What problem does it solve? AI browsing agents increasingly attempt to complete tasks like booking, buying, and registering on websites, but most sites lack machine-readable action declarations, causing agents to fail mid-flow. This Skill audits whether agents can actually complete high-value task flows on your site and implements WebMCP declarative and imperative patterns to fix failures. ## Core Features & Use Cases - WebMCP Readiness Audits: Test task flows with real browser agents and produce scorecards measuring discoverability, initiatability, and completion rates per flow. - Declarative & Imperative Implementation: Add data-mcp-action attributes to HTML forms or register dynamic actions via navigator.mcpActions.register(), plus publish a /mcp-actions.json discovery endpoint. - Agent Friction Mapping: Identify step-by-step where agents drop or fail, classifying issues like custom JS date pickers, CAPTCHA gates, and auth walls with specific fixes. - Use Case: A SaaS company finds AI agents cannot complete its signup flow. The audit reveals a custom JavaScript date picker blocks agents at step 2; replacing it with a native input and adding declarative markup raises task completion from 20% to 80%. ## Quick Start Audit my website's booking and contact forms for WebMCP readiness and show me which task flows AI browsing agents cannot complete.
