What problem does it solve? Verifying a WASM-canvas dashboard is hard because there is no DOM to query, and renderer regressions (WebGPU failures, frame pacing stalls, broken interactions) slip through ordinary unit tests. This Skill provides a repeatable end-to-end QA workflow for the Station dashboard using a zero-dependency Node + Chrome DevTools Protocol harness. ## Core Features & Use Cases - Renderer health probes: Run named probes (status, rendered, fps, webgpu, smooth, debug-json) against a launched dashboard on headless CI boxes or headed GPU hosts, with configurable fps and frame-gap thresholds. - Agentic canvas interaction: Drive the canvas through station.debug_json() and station.activate(name), or use pointer-path interaction probes that click hotspot zones and verify status text. - Perf regression gate: Run --station-perf-eval before and after renderer changes and diff the JSON reports (fps, p95 frame gaps, interaction latencies) to catch regressions. - Use Case: Before merging a renderer change, run the headed GPU acceptance command on the dual-GPU host with Vulkan flags, capture the perf report, and diff it against the baseline to confirm no frame pacing or activation regressions. ## Quick Start Run the Station dashboard validator with the headless gate probes on a throwaway port and show me the JSON pass/fail result.
