What problem does it solve? When WebMobAI itself breaks — MCP tools missing from Claude, browsers failing to launch, Playwright engines not installed, or the desktop app flagged as damaged — this Skill provides a decision-tree workflow to isolate whether the fault is in the environment, the MCP wiring, or the live browser session. ## Core Features & Use Cases - Doctor-driven diagnosis: Runs webmobai-doctor first and branches on its six/seven checks covering Node version, Playwright engines, Lighthouse, AI API key, and storageState validity. - Layered fault isolation: Distinguishes environment failures (Node < 18, missing Chromium), wiring failures (MCP server not appearing in Claude Desktop or Claude Code), and session-state failures (browser not launched, already running, idle-timeout auto-close). - Use Case: A user says "none of the webmobai tools show up in Claude Code." The Skill walks the .mcp.json config, validates JSON and command paths, confirms dist/index.js is built, and has the user run npx -y webmobai-mcp by hand to read the real stderr error. ## Quick Start Run webmobai-doctor and use the troubleshooting skill to diagnose why my WebMobAI tools are not working.