What problem does it solve? Producing a believable UGC-style talking-head ad requires getting many details right at once: a photoreal actor, a script that sounds like a real person, natural voice delivery, and two generated clips that join without a visible seam. This Skill encodes those requirements as machine-enforced validation gates so every generation prompt is assembled and checked before any paid video or image call is made. ## Core Features & Use Cases - Gated prompt generation: The agent writes structured JSON for casting and each video clip; gate scripts validate it against data glossaries (age-correct skin physiology, a 32-entry micro-behavior glossary, a six-category voice spec, color-grading vocabulary) and assemble the final prompt only on PASS. - End-to-end ad pipeline: Researches the product, tears down 5 competitor Meta ads and 5 organic TikToks on the MaxFusion canvas, writes a pain-driven script, reverse-prompts a rights-clean actor from a user-supplied photo, and generates two chained 10-second Gemini Omni clips. - Seamless stitching: adkit.py measures speech, trims dead air, extracts clip 1's final frame to seed clip 2, verifies the seam (under 5/255 mean pixel difference), and stitches one 9:16 MP4 deliverable. - Use Case: Ask for an ad for your app, supply an actor reference photo and competitor leads, and receive a finished vertical MP4 of a realistic person delivering a natural-sounding testimonial. ## Quick Start Ask the assistant to make a UGC ad for your product and provide an actor reference photo when prompted.
