Magnific adds Fable 5 MCP workflow for branded product landing pages

• Magnific published a repeatable site-build workflow that starts with product context in its Library, passes that context through the Magnific MCP, and uses Fable 5 to assemble a branded landing page, according to Magnific's workflow overview and Magnific's Library step.
• The page prompt is unusually concrete: Magnific's prompt template asks for a product showroom landing page tied to a brand's guidelines and four named products from the user's Magnific library.
• Motion is a separate layer, not an afterthought. In Magnific's video prompt, the company wires two Seedance 2.0 4K hero videos into the page, one scroll-scrubbed and one cursor-reactive.
• The final demo in Magnific's result post shows the pitch clearly: product pages with interactive motion that stays visually on-brand because the same product context travels through the workflow.
• Magnific is also spreading the same context stack across other entry points. Magnific's Claude connector post, Magnific's Gemini Omni announcement, and Magnific's editor plugin announcement all point to the same bet, AI creative work should happen inside a shared asset and model layer.

You can watch the full workflow thread, grab the exact page prompt, and inspect the Seedance motion prompt without much guesswork. The interesting bit is how much of the page is specified up front: product names from the Library, brand guidelines from a Space, then two explicitly scripted interactions for motion. Magnific is also framing the same stack as portable, from Claude via MCP to editing apps and model drops like Gemini Omni.

Magnific's first step is not page generation. It is asset normalization. In Magnific's Library step, the company says uploading a product to the Magnific Library lets "every tool" know products by name, including the MCP, the image generator, and any workflow.

That makes the Library the story's real primitive. Instead of re-describing a bottle, cap, packaging system, or SKU family in every prompt, the workflow treats those items as named references that can be reused across page layout, image generation, and video generation.

The companion thread in Magnific's workflow overview frames the end goal as a full website with cursor-reactive videos that keeps visuals on-brand through the Magnific MCP.

The page assembly prompt is short enough to matter. According to Magnific's prompt template, the base instruction is: build an interactive product showroom landing page for the brand in a named Space's brand guidelines, then use four specific products from the Magnific library.

That prompt bakes three constraints into one request:

1. Brand rules come from a Space.
2. Product identity comes from the Library.
3. Page structure comes from Fable 5.

Magnific does not present Fable 5 here as a general-purpose chatbot. It is the assembly layer that turns brand context and product context into a concrete landing page skeleton.

The motion prompt is more detailed than the page prompt, and that is where the workflow stops looking like a generic "AI website" demo. In Magnific's video prompt, the company specifies two different Seedance 2.0 4K sequences and tells the model how each one should behave inside the page.

The two hero blocks are:

• Scroll-scrubbed hero: a centered bottle rotates 360 degrees, the cap closes, translucent mist membranes "breathe" around it, and the sequence stays pinned for about three screen heights.
• Cursor band: a centered bottle stays in view while translucent mist membranes gather left or drift right based on mouse or touch movement.

That is a cleaner division of labor than most creative AI demos show. Fable handles page structure. Seedance handles motion shots. The prompt then binds those shots to specific interaction rules.

Magnific's final post in the result reveal shows the composite output, including the polished page and the motion-led product presentation.

The workflow lands harder because Magnific keeps presenting the same system through multiple surfaces. Magnific's Claude connector post makes the Claude connection explicit, while Magnific's Gemini Omni announcement and Magnific's Gemini Omni feature post pitch plain-language, memory-backed scene editing through the same Magnific layer.

That portability shows up in the editor side too. In Magnific's editor plugin announcement, the company says its plugin is available inside Premiere Pro, After Effects, DaVinci Resolve, and Final Cut Pro, with generation, upscaling, relighting, reframing, background removal, AI music and voice, and access to 250 million stock assets.

For creators, the product-page demo reads less like a one-off campaign build and more like a storefront kit sitting on top of shared brand assets, shared product references, and a growing menu of model backends.