What problem does it solve?
Producing polished product videos, landing-page loops, and screenshots for Kandev requires deterministic demo data, pixel-accurate screen capture, smooth cursor-follow camera motion, and rigorous multi-format QA, which is error-prone to do by hand.
Core Features & Use Cases
- Isolated Capture Pipeline: Records continuous 1x raw masters from a clean origin/main worktree using Xvfb, Chrome for Testing, Playwright/CDP, and FFmpeg x11grab at exact profiles (3840x2400 desktop, 1290x2796 native mobile, 25 fps).
- Reversible Semantic Camera: Builds center-biased widen-pan-tighten camera tracks from dense pointer/touch metadata, then encodes WebM, MP4, and WebP through the landing repo's tested camera/encoder scripts with a 2.0x zoom cap.
- Acceptance Gate: Enforces decoded-pixel dimension proofs, zero duplicated/dropped frames, actual-size readability audits, SHA-256 hashes, provenance records, and full teardown evidence before any promotion to production assets.
- Use Case: When asked for a new landing-page feature loop, the skill seeds demo data via product-demo-seeding, records desktop and mobile masters, applies the landing-editorial camera profile, and stages validated WebM/MP4/WebP candidates for review.
Quick Start
Ask the agent to capture a new Kandev product demo video of a feature for the landing page, and it will seed demo data, record desktop and mobile masters, and stage encoded WebM, MP4, and WebP deliveries for review.