ab-test-cardlisted
Install: claude install-skill ali-demirbas/ab-test-playbook
# ab-test-card — Scenario Card Rendering
> **Language:** Output always matches the language you write in (CLAUDE.md rule 7).
The visual language is defined in `${CLAUDE_PLUGIN_ROOT}/knowledge/mockup-style.md` — read it before producing anything. Template: `${CLAUDE_PLUGIN_ROOT}/templates/scenario-card.html`.
This skill runs automatically for EVERY scenario that `ab-test-suggest` or `ab-test-design` produces in a turn (CLAUDE.md rule 9) — the user doesn't need to ask separately. The full content of the three boxes ("What to test" / "Primary KPIs to track" / "Never do") lives only in this card; the same content isn't also written to chat as text — chat only keeps the title, source tag and a one-sentence summary. 2-5 scenarios in a turn become cards directly; if there are more than 5 strong candidates, they aren't all produced without asking (rule 9). The same flow runs if the user directly says "make a card."
## Flow
0. **Brand source (once per session).**
- **Don't ask if the user shared a screenshot/page:** pull the brand color, logo text and button style straight from the image and use them. Add a one-line note under the card: "I took the colors from the screenshot; send me the official guide and I'll update it." Don't stop the flow and wait for an answer.
- **Ask if there's no visual source:** "Should I prepare the card to your brand guide (logo, color palette), or use a neutral style?" Wait at this step until an answer comes. If the user gave a URL, that also co