creator-ad
Skill7 starsChanged 16 days agoMITView on GitHub
Make short-form video ads and UGC-style ads for a product or app with whatever providers the user has: hook-first script, presenter or product shots, voiceover, captions, music, and placement-ready exports (9:16, 1:1, 16:9). Uses Higgsfield Marketing Studio when available, otherwise builds the ad from generated clips or animated stills plus voiceover. Use when the user asks for a video ad, UGC ad, TikTok/Reels/Shorts ad, product promo, unboxing, testimonial-style, before/after, or ad variations/hooks to test. Not for static product photos (creator-product-photo) or long explainers (creator-explainer).
In this repository10 other files it gives its agents.
- AGENTS.mdAGENTS.md
- creator-audioSkill
- creator-brandkitSkill
- creator-characterSkill
- creator-explainerSkill
- creator-imageSkill
- creator-product-photoSkill
- creator-setupSkill
- creator-thumbnailSkill
- creator-videoSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 31
Write 3 hooks × 1 body using references/ad-formats.md (formats: problem→solution, UGC testimonial, unboxing, before/after, 3 reasons, demo).
- Distinctive: line 70
No fake testimonials presented as real customers, no invented stats, prices, or claims.
- Distinctive: line 72
No real people's likeness without consent.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Creator Ad
Project folder: ads/<product-slug>/.
Step 1 — Brief
Collect (ask once, together, only what's missing): product (photos or URL content the
user pastes), audience, single key benefit, offer/CTA, length (default 15 s), placements
(default 9:16). Read config brand.
Step 2 — Script variants
Write 3 hooks × 1 body using references/ad-formats.md (formats: problem→solution,
UGC testimonial, unboxing, before/after, 3 reasons, demo). Each script: hook (0–3 s), body,
proof, CTA, ≤ 40 words for 15 s. Show them; user picks or approves all.
Step 3 — Route
python3 scripts/creator.py detect| Available | Build |
| Higgsfield signed in | Marketing Studio: higgsfield generate create marketing_studio_video --prompt … --mode ugc --aspect_ratio 9:16 --duration 15 --wait (product import / avatars via higgsfield marketing-studio --help). Captions/CTA end card added locally |
| Video provider | Per beat: product-locked still (creator.py image --ref product.jpg) → creator.py video --start |
| Image provider only | Per beat: product-locked stills → Ken Burns via media.py assemble |
Voiceover per beat with creator.py speech (pinned voice, energetic but natural style;
--lang for the market's language — Arabic dialect ads: references/arabic-voice.md;
add tashkeel to meaning-changing words before any TTS, or speech stops with needs_tashkeel).
Presenter faces: use creator-character packs only for the user or consenting people.
Step 4 — Assemble each variant
scenes.jsonper variant (hook scene differs, body shared), thenmedia.py assemble --ratio 9:16 --music music/bed.mp3.- Captions: big, centered-lower, ≤ 5 words per line →
media.py captions. - End card: product hero still + CTA via
media.py text, 1.5–2 s scene. - Other placements:
media.py reframe --ratio 1:1|16:9 --mode blurand re-check that product and captions stay in frame.
Step 5 — QA and deliver
- Hook visible and audible in the first 1–2 s; product appears by 3 s; CTA on screen ≥ 1.5 s.
- Captions inside safe zones (avoid top 12% and bottom 20% on 9:16 for platform UI).
- Product matches the reference in every shot.
- Deliver files per variant/placement with one line on what each tests.
Rules
- No fake testimonials presented as real customers, no invented stats, prices, or claims; label AI presenters if the platform or law requires it.
- No real people's likeness without consent; no competitor trademarks in visuals.
References
references/ad-formats.md— formats, hook bank, timingreferences/audio.md,references/video-prompting.md,references/media-ops.md,references/providers.md
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

