Street Interview Skill — Design skill for Claude Code
Claude skill that builds street-interview style video ads for Seedance 2.0 — question ladder, casting prompts, and every video prompt with the reference chain mapped.
How to install Street Interview Skill
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open mikefutia/street-interview-skill and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Street Interview Skill does
Claude skill that builds street-interview style video ads for Seedance 2.0 — question ladder, casting prompts, and every video prompt with the reference chain mapped.
Alternatives in Design
- Pack — Bundle every designlang output (DTCG tokens, Tailwind, shadcn, Figma vars, motion, anatomy, Storybook, prompts 3.6k ★
- Drawbridge — Design editor for Claude Code and Cursor 960 ★
- Scaffold Project — Scaffold a new design-product project that matches the recommended Claude Code layout (the reference structure 781 ★
README
Street Interview Skill
A Claude skill that builds street-interview style video ads for Seedance 2.0 — the format where somebody gets stopped on the sidewalk, gets asked an uncomfortable question, and ends up trying the product on camera.
Hand it a product. It returns the interview script, image prompts to generate the person being interviewed, and every Seedance prompt in order with the reference chain worked out.
What it gives you
- A question ladder, not marketing beats — Concession → Dead product → Handover → Ruling
- Three casting prompts for ChatGPT Images 2.0, written to fight the poreless AI-model look
- A prompt per segment, each with its own attachment list
- The reference chain, so the face, the voice, and the street stay identical across every clip
Install
- Download this repo (Code → Download ZIP)
- In Claude, go to Settings → Capabilities → Skills
- Upload the folder containing
SKILL.md
Use it
In a normal Claude chat:
Make me a street interview ad for [product] — [one line on what it is]
It asks three questions — angle, who gets stopped, and how long — then builds the whole thing in one reply.
What you'll need
- Claude for the script and prompts
- ChatGPT Images 2.0 (or Nano Banana) to generate the interviewee
- Seedance 2.0 via Higgsfield or fal.ai to generate the video
- A real photo of your product
- CapCut or any editor to cut for pace
How the continuity works
Seedance caps every generation at 15 seconds, so a 60-second ad is four separate clips. The trick is that each clip after the first uses a trim of the **previous clip** as its video reference. That carries the face, the wardrobe, the street, the light, and both voices forward — which is what makes four generations read as one conversation instead of four shoots.
The skill maps this out per segment so you know exactly what to attach and when.
Want the full walkthrough?
The skill is free — everythi
Related Skills
C'est Quoi ? What Is This?
15 compétences Claude spécialisées conçues pour Seedance 2.0 (Higgsfield). Chaque compétence transforme Claude
Vox Explainer Skill
A Claude Code skill that turns one topic prompt into a finished Vox-style explainer video. Script, keyframes,
Nano Banana UI Prompts
Claude Code skill that adapts the 12k-prompt Nano Banana Pro library for UI workflows — Google Stitch MCP scre
Best Practices Research
Reference guide for researching and applying UI/UX, prompt engineering, and technical implementation best prac
Case Practice
Mock case interview practice (McKinsey/BCG/Bain style)
Design Clarify
Interactive command that transforms vague requirements into optimal concrete design via dynamic question gener
Related Agents
Scene Prompt Smith
Use this agent when drafting kie.ai Nano Banana 2 image prompts for the scenes of a short video. It turns a sc
Evaluator Svm Agent
SVM (Solana) smart-contract vulnerability evaluator — spawned per smart_contract surface with chain_family=svm
Marketing Manager
Marketing strategist covering SNS, ads, content, and campaigns. Orchestrates creative production with image/vi