Your HTML forms already describe structured inputs - what if an AI agent could fill them out, submit them, and get structured results back, all without a single new API endpoint? Declarative WebMCP turns annotated form elements into browsable tool surfaces by synthesizing schemas from labels, required fields, and select options. This hands-on session walks through annotating real forms for agent discovery, intercepting agent-invoked submissions, returning validation errors the agent can act on, and styling the agent interaction with preview pseudo-classes. In a live demo, we'll see a standard search-and-checkout flow operating as a fully agent-driven experience - and show how the WebMCP AI agent skill guides developers through every compatibility edge case.

Learning Outcomes:
- Understand what WebMCP is, how it differs from server-side MCP, and why the browser is a natural tool-hosting surface for AI agents.
- Annotate existing HTML forms with declarative WebMCP attributes to expose them as structured agent tools with zero JavaScript.
- Handle agent-invoked form submissions by detecting the agent path, returning structured results, and providing corrective validation errors.
- Distinguish between imperative API features and declarative behaviors to write forward-compatible code.
- Keep the human user path and the agent path synchronized on the same visible form state, avoiding hidden agent-only interaction surfaces.
- Use the WebMCP AI agent skill's declarative API reference and troubleshooting guide to diagnose form annotation, schema synthesis, and submit-handling issues in real time.