founder-product-video

Generate a 65-second founder-style product video from a product URL and user imagery via Pika MCP.

1|Updated Jun 20, 2026
One-click install
npx skills add https://github.com/BeanyZoldyck/moggie --skill founder-product-video-beanyzoldyck
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: founder-product-video
Source: https://github.com/BeanyZoldyck/moggie/tree/main/.agents/skills/founder-product-video
Command: npx skills add https://github.com/BeanyZoldyck/moggie --skill founder-product-video-beanyzoldyck

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Producing a founder-style product pitch video normally requires filming, editing, motion graphics, and music licensing. This Skill automates the entire pipeline: it turns a product URL plus founder photo and brand assets into a 65-second 1080p MP4 with four AI-generated talking-founder acts, a branded end card, captions, and background music. ## Core Features & Use Cases - Brief and asset analysis: Derives product name, tagline, features, tone, and CTA from the product URL, then classifies each supplied image (digital screen, apparel, physical object, consumable) to drive the correct product-reveal shots. - Script and acting direction: Writes a 4-act founder pitch script with per-line emotion and physical beats, wardrobe locking, and camera transition choreography for SeeDance video generation. - Full MCP render pipeline: Generates four 15-second SeeDance acts in parallel, concats them, generates instrumental music, burns in captions, renders a branded HTML end card, and mixes audio into a final ~65-second MP4. - Use Case: A startup founder pastes their product URL, name, role, and a photo, and receives a ready-to-share 16:9 pitch video where their actual product screenshots appear on the founder's phone in reveal shots. ## Quick Start Run the founder-product-video skill with your product URL, founder name and role, and a founder photo to generate a 65-second branded pitch video.

Frequently Asked Questions about founder-product-video

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate a founder-style product video from a URL?

Invoke the skill with the product URL, founder name and role, and a founder photo (local path, URL, or the literal 'generate'). The pipeline analyzes the product brief, writes a 4-act script, generates four 15-second SeeDance acts, and assembles a 65-second MP4 with music and an end card.

What inputs are required to make a talking founder explainer video?

Three inputs are required: a product URL, founder identity (name and role), and a founder photo or the 'generate' option. A brand kit folder is also required; if none exists, the build-a-brand skill is invoked first. Optional extras include custom screenshots, music, aspect ratio, and location.

Does the video show my real product screenshots or AI-invented UI?

Your actual product screenshots are passed as reference images to the SeeDance generation, so reveal shots show the real UI on the founder's phone. The skill classifies each asset by type (digital screen, apparel, physical object, consumable) to pick the correct reveal pattern.

Can I run the founder video pipeline non-interactively from CI?

Yes. Pass --quick or --config <path> with a JSON file containing url, founder fields, brand kit path, and optional extras. The fast lane skips all confirmation prompts, auto-builds the brand kit if omitted, and records assumptions inline.

Why does the generated founder look frozen or wear different clothes between acts?

Frozen performances come from beats with only facial micro-expressions; every beat needs a hand gesture, torso shift, or larger head movement. Wardrobe drift is fixed by repeating an explicit wardrobe_lock sentence in all four act prompts, since each 15-second act is generated independently.

What are the limitations of the Pika MCP video pipeline?

SeeDance renders one environment per 15-second clip, so multi-shot acts need explicit camera transitions or hard cuts. Real-person photos can trip likeness moderation, SVG logos cannot be uploaded, and arbitrary transparent lower-third overlays still require one local ffmpeg pass.