All skills
anthropics avatar

/quickdesign

@710a472

Use the `quickdesign` CLI to generate AI media — UGC promo videos, image edits, product creatives, video upscales — through Seedance, Kling, Sora2, Nano Banana, and GPT Image. Invoke this skill whenever the user asks for a talking-avatar video, multi-segment ad / promo / explainer, image edit (object swap, angle change, state change), product photoshoot, or video upscale via QuickDesign.

Use this Skill: https://skilld.dev/gh/anthropics/claude-plugins-community/quickdesign

This session only. Nothing lands on disk.

referencesauto-subtitle.md

≈1.5k tokens on demand. Your agent reads this file only when SKILL.md points to it.

When to use

Add subtitles in post-production whenever the user wants:

  • Burned-in captions on a UGC reel / TikTok / Reels content
  • Karaoke-style word-by-word highlights
  • Multi-language transcription with custom font / color styling
  • Text overlay that perfectly matches the spoken audio

NEVER let Seedance burn captions via the prompt. Seedance's prompt-derived subtitles hallucinate (partial sentences, paraphrased words, wrong timing). The dedicated subtitle endpoint uses real ASR on the generated audio — bulletproof accuracy.

CLI usage

# Default: TikTok-style, English, single-word karaoke highlight
quickdesign video subtitle ./final.mp4 -o ./final-subbed.mp4 --wait

# Different style + language
quickdesign video subtitle ./final.mp4 \
  --style minimal \
  --language tr \
  -o ./final-tr-subbed.mp4 --wait

# Style override (preset + custom flags)
quickdesign video subtitle ./final.mp4 \
  --style tiktok \
  --font Poppins \
  --highlight-color cyan \
  --words-per-line 2 \
  -o ./final-custom.mp4 --wait

Style presets

The --style <preset> flag selects a baseline; any explicit --<flag> overrides individual fields.

Preset Font Size Weight Color Highlight Stroke Pos Words/line Anim Use case
default (default) Montserrat 65 bold white yellow 2 black bottom +60 5 on Balanced, readable, works for most content
tiktok Montserrat 100 bold white purple 3 black bottom +75 1 on UGC creator content, vertical reels, single-word pop
minimal Inter 60 normal white white (none) 1 black bottom +50 8 off Editorial, brand ad, professional
karaoke Montserrat 80 bold white yellow 2 black center 0 3 on Music / lyric content
reels-pop Bebas Neue 110 black yellow red 4 black center -50 1 on Punchy hook content, attention-grab

Override flags (apply on top of preset)

--language <code>          en | tr | es | fr | de | ... (or 3-letter ISO)
--font <name>              Any Google Font name (Montserrat, Inter, Poppins, Bebas Neue, etc.)
--font-size <n>            px
--font-weight <w>          normal | bold | black
--color <c>                white | black | red | green | blue | yellow | orange | purple | pink | brown | gray | cyan | magenta
--highlight-color <c>      same color enum
--stroke-width <n>         px (0 = no outline)
--stroke-color <c>         outline color
--background-color <c>     same enum + 'none' / 'transparent'
--background-opacity <n>   0.0–1.0
--position <p>             top | center | bottom
--y-offset <n>             vertical px (positive = down)
--words-per-line <n>       1 = single-word (TikTok), 3 = phrase, 8-12 = full sentence
--no-animation             Disable bounce-in animation

Cost

$0.03 per minute of source video on fal.ai → 1 credit per second at our standard rate (60cr/min, ~50% margin). Minimum 20cr per call to cover infra overhead.

Source duration Cost (cr)
8s 20 (min)
15s 20 (min)
30s 30
44s 44
60s 60
90s 90

How to apply

  1. Default flow: generate the video clean (no Seedance subtitles per no-music-no-subtitles.md rule), then run quickdesign video subtitle on the final concat'd output.

  2. Multi-segment UGC pipelines: subtitle the FINAL concat'd video, NOT each individual segment. This keeps timing continuous — no caption restart at segment cuts.

  3. Style picking:

    • User says nothing specific about style → omit --style (uses default: 5 words/line, font 65, bottom-positioned, yellow highlight). Good baseline for talking-head, narration, mixed content.
    • User says "TikTok-style" / "creator selfie" / "single word pop" → --style tiktok.
    • User says "clean", "professional", "minimal", "editorial" → --style minimal.
    • User says "music video", "lyric video", "karaoke" → --style karaoke.
    • User says "punchy", "attention grab", "hook style", "viral" → --style reels-pop.
    • User specifies custom font / color / size → apply preset + override flags.
  4. Language: always specify --language for non-English content. Default is en. Common: tr, es, fr, de, it, pt, nl, ja, zh, ko. Any 2-letter or 3-letter ISO code accepted.

  5. Duration: auto-detected via ffprobe for local files. For remote URLs, pass --duration-seconds <n> explicitly.

  6. Plan summary must mention subtitle as a separate cost line:

    "Plan: 3-segment UGC video (Seedance R2V) → 1500cr + auto-subtitle on final 36s concat → 36cr. Total: 1536cr. OK?"

What NOT to do

  • ❌ Don't let Seedance burn its own subtitles via the prompt. Always include the No subtitles, no captions, no on-screen text overlays line in every Seedance prompt (per no-music-no-subtitles.md).
  • ❌ Don't subtitle each multi-segment piece individually — subtitle the final concat for continuous timing.
  • ❌ Don't set --words-per-line 1 for minimal style (defeats the purpose; that's a tiktok style choice).
  • ❌ Don't pass a non-Google-Font name to --font — fal pulls from fonts.google.com only. If the user wants a custom font, fall back to a similar Google Font.

Pipeline integration (future)

For a one-command full UGC ad with captions, the --auto-subtitle <style> flag on video generate will chain auto-subtitle automatically after the final concat. Use --no-auto-subtitle to opt out if needed. (Status: planned — not yet shipped at time of writing.)

Source: SKILL.md on GitHub

2 warnings2mo3 checks · Risk SAFE
  • Gen Agent Trust Hub2mo

    This skill provides detailed instructions for using the `quickdesign` CLI to create and edit AI-generated media. It follows secure practices by implementing confirmation gates for paid tasks and provides guidance on interacting safely with downstream AI models. No security issues were detected.

  • Socket2mo

    1 alert: gptSecurity

  • Snyk2mo

    Risk: MEDIUM · 1 issue

Signed by skilld at 710a472. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub last month.

Steadyupdated 3 months ago

README badge

README badge for anthropics/claude-plugins-community/quickdesign