Communitygithub.com

tiktok-photo-mode

>-

tiktok-photo-mode とは?

tiktok-photo-mode is a Claude Code agent skill that >-.

対応~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/social-media-skills/skills/tree/main/skills/tiktok-photo-mode

お気に入りのAIに質問する

このエージェントスキルを事前に読み込んだ状態で新しいチャットを開きます。

ドキュメント

tiktok-photo-mode は何をしますか?

A format-execution skill — produce a TikTok-native Photo Mode carousel (craft for completion) and the exact WoopSocial PHOTO publish config. The agent writes the plan + overlay copy + caption + config; the images are made by a human or an image skill; WoopSocial publishes the native PHOTO post.

The POV: win completion, don't chase a reach hack

TikTok treats Photo Mode like video and ranks it on completion rate — what % swipe to slide 2+ and through the end. So every choice serves "make them swipe to the end": a slide-1 hook that promises a payoff, one idea per slide with cliffhangers, a reward + one CTA at the end. But be honest: photo mode is a high-ceiling, niche-dependent format, still outnumbered ~5:1 by video with declining average per-post performance as it saturates — it can punch far above its weight for the right content type, and underperforms video when forced. Never promise virality. Design 9:16 (1080×1920), not Instagram's square, and publish it as a native PHOTO post (not an MP4 slideshow).

Read these first

  1. brand-profile — niche + voice (does a swipeable format fit?).
  2. goals-and-kpis — the goal (saves / comments / follows / reach).

The framework: SWIPE

(Depth: references/the-swipe-framework.md.)

  • S — Stop the scroll on slide 1: the thumbnail hook (bold visual + overlay question/claim) that promises a payoff.
  • W — Win the niche fit: pick a swipeable type (listicle / before-after / tutorial / hot take / photo dump / story / myth-bust); don't force it — photo mode is niche-dependent.
  • I — Itemize for completion: one idea per slide, progressive reveal, cliffhangers; 5–7 slides (max 35); 9:16 1080×1920, content centred in the safe zone.
  • P — Pay off + prompt: reward on the final slide + one CTA; question in the caption; don't beg for likes.
  • E — Equip the publish config: postType=PHOTO, privacyLevel (from platform-inputs), allowComment, autoAddMusic (applies to PHOTO), truthful isYourBrand/isBrandedContent, DIRECT_POST vs MEDIA_UPLOAD; 1–2 hashtags + overlay keywords (TikTok SEO); validate before publish.

The reality (verify-quarterly)

The honest performance tension (high-ceiling vs video-dominance + declining averages); completion rate as the primary signal; 9:16 1080×1920 + safe zone; 5–7 slides; hook / progressive reveal / cliffhanger / reward; sound (autoAddMusic) + text overlays as TikTok SEO; 1–2 hashtags; native-vs-MP4; winning content types + engagement tactics: references/photo-mode-2026-reality.md. Content-type menu, slide blueprint, the full PHOTO publish-config block + worked examples: references/carousel-recipes-and-publish.md.

Honest scope (never violate)

  • The agent writes the slide plan + overlay copy + caption + publish config. The images are produced by a human or an image skill (image-prompt/ideogram/nano-banana) — WoopSocial doesn't generate images.
  • WoopSocial publishes the native PHOTO post (postType=PHOTO, raw-bytes upload) and auto-discloses AI on TikTok. No update endpoint → edit = delete + recreate (only while NOT_STARTED).
  • Disclosure flags set truthfully (isBrandedContent when paid/gifted; isYourBrand for own-brand promo) — never falsely "organic." Never promise virality / a "5×"; never fabricate metrics (native analytics only). (Scope, distinctions + connections: references/scope-and-connections.md.)

Distinct from its siblings (route correctly)

tiktok-photo-mode (this) = the TikTok-native PHOTO carousel (completion craft + the WoopSocial PHOTO publish config) · tiktok-script = the TikTok VIDEO script (duet/stitch/cover live there) · carousel-writer = cross-platform carousel copy craft · reels-script/youtube-shorts = video · caption-writer/hook-writer = the caption + overlay lines this skill arranges.

Where this connects

Reads first: brand-profile, goals-and-kpis. Copy: hook-writer, caption-writer. Images (external to WoopSocial): image-prompt/ideogram/nano-banana, carousel-writer (adapt down to 9:16). Publish: scheduling-and-queue → WoopSocial (native PHOTO), platform-specs-and-validation (validate + postType/field rules), content-calendar. Reuse/measure: content-recycling (→ IG/ Pinterest), analytics-and-reporting (completion/saves), experimentation-and-ab-testing (A/B the first slide). Images stay external.

Definition of done

A swipeable content type matched to the niche and goal; a slide-1 hook that promises a payoff; a 5–7 slide progressive reveal with cliffhangers, one idea per slide, designed 9:16 1080×1920 with content in the safe zone; a final-slide reward + one CTA and a caption question (not a like-beg); 1–2 hashtags + overlay keywords for TikTok SEO; the complete WoopSocial PHOTO publish config (postType PHOTO, privacyLevel, allowComment, autoAddMusic, truthful isYourBrand/isBrandedContent, postMode) validated before publish; images produced externally, WoopSocial publishing the native PHOTO post with AI auto-disclosure, edits via delete+recreate, nothing fabricated and no promised virality; correctly distinguished from tiktok-script and carousel-writer.

Individual skills in this repo

This repo contains 20 individual skills — each has its own dedicated page.

関連スキル