Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Brief gap cho designer khi campaign dang chay va can asset moi trong ngay (creative fatigue, CTR drop, test angle moi): rut gon input, nhan ban tu creative winner, template 1 trang, SLA ro rang. Kich hoat khi nhac: brief nhanh, can asset gap, creative fatigue, quick brief, asset trong ngay.
.claude/skills/minhnv0807-48-quick-visual-brief/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 49% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 53% | 0% |
> Dung khi campaign dang chay va can asset gap: toc do quan trong hon perfection. Variant moi chay duoc trong 4 tieng tot hon asset hoan hao xong sau 2 ngay. Brief du de lam, khong can du de dep.
| Tinh huong | Dung skill nao | |------------|----------------| | Creative fatigue, CTR drop, frequency cao, can variant trong ngay | Skill nay (48) | | Test angle moi tren nen creative dang work | Skill nay (48) | | Asset moi hoan toan, concept moi | 42-brief-hinh-anh (brief day du) | | Landing page, video, email | 12 / 44 / 49 (khong quick duoc) | | Khong co creative winner nao de base vao | 42-brief-hinh-anh |
Gop trong 1 tin nhan, toi da 4 cau:
21-audit-ads-performance.Creative nao CTR cao nhat dang chay? Do la base de nhan ban. Lay link file goc (Canva/Figma) — luon nhan ban tu file goc, khong tu file export.
Dung template ben duoi. Chi dien bang "thay doi cu the" — moi thu khac giu nguyen theo base.
Duplicate base → thay dung element da ghi → export dung kich thuoc → nop.
Message dung? Kich thuoc dung? CTA co? Dat → day len ads ngay. Khong co feedback sau 1 tieng → designer coi la approved va nop file final.
Ten file: quick-brief-[campaign]-[YYYYMMDD].md
markdown# QUICK BRIEF — [Ten asset] — [ngay] Deadline: [gio:phut ngay X] (khong flexible) Lam trong: 2-4 tieng Base tu: [link Canva / ten file goc — creative CTR cao nhat] ## Base template - Duplicate tu: [link] - Giu nguyen: [layout / background / logo / font] - Chi thay: [copy / mau / image / CTA — chon 1 thay doi chinh] ## Thay doi cu the | Element | Noi dung cu | Noi dung moi | |---------|-------------|--------------| | Headline | | | | Subtext | | | | CTA | | | | Visual / image | | | ## Kich thuoc can - [ ] 1080x1080px — FB Feed - [ ] 1080x1920px — Story (neu can) ## Export & nop - Format: JPG - Nop qua: [Zalo / Drive folder] - Ten file: QUICK_[Campaign]_[Ngay]_v1.jpg ## Review - Check 3 diem: message dung / kich thuoc dung / CTA co - Feedback trong: 30 phut sau khi nop - Khong feedback sau 1 tieng = approved ## Context (1-2 cau) [CTR dang drop tu X% / frequency vuot nguong / test angle moi: ...]
| Buoc | Ai | Thoi gian | |------|----|-----------| | Viet quick brief | Leader / Media Buyer | 30 phut tu khi phat hien van de | | Lam asset | Designer | 2-4 tieng | | Review | Leader / Media Buyer | 30 phut sau khi nop | | Khong feedback | — | Sau 1 tieng = approved | | Launch day standby | Designer | Phan hoi yeu cau trong 2 tieng |
21-audit-ads-performance: nguon data xac dinh creative fatigue (frequency, CTR) va winner de base.19-ab-test-setup: variant moi tu quick brief nen vao khuon A/B test — 1 thay doi = 1 bien.41-campaign-asset-list: them dong asset moi vao list + update trang thai.42-brief-hinh-anh: fallback khi khong co winner de nhan ban — brief day du.50-asset-resize: khi variant thang, resize ra cac kich thuoc con lai.47-design-review: quick asset chi check 3 diem — khong can full review 40 diem.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 19,291 | 15,256 | -21% | 1 | 1 | 0% | 2,674 | 3,466 | +30% | 0 | 0 | — |
case-02 | fail→fail | 14,987 | 9,314 | -38% | 1 | 1 | 0% | 1,953 | 3,291 | +69% | 0 | 0 | — |
case-03 | fail→fail | 19,537 | 11,300 | -42% | 1 | 1 | 0% | 2,585 | 3,664 | +42% | 0 | 0 | — |
case-04 | fail→pass | 31,522 | 18,188 | -42% | 1 | 1 | 0% | 4,095 | 3,601 | -12% | 0 | 0 | — |
case-05 | fail→pass | 24,551 | 23,975 | -2% | 1 | 1 | 0% | 3,905 | 5,242 | +34% | 0 | 0 | — |
case-06 | fail→pass | 33,862 | 31,287 | -8% | 1 | 1 | 0% | 4,550 | 5,503 | +21% | 0 | 0 | — |
case-07 | fail→fail | 20,896 | 27,585 | +32% | 1 | 1 | 0% | 3,174 | 5,016 | +58% | 0 | 0 | — |
case-08 | fail→fail | 19,292 | 16,662 | -14% | 1 | 1 | 0% | 2,269 | 3,202 | +41% | 0 | 0 | — |
case-09 | fail→pass | 19,170 | 11,368 | -41% | 1 | 1 | 0% | 2,331 | 3,464 | +49% | 0 | 0 | — |
case-10 | fail→pass | 19,645 | 22,301 | +14% | 1 | 1 | 0% | 2,734 | 4,177 | +53% | 0 | 0 | — |
case-11 | fail→pass | 20,979 | 22,284 | +6% | 1 | 1 | 0% | 2,337 | 4,130 | +77% | 0 | 0 | — |
case-12 | pass→pass | 12,743 | 15,868 | +25% | 1 | 1 | 0% | 1,835 | 3,195 | +74% | 0 | 0 | — |
case-13 | pass→pass | 18,744 | 12,728 | -32% | 1 | 1 | 0% | 2,188 | 2,705 | +24% | 0 | 0 | — |
case-14 | pass→pass | 14,777 | 14,745 | -0% | 1 | 1 | 0% | 2,509 | 3,897 | +55% | 0 | 0 | — |
case-15 | fail→pass | 19,476 | 11,776 | -40% | 1 | 1 | 0% | 2,049 | 3,181 | +55% | 0 | 0 | — |
case-16 | fail→pass | 15,296 | 4,198 | -73% | 1 | 1 | 0% | 1,720 | 1,958 | +14% | 0 | 0 | — |
case-17 | fail→pass | 19,118 | 8,469 | -56% | 1 | 1 | 0% | 2,092 | 2,084 | -0% | 0 | 0 | — |
case-18 | pass→pass | 11,287 | 3,587 | -68% | 1 | 1 | 0% | 1,662 | 2,060 | +24% | 0 | 0 | — |
case-19 | fail→pass | 19,851 | 19,219 | -3% | 1 | 1 | 0% | 1,961 | 2,829 | +44% | 0 | 0 | — |
case-20 | fail→fail | 14,065 | 1,944 | -86% | 1 | 1 | 0% | 1,447 | 1,810 | +25% | 0 | 0 | — |
case-21 | fail→pass | 20,176 | 21,809 | +8% | 1 | 1 | 0% | 2,332 | 4,176 | +79% | 0 | 0 | — |
case-22 | fail→pass | 16,262 | 15,215 | -6% | 1 | 1 | 0% | 1,876 | 3,342 | +78% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +55 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.