
video-delivery-specs
This skill should be used when the user asks "what are the specs for Instagram Reels / TikTok / YouTube / broadcast", "export settings for social", "what aspect ratio and bitrate", "how do I repurpose 16:9 to 9:16", "safe area / title-safe for vertical", "pre-delivery checklist", or "collect files for handover". It outputs per-platform spec sheets, a cutdown matrix, captioning/accessibility notes, and a QC + collect-files checklist.
- Installs
- 613
- Last 24h
- +14
- Trending rank
- #7194
- Security
- Not audited
Install
Works with every agent the skills CLI supports.
npx skills add https://github.com/iart-ai/freelance-motion-skills --skill video-delivery-specs- 01
Install
Run the command in your project. The CLI asks which agents to install for.
- 02
Describe the task
Ask your agent for something this skill covers. It loads video-delivery-specs automatically when relevant.
- 03
Review the result
The agent follows the SKILL.md steps, scripts and references bundled with the skill.
SKILL.md
The instructions your agent reads when the skill is triggered.
Video Delivery Specs
Deliver motion content correctly across platforms: produce per-platform spec sheets, repurpose one master into every aspect ratio safely, handle captions and accessibility, and run a pre-delivery QC and collect-files pass before handover.
When to use
Use when exporting or delivering finished motion, when targeting multiple platforms from one master, when checking safe areas for UI-overlay platforms, when deciding caption strategy, or when packaging a project for handover/archive.
Decision tree — master strategy and spec path
Two judgments: what to build as the master so it survives every crop, and which delivery path each platform needs. Decide the master before animating — re-cropping after the fact is where deliveries break.
How many ratios does the brief's distribution need?
├─ One ratio only → animate native to that ratio; no spine needed.
└─ Multiple ratios → build a 16:9 master with a centre 1:1 "title-safe
spine"; expect to RE-LAYOUT (not just crop) for 9:16.
Per target, which path?
├─ Social (Reels/TikTok/Shorts/Story) → 9:16 1080×1920, H.264 8-12 Mbps,
│ burn in captions, keep action out of top/bottom/right UI zones.
├─ Feed → 4:5 (1080×1350) for most screen real estate.
├─ YouTube long-form → 16:9, upload high-bitrate / ProRes, .srt sidecar.
├─ Broadcast → follow the broadcaster's spec PDF EXACTLY: ProRes 422 HQ,
│ region frame rate, R128/A85 loudness, title-safe 90%, textless + slate.
├─ OOH/signage → pixel-exact to the screen, silent, oversize type.
└─ Web hero → MP4+WebM, muted autoplay loop, compress hard, poster frame.
Rule of thumb: if a platform overlays UI (all vertical social), the safe-area map governs layout before the spec governs export. Get the layout wrong and no bitrate fixes it.
Workflow
- List target platforms from the brief's distribution line (Reels, Feed, Story, TikTok, YouTube, Shorts, broadcast, OOH, web).
- Pull each platform's spec (table below; full detail in references/platform-specs.md).
- Plan the master + cutdowns using the repurposing matrix so the 16:9 master survives the crop to 9:16 and 1:1.
- Decide captions/accessibility — burned-in vs uploaded, styling, safe placement.
- Run QC against the checklist, then collect files for handover/archive.
Per-platform spec sheet (essentials)
| Platform | Ratio | Pixels | Duration | Container/codec | Notes |
|---|---|---|---|---|---|
| IG Reels | 9:16 | 1080×1920 | ≤90s (up to 3min) | MP4/MOV, H.264 | UI overlays bottom + right; keep action centre |
| IG Feed | 4:5 or 1:1 | 1080×1350 / 1080×1080 | ≤60s | MP4, H.264 | 4:5 gives most feed real estate |
| IG Story | 9:16 | 1080×1920 | ≤60s/card | MP4, H.264 | Top + bottom ~250px reserved for UI |
| TikTok | 9:16 | 1080×1920 | 15s-10min | MP4/MOV, H.264 | Heavy right-side + bottom UI; caption-safe centre |
| YouTube (horiz) | 16:9 | 1920×1080 (4K 3840×2160) | flexible | MP4, H.264/H.265 | High bitrate; full frame usable |
| YouTube Shorts | 9:16 | 1080×1920 | ≤60s | MP4, H.264 | Bottom UI for title/controls |
| Broadcast (HD) | 16:9 | 1920×1080 | exact to spec | ProRes 422 (HQ) / per broadcaster | Title-safe 90%, action-safe 93%; legal/loudness |
| OOH / digital screen | varies (often 9:16 or custom) | per-site | loop, often 10-15s | MP4/ProRes per vendor | No sound; high legibility; supplied pixel-exact |
| Web (hero/embed) | 16:9 or custom | responsive | short loop | MP4 (H.264) + WebM; poster frame | Compress hard; muted autoplay; provide fallback |
Bitrate guidance (H.264): 1080p social 8-12 Mbps; 1080p high-quality 16-20 Mbps; 4K 35-45 Mbps. Audio AAC 320 kbps stereo, 48 kHz. For broadcast/archive use ProRes 422 HQ, not H.264. Full codec/bitrate detail in references/platform-specs.md.
Safe areas and UI danger zones
Social platforms overlay UI on top of the video. Keep titles and key action out of these zones:
- 9:16 (Reels/TikTok/Shorts/Story): reserve top ~250px (handle/close) and bottom ~250-400px (caption, CTA, controls). On TikTok also reserve the right ~120px column (icons). Keep essential text within the centre ~1080×1080 region.
- 16:9 broadcast: title-safe = inner 90%, action-safe = inner 93%. Nothing critical outside title-safe.
- OOH: follow the vendor's exact pixel map and bleed; assume viewing distance — oversize type.
A visual map of each zone is in references/safe-areas.md.
Cutdown / repurposing matrix (one master → all ratios)
Design the 16:9 master so the centre survives cropping. Plan it before animating, not after.
| From → To | Method | Watch for |
|---|---|---|
| 16:9 → 1:1 | Crop sides to centre 1080 | Text near left/right edges gets cut |
| 16:9 → 9:16 | Crop hard to centre + re-stack elements vertically | Wide compositions break; reflow text/logos |
| 16:9 → 4:5 | Light side-crop | Usually safe if action is centred |
| Master → 6s/15s cutdown | Pick the hook + payoff; drop middle | Re-time, don't just trim; keep brand + CTA |
Best practice: build a "title-safe 1:1 spine" down the centre of the 16:9 master so every crop keeps the message. For 9:16, expect to re-layout rather than crop — budget time for it. Full matrix and a reflow recipe in references/repurposing-matrix.md.
Captions and accessibility
- Open (burned-in) captions: baked into the video. Default for social — most users watch muted, and platform auto-captions are unreliable. Always burn captions for Reels/TikTok/Shorts.
- Closed captions (.srt/.vtt sidecar): uploaded separately; user-toggleable. Required for YouTube long-form and broadcast accessibility compliance; good practice everywhere.
- Styling for burned-in: high contrast (white text, subtle shadow or box), legible weight, 1-2 lines max, placed inside the caption-safe zone (above the bottom UI band, not under it). Sync to speech; ~1-2s per line minimum.
- Beyond captions: avoid conveying meaning by colour alone; keep flashing under 3 flashes/sec (photosensitivity); provide a descriptive title/transcript for screen readers where the platform allows.
Pre-delivery QC checklist (top items)
- Correct dimensions, ratio, and duration for each target.
- First and last frames clean (no black flash, no held render frame).
- Audio present, levels sane, no clipping; broadcast: correct loudness target.
- All text spell-checked; legal/super present and held long enough.
- Logo lockup correct; brand colours match; no placeholder assets.
- Safe areas respected; nothing critical under platform UI.
- Captions accurate and inside the caption-safe zone.
- Export matches codec/bitrate spec; file plays on a fresh machine.
- File names follow convention; versions not mixed up.
Full QC + collect-files checklist in references/qc-checklist.md.
Worked examples
GOOD — one master planned for every crop:
Brief needs 16:9 (web), 1:1 (feed), and 9:16 (Reels + TikTok). You animate a 16:9 master keeping all titles and the logo inside a centre 1:1 spine, then crop to 1:1 cleanly and re-layout (not crop) to 9:16 — restacking the headline above the product, captions burned in above the bottom UI band, action clear of TikTok's right column. Each export matches its platform's bitrate; captions sit inside the caption-safe zone.
ANTI-PATTERN — "we'll just crop it later":
A wide 16:9 hero with the logo bottom-left and a tagline spanning the full width gets center-cropped to 9:16 for Reels. The logo is gone, half the tagline is cut, and the burned-in captions land under the platform's CTA bar where no one can read them.
Why it fails: the master wasn't designed for the crop, and the safe-area map was ignored. The fix is the master strategy above — centre spine, re-layout for vertical, captions above the UI band — decided before animation, not after delivery is due.
Common mistakes
| Symptom | Why it happens | Fix |
|---|---|---|
| Text/logo cut off after crop | Composed wide, no centre spine | Build a centre 1:1 title-safe spine; re-layout for 9:16 |
| Captions hidden under platform UI | Placed at frame bottom, not caption-safe zone | Keep captions above the bottom UI band, inside caption-safe |
| Broadcast delivery rejected | Used H.264 / wrong loudness / no textless | Follow the broadcaster spec PDF: ProRes, R128/A85, slate, textless |
| Video looks soft after upload | Exported at low bitrate; platform re-compresses | Upload high-bitrate / ProRes; let the platform downscale |
| Black flash at start/end | Held render frame or stray black frame | QC first/last frames; trim to intended frames |
| Colour shifts outside the editor | Gamma/colour-management mismatch | Spot-check the export in a neutral player before handover |
| Wrong file delivered | Versions mixed in the folder | Enforce the naming convention; flag CAPTIONED/TEXTLESS/BROADCAST |
| Autoplay web hero won't play | Has audio / too large / no fallback | Muted MP4+WebM, compress hard, poster frame |
Deliverable spec — what a good handover contains
The output is a packaged, dated delivery folder plus the QC pass behind it (full structure in references/qc-checklist.md). A good handover:
- Has every deliverable at the correct ratio, pixels, duration, codec, and bitrate for its target.
- Separates final exports, captioned versions (+ .srt/.vtt), textless masters, and stills.
- Includes a README listing deliverables, specs, and every font/music/stock licence with its usage scope.
- Uses the file-naming convention consistently, with the final version clearly flagged.
- Has passed the full pre-delivery QC checklist before it leaves.
Before you finish — checklist
- Each target's dimensions, ratio, duration, codec, and bitrate verified.
- First/last frames clean; no dropped frames or artifacts.
- Audio correct (or correctly absent); broadcast loudness met.
- Text spell-checked; legal/super held long enough; logo lockup correct.
- Safe areas respected; nothing critical under platform UI.
- Captions accurate, synced, inside the caption-safe zone.
- Plays on a fresh machine; colour correct outside the editor.
- File names follow the convention; final version flagged; old versions removed.
- Handover folder structured; README + licence/rights notes included.
Quick reference
| Need | Answer |
|---|---|
| Reels/TikTok/Shorts ratio | 9:16, 1080×1920 |
| Feed best ratio | 4:5 (1080×1350) |
| Social bitrate (1080p) | 8-12 Mbps H.264 |
| Broadcast codec | ProRes 422 HQ, not H.264 |
| Captions for social | Burn them in (open) |
| TikTok danger zones | Bottom + right column |
| Broadcast title-safe | Inner 90% |
| One master strategy | Centre 1:1 spine in the 16:9 |
Reference files
references/platform-specs.md— full per-platform spec sheet: exact pixels, durations, codecs, bitrates, audio, and file-size limits.references/safe-areas.md— ASCII safe-area maps per ratio with pixel measurements for UI danger zones.references/repurposing-matrix.md— full cutdown matrix and a step-by-step 16:9→9:16 reflow recipe.references/qc-checklist.md— complete pre-delivery QC checklist plus the Collect-Files / handover packaging list and naming convention.
FAQ
What does the video-delivery-specs skill do?
This skill should be used when the user asks "what are the specs for Instagram Reels / TikTok / YouTube / broadcast", "export settings for social", "what aspect ratio and bitrate", "how do I repurpose 16:9 to 9:16", "safe area / title-safe for vertical", "pre-delivery checklist", or "collect files for handover". It outputs per-platform spec sheets, a cutdown matrix, captioning/accessibility notes, and a QC + collect-files checklist.
How do I install video-delivery-specs?
Run npx skills add https://github.com/iart-ai/freelance-motion-skills --skill video-delivery-specs in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/video-delivery-specs/ (personal) or .claude/skills/video-delivery-specs/ (project).
Which AI agents can use video-delivery-specs?
Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.
Is video-delivery-specs safe to install?
No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.
Who maintains video-delivery-specs?
It is published by iart-ai in iart-ai/freelance-motion-skills. It has 613 recorded installs.
More from iart-ai/freelance-motion-skills
brand-motion-guidelines
iart-ai/freelance-motion-skills
This skill should be used when the user asks to "write motion guidelines", "create a brand motion system", "define easing and timing tokens", "document our animation principles", "build a motion language like Slack/IBM/Klarna", "set up entrance/exit/transition rules", or "add a reduced-motion / accessibility section". It outputs a fill-in-ready brand motion guidelines document.
creative-brief
iart-ai/freelance-motion-skills
This skill should be used when the user asks to "write a creative brief", "build a motion brief", "the client doesn't know what they want", "turn this vague client request into a brief", "what questions should I ask the client", or "make a brief I can get sign-off on". It turns a fuzzy client ask into a structured, sign-off-ready motion design brief and the exact clarifying questions to send back.
client-revisions
iart-ai/freelance-motion-skills
This skill should be used when the user asks to "stop scope creep", "limit revision rounds", "write SOW revision language", "the client keeps asking for changes", "how do I say this is out of scope", "translate vague feedback", "client said make it pop", or "write a change-order email". It provides revision-cap contract language, a consolidated-feedback request, a vague-feedback translator, and a polite out-of-scope email.
motion-pricing
iart-ai/freelance-motion-skills
This skill should be used when the user asks to "quote a motion project", "estimate this animation job", "how much should I charge for this video", "what's my day rate", "build a quote", "price a 30-second explainer", or "is this estimate fair". It explains motion-design pricing models, the day-rate math, the cost drivers, and outputs a line-itemed quote with a worked example.
Popular media & creative skills
hyperframes
heygen-com/hyperframes
Read first to make, edit, animate or render any video with HyperFrames (HTML rendered to video), including launch films, explainers, music videos, charts, maps, vertical clips, title cards, captions and overlays. Holds the render contract, the work loop and what each kind of video needs.
media-use
heygen-com/hyperframes
Get or make any media for a HyperFrames video - music, sound effects, images, icons, logos, voiceover, avatar and generated video, transcripts, cutouts, colour grades and looks - with HeyGen or free local tools, and mix placed audio. Also for footage that looks dark, flat or dull, or should be stylized or hidden.
hyperframes-registry
heygen-com/hyperframes
About 450 ready-made HyperFrames blocks and components installed as editable source - transitions, overlays, titles, kinetic type, captions, charts, map overlays, 3D, camera moves. Use it before hand-building any named visual (CRT scanlines, glitch, film grain, a chart); also for hyperframes add and catalog.
hyperframes-creative
heygen-com/hyperframes
Non-animation creative direction for HyperFrames videos. Use for design spec (frame.md / design.md) handling, palettes, typography, narration, beat planning, audio-reactive visuals, composition patterns, and brand / style decisions. For atomic motion patterns and scene blueprints, use hyperframes-animation.
ai-video-generation
101-skills/superpowers
Generate AI videos with Google Veo, Seedance 2.0, HappyHorse, Wan, Grok and 40+ models via inference.sh CLI. Models: Veo 3.1, Veo 3, Seedance 2.0, HappyHorse 1.0, Wan 2.5, Grok Imagine Video, OmniHuman, Fabric, HunyuanVideo. Capabilities: text-to-video, image-to-video, reference-to-video, video editing, lipsync, avatar animation, video upscaling, foley sound. Use for: social media videos, marketing content, explainer videos, product demos, AI avatars. Triggers: video generation, ai video, text to video, image to video, veo, animate image, video from image, ai animation, video generator, generate video, t2v, i2v, ai video maker, create video with ai, runway alternative, pika alternative, sora alternative, kling alternative, seedance, happyhorse
ai-image-generation
101-skills/superpowers
Generate AI images with GPT-Image-2, FLUX, Gemini, Grok, Seedream, Reve and 50+ models via inference.sh CLI. Models: GPT-Image-2, FLUX Dev LoRA, FLUX.2 Klein LoRA, Gemini 3 Pro Image, Grok Imagine, Seedream 4.5, Reve, ImagineArt. Capabilities: text-to-image, image-to-image, inpainting, LoRA, image editing, upscaling, text rendering. Use for: AI art, product mockups, concept art, social media graphics, marketing visuals, illustrations. Triggers: flux, image generation, ai image, text to image, stable diffusion, generate image, ai art, midjourney alternative, dall-e alternative, text2img, t2i, image generator, ai picture, create image with ai, generative ai, ai illustration, grok image, gemini image, gpt image, openai image, chatgpt image