What problem does it solve? Product and competitor analysis often relies on memory, guesswork, or uncited opinions, making teardowns untrustworthy and impossible to compare across products. This Skill captures the screens itself (via the ui-capture engine) when they do not exist yet, then produces a UX and feature teardown where every claim is cited to a specific screenshot and the analysis confidence is derived from actual capture coverage. ## Core Features & Use Cases - Capture-then-analyse in one command: Say "teardown of X" and it routes automatically — an existing capture run goes straight to analysis, a single page triggers directed capture, and a whole product or competitor triggers explore-mode capture with a coverage verification pass. - Evidence-bound analysis: A ten-section rubric (onboarding, IA, key flows, visual design, AI angle, weaknesses, and more) plus a controlled-vocabulary feature matrix where no feature is marked present without a screenshot proving it. - Honest confidence scoring: Coverage under 50% yields a Low confidence read instead of a confident wrong one, and uncaptured surfaces are named as gaps rather than invented. - Use Case: Run "teardown of Pipedrive" before a design sprint to get a rubric-structured teardown, a feature matrix comparable with past runs, and 3-6 decision-shaped takeaways framed against your own product strategy. ## Quick Start Ask the agent to run a teardown of a product, page, or competitor by saying "teardown of Pipedrive" and it will capture the screens and write the cited analysis into the capture run folder.
