Parallel visual generation across multiple AI + stock search + website mockups & video from a URL.
Uses
APIs to connect
This skill needs keys for these services. The link takes you where to get each one:
What it does
Parallel visual generation across several AI providers (Recraft v3, Gemini Nano Banana, Imagen 4, OpenAI, Flux) + stock search (Unsplash / Pexels / Pixabay), a browser preview grid, optional Cloudinary upload and vectorize/compress. 3 price tiers. Plus a `/content-mockup` branch: from a website URL it produces a batch of polished mockups — a screenshot in a browser / iPhone / iPad / MacBook frame, or a device in a themed context (laptop in a sauna, phone in hand), with a Tinder-style pick of the visual direction and volume adapted to the page's image slots (desktop + mobile). New — photorealistic quality: the device sits in a generated scene with real lighting (the scene's glare, shadows and reflections are transferred onto the inserted screenshot, screen pixel-perfect), the screenshot is captured to the exact screen aspect (no cropping), and fingers correctly occlude the content. Plus pro PSD templates via the Dynamic Mockups API — in a blind comparison the local engine beat the paid service on realism. New `/content-screencast` branch: from a URL it produces ready website videos in a device frame — a perfectly smooth 60fps scroll (the trick: not a recording of the real, janky scroll, but a smooth pan across one long screenshot), recorded real clicks and transitions via a scenario, gifs and posters for previews, with every step verified and auto-uploaded to Cloudinary.
How to use
- 1“generate an image of X” or
/content-gen <prompt>. - 2Pick a tier: explore / mid / premium.
- 3Browse the variant grid, pick the best.
- 4Mockups:
/content-mockup <url>→ website screenshot → Tinder of directions → batch fitted to slots. - 5Photoreal mockups: own scene + scene-light transfer onto the screen (local, free) or a PSD via Dynamic Mockups.
- 6Website video:
/content-screencast <url>→ smooth 60fps scroll in an iPhone / browser / MacBook frame (desktop + mobile), optional scripted click recording, gifs and posters. - 7Optional: vectorize to SVG / compress, upload to Cloudinary.
When it triggers
Example
/content-mockup https://my-site.com — screenshots the site, shows a Tinder of directions, builds a batch of mockups in device frames and in context (desktop + mobile).
Updates
New `/content-screencast` branch — website video from a single URL. Smooth 60fps scroll in an iPhone / browser / MacBook frame (the trick: not a recording of the real, janky scroll, but a smooth pan across a long screenshot), recorded real clicks and transitions via a scenario, gifs and posters for previews. Every step is verified (broken video can't slip through), idempotent auto-upload to Cloudinary with no duplicates.
Mockups leveled up to photorealism. The device now sits in a generated scene with real lighting: the scene's glare, shadows and reflections are transferred onto the inserted screenshot (screen pixel-perfect, no diffusion text artifacts). The screenshot is captured to the exact screen aspect — the site fits fully with no cropping, fingers correctly occlude the content, corners are rounded. In a blind comparison the local engine beat the paid Dynamic Mockups on realism; the PSD route via their API stays for ready-made third-party templates.
Open-sourced — Redloft/content-gen repository (MIT). Ready-made Claude Code install prompt via the button on the card.
New `/content-mockup` branch: from a site URL — a batch of mockups (browser/iPhone/iPad/MacBook frames and a device in a themed context), a Tinder-style direction pick, volume fitted to page slots (desktop + mobile).
Added to the RedSkills catalog.
Liked this skill? New breakdowns and updates land in the channel.