Skip to content
← Tools / Skills
Skill · .md packageVincentzdq/10 trustMetrics-checked
Copywriting Interview logo

Copywriting Interview

Interview-first copywriting workflow: the AI asks before it writes

Visit official site →

About

A single-skill package that changes how agents write copy: instead of drafting immediately, the AI interviews the user under a four-scenario protocol to surface real intent and material first. Covers vague ideas, no ideas, cold starts, and field-only context, with one-question-at-a-time discipline and a requirement brief you sign off. Ships bilingual in English and Chinese.

What it can do

  • Interview the user with a four-scenario protocol before writing anything
  • Convert interview answers into a content framework and requirement brief
  • Trace every fact in the draft to a source instead of fabricating

Setup

git clone https://github.com/Vincentzdq/copywriting-interview ~/.claude/skills/copywriting-interview

For Codex CLI: clone into ~/.codex/skills/copywriting-interview

For claude.ai: upload copywriting-interview.skill under Settings → Skills

Prerequisites

Claude Code, Codex CLI, or claude.ai web access

Use cases

writing marketing copydrafting video scriptsbrainstorming content topics

Works with

Claude CodeCodex

Trust score — 5/10

/10 trustMetrics-checked

Scored 5/10 (metrics tier, metrics-v1). Evidence: Interview-first copywriting skill (1 star, MIT, 2 commits); four-scenario protocol and install paths verified from the README on 2026-10-13.; The 12/12 vs 7/12 A/B test is author-reported on a small sample, not independent evidence.; Bilingual English/Chinese; no harness plugin — install is a manual git clone or .skill upload.. License: MIT per GitHub license field..

★ 1 GitHub stars

How the trust score is built

Full methodology →
Adoption · stars/downloads, log-scaled vs peers
4.0/10
Maintenance · recency, releases, issue load
4.0/10
Openness · license + docs
8.0/10
Manual review · sanity read + red-flag check
6.0/10
  • · Interview-first copywriting skill (1 star, MIT, 2 commits); four-scenario protocol and install paths verified from the README on 2026-10-13.
  • · The 12/12 vs 7/12 A/B test is author-reported on a small sample, not independent evidence.
  • · Bilingual English/Chinese; no harness plugin — install is a manual git clone or .skill upload.

Sources

Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.