agentleFS
akachandan1akachandan1/video-studio/.claude/skills/ugc-ads/SKILL.md

ugc-ads

Skill2 starsChanged 29 days agoMITView on GitHub

UGC / creator-style ad sets for Meta, TikTok, Reels — research the category's ads first, cast an AI presenter, generate talking clips with Higgsfield, cut to real-product cutaways, and build hook × ratio variants (A/B/C × 9:16 + 4:5) in HyperFrames with cuts anchored to spoken words. Also green-screen tip videos and statics.

3 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions3 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 3

    1. Pika-Labs/Pika-Plugins★ 40same name
    2. Advibly/skills★ 3same name
    3. ofoxai/skills★ 2same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository10 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Before designing anything, look at what the category runs: Meta Ad Library / TikTok Creative Center / LinkedIn ads (Apify …

    line 9
  2. Very specific:

    Use the ugc-actor skill for the profile, then generate the still.

    line 16
  3. Distinctive:

    No fake testimonials, no invented customer numbers.

    line 38

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

UGC ad sets (HyperFrames + Higgsfield)

1. Research first — always

Before designing anything, look at what the category runs: Meta Ad Library / TikTok Creative Center / LinkedIn ads (Apify apify/facebook-ads-scraper if connected, or the Arcads spy-competitor-ads skill). Write research/AD_RESEARCH.md: ~50–150 ads, the patterns that recur in long-running ones (first-frame artifact, first-person native feel, numbered steps, demo-not-claim, one segment per ad, CTA style), and 3–6 concepts. The user picks the concept(s). Brand guides supply logo/palette/fonts, not layout rules.

2. Cast the presenter

Use the ugc-actor skill for the profile, then generate the still. Be hyper-specific about region, features, styling and setting (e.g. Indian market: North Indian Punjabi features, kajal, chikankari kurta, builder-floor jaali window; "iPhone photo, unretouched, no beauty filter, no text"). Generic prompts give generic, wrong-market faces. Models: nano_banana_pro (best ethnic accuracy, ~2 cr), soul_2 (stylised realism). Show 3–4 candidates; the user picks one.

3. Talking clips (Higgsfield) — quote cost first

  • seedance_2_5 omni_reference, 720p, 15–20 s, native audio (~105–120 credits/clip): portrait + script.
  • kling3_0 pro talking clips handle Hinglish lip-sync (~14 cr / 7 s), but mangle pure Hindi.
  • Hindi/other languages: write Devanagari + English in Latin, TTS with a cloned voice (Higgsfield text2speech_v2 elevenlabs variant), then audio-first lip-sync (wan2_7, start image + audio).
  • Brand names get mispronounced (e.g. an invented brand read as a common word) — spell phonetically in the script.
  • Presenters often hold up a phone with a generated (fake) screen: always cover it with a real product cutaway.
  • Green-screen tips: npx hyperframes remove-background clip.mp4 → transparent webm over designed screens.

4. Build variants

  • Transcribe each take: npx hyperframes transcribe clip.mp4 (use medium.en if words are off); fix brand words.
  • build.py pattern: anchor every cut to a spoken word from the transcript (not to seconds), so a new take only needs a new transcript. Generate variants: hooks A (face-first) / B (product-first) / C (proof/number-first) × 9:16 + 4:5 → one HyperFrames root per variant (write each to index.html and render in a loop — a project allows one root composition).
  • Cutaways: real screen recordings (cropped tall, sped up) or HTML rebuilds of the real UI labelled "Demo". No fake testimonials, no invented customer numbers; claims = product mechanics.
  • Captions via caption-video styles (Clean Box / Karaoke work best for UGC), CTA end card.
  • Render all, loudnorm −14 LUFS, name Brand_Concept_Hook_Ratio.mp4.

5. Statics (optional)

HyperFrames still frames exported with npx hyperframes snapshot . --at <t>. For premium statics: AI lifestyle scene with a chroma-green screen/paper, then composite the real product art in with an OpenCV homography — never let an image model draw the product/artwork itself.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.