What problem does it solve? Reproducing a reported UI bug or proving a fix works requires manually driving VS Code and assembling evidence by hand. This Skill automates that by running a scripted scenario against a real VS Code window and producing a captioned video, per-step screenshots, a Playwright trace, and an HTML report ready to attach to an issue or pull request. ## Core Features & Use Cases - End-to-end scenario execution: Write a small JavaScript scenario file with steps that drive the workbench via Playwright, then run it against an installed Insiders/Stable build, a dev build, a specific install, or web. - Automatic evidence capture: Every run produces an annotated MP4 with per-step captions, per-step screenshots, a Playwright trace, a manifest.json, and a report.html. - Honest handling of unautomatable steps: Steps that need a human or a missing harness capability are explicitly skipped, recorded, and surfaced in the report rather than silently dropped. - Use Case: Given a GitHub issue claiming Settings search fails to match a setting by its description, write a scenario that opens the Settings editor, searches, scrolls the virtualized list, and asserts the setting appears — then attach the failing run's video to the issue as reproduction proof. ## Quick Start Ask the AI to reproduce a specific VS Code issue by writing and running a UI scenario, for example: reproduce issue 250159 and attach the recorded evidence.
