UGC Content Studio · Open Skill 公開スキル
UGC Content Studio
UGC Content Studio は Digen 上の公開 Open Skill です。開くと、このワークフローが何を作るか分かります。
SocialAd
5 1 12
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/3521128/2026-09-01/bftrwt9y.jpg
# UGC Content Studio ## When to use - The user wants authentic "user-generated content" style short videos — the raw, shot-on-a-phone, one-creator-talk-to-camera look used in ads and organic posts — for TikTok, Instagram Reels, or YouTube Shorts (default 9:16 vertical). - The user has a product, brief, brand, script, idea, reference video link, or product photos, and wants a complete production: trend-aware script → shot list → visual stills → short clips → music/voiceover → final stitched video. - The user already has one of these steps done and only wants to advance or change a single step (rewrite the script, generate the next clip, restyle a still, remix the final video, etc.). - The user asks for several 10-second-ish UGC episodes or a UGC series in one session. ## Goals - Deliver a complete, ready-to-post vertical UGC video plus a production package (script, on-screen caption lines, shot list, style & posting notes) in one workflow. - Keep the UGC aesthetic: handheld phone-camera feel, natural lighting, imperfect framing, realistic people and textures, casual conversational tone — not glossy studio advertising. - Be transparent that intermediate stills/clips are visible on the canvas as they are made (user skills cannot hide them), and report honestly if any generation misses the brief. ## Steps ### 0. Lock the brief (compact, 1 round) Ask for anything not yet known, using defaults when the user is in a hurry: - Product/brand and what it does (if the user pastes a product URL, use `search_web` to extract basic facts, but do not invent specs). - Goal: awareness, sales/CTA, tutorial/demo, testimonial/social proof, or entertainment. - Target platform (default: TikTok + Reels, 9:16 vertical), target length (default ~30s; 15s or 60s variants ok). - Creator vibe: ordinary consumer, specialist/enthusiast, or "server/host" type; age/appearance/voice if the user cares; otherwise pick a realistic, diverse default and state it. - Tone: hype, calm/trustworthy, funny, honest-review, curiosity-driven. - Must-have talking points and the CTA (e.g. "link in bio", "comment CODE", "follow for more"). - If the user provides product photos or a creator photo, collect them — they become generation references. If not, generate a generic but realistic product (never fake a trademarked logo). - Language of script/captions (default: same as the user's message language). - If the user requests adult/NSFW UGC, call `switch_model(tier="uncensored", reason=...)` first and note that the .nsfw channels/variants will be used for any video generation. ### 1. Trend research (optional but recommended, max 2 searches) - `search_web` for current hooks/trends: e.g. `<niche or product category> UGC ad hooks 2026`, or `TikTok <niche> trend this month`. - Distill 1–3 usable trend angles into the script. If the user shared a reference video link (YouTube/TikTok/Instagram), get its download link with `download_video` (the returned `downloadUrl` may look like `shorturl://...` — pass it verbatim to the next tool), then analyze it with `recognize_video` to extract hook style, pacing, shot types, and on-screen text patterns to imitate (structure only — never copy protected audio/visuals wholesale). ### 2. Write the production package (script + shot list) Write a markdown production package with this structure: - **Concept**: one-line idea + why it fits the platform trend. - **Creator lock** (a copy-paste block reused in every image prompt): one specific human (gender, age range, hair, outfit, setting, phone in hand, natural window light) — e.g. "female creator, early 20s, wavy brown hair, oversized white t-shirt, cozy bedroom lit by window, holding smartphone". Keep this identical across all stills for consistency. - **Beat-by-beat script** (default 5 beats, ~30s): 1. Hook (0–3s): pattern interrupt / bold claim / direct question at camera. 2. Problem & agitation (3–9s). 3. Solution intro + quick demo (9–20s). 4. Proof / social proof (20–25s). 5. CTA (2