-based AI agents cannot access outdated or irrelevant website actions during page rendering.

You validate if your website is ready for AI agent discovery by scanning the deployed site with the Is It Agent Ready API. This verifies that your WebMCP tool registration passes discovery validation for browser-based AI agents.

Yes, you need JSON Schema to define the input parameters for WebMCP tools exposed to AI agents. The navigator.modelContext.registerTool method requires JSON Schema inputs alongside execute callbacks to enable browser automation.

Yes, you can expose search, navigation, and data retrieval capabilities to AI agents through the WebMCP API. By applying tool registration to web applications, browser-based agents can discover and execute these site actions during page rendering.