make-a-video
Skill136 starsChanged 17 days agoMITView on GitHub
Beginner-friendly end-to-end video creator for HyperFrames. Use when the user says "make a video", "create a video", "new video", "build a video", "video from scratch", "I want to make a video", "help me create a video", or when someone who's never used HyperFrames before arrives with a concept, script, or rough idea and wants a finished MP4. Interviews the user in one pass, then builds the full video with mandatory preview and visual-verification gates.
11 other projects use a version of this, 7 of them edited.
Adopted by
Most recently taken up by
- AI-Biz-Guy/cast-studio-plugintheir own version, first seen Nov 1, 2026
- F0rger123/claude-video-editing-god-modeedited, first seen Oct 25, 2026
- NexumAI-online/skills-ediciontheir own version, first seen Oct 25, 2026
- tony-hoang-ai-automation/mkt-skillsedited, first seen Oct 25, 2026
- nateherkai/herk-edit-kitedited, first seen Oct 15, 2026
- nateherkai/hyperframes-student-kitedited, first seen Oct 14, 2026
Used with
Projects with it also use short-form-edit, cut-mistakes and 6 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
NexumAI-onlineSkillshort-form-edit
Convertí una grabación hablada en un reel, YouTube Short o anuncio corto terminado, con apertura que genere curiosidad, resolución respaldada por el contenido, cortes narrativos, gráficos sincronizados, B-roll en movimiento, subtítulos, música sutil y efectos. Para videos largos usá edit-video; para quitar sólo pausas, cut-silences.
Also in 4 projects that use this
nateherkaiSkillcut-mistakes
Agent 2 of the video editing pipeline. Finds and removes spoken mistakes — stutters, repeated words, false starts, and retakes (re-recorded lines) — from a talking-head recording. Use after cut-silences, when asked to cut mistakes, remove stutters / repeats / filler restarts, clean up flubs, or keep the best take. Works review-gated: it proposes every cut with context and a reason for approval, then renders only the approved cuts via ffmpeg. Requires a word-level transcript.
Also in 4 projects that use this
nateherkaiSkillcut-silences
Agent 1 of the video editing pipeline. Removes silences and dead air from a talking-head recording. Use when asked to cut silences, trim pauses, remove dead air / gaps, or tighten the pacing of a raw video, given a word-level transcript. Produces an edit list (EDL), a re-timed transcript for downstream agents, and optionally the cut video via ffmpeg. Does NOT cut mistakes, repeats, or false starts — that is the cut-mistakes agent.
Also in 4 projects that use this
Google CloudSkillvideo-storytelling
Cinematic video ad storytelling expertise. Covers 3-act narrative structure, emotional arcs, camera motion, scene transitions, and voiceover pacing for 10-second commercial video ads using Gemini Omni. Load this skill when generating video ads or storylines.
Also in 4 projects that use this
ConardLiSkillmotion-showreel
用代码制作图形、文字或界面动画视频,输出 MP4 和可编辑源码,最长 20 分钟。适用于动效作品集、产品片、UI 动效、动态排版、Logo 片头、生成艺术与讲解视频;支持独立选择 BGM/音效来源、用户口播,以及用用户提供的口播稿通过 TTS 合成语音。不用于实拍制作、写实人物生成或一般视频剪辑。
Also in 4 projects that use this
GoldLegendW80Skilledit-video
Edit one chapter or section of an already-generated video without redoing the whole thing. Change the script, visuals, captions, or timing of a single scene, then re-render and re-verify - iterate on AI-generated videos, social clips, promos and explainers. Companion to make-video. Use when asked to change, fix, redo or tweak a section, scene or chapter of an existing video. Trigger - /edit-video <project-id> <chapter-id> "<instruction>".
Also in 4 projects that use this
- +2 more from nateherkai
Other versions11 projects use a version of this.
Word for word · 1
- notivn/AIEV★ 132100% same
Edited · 7
- nateherkai/hyperframes-student-kit★ 2.2k97% same
- nateherkai/herk-edit-kit★ 1.5k97% same
- coltonjosephdean-rgb/Hyperframes-colton.ai.dean★ 1999% same
- assafkip/claude-video-editor★ 1799% same
- tony-hoang-ai-automation/mkt-skills★ 499% same
- F0rger123/claude-video-editing-god-mode★ 197% same
- chriswestt/claude-skills★ 099% same
Wrote their own version · 3
- hubeiqiao/how-i-make-videos-with-ai★ 4same name
- NexumAI-online/skills-edicion★ 4same name
- AI-Biz-Guy/cast-studio-plugin★ 0same 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 repository20 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- ai-illustrationsSkill
- auto-cutSkill
- background-musicSkill
- color-gradingSkill
- gsapSkill
- hyperframes-cliSkill
- hyperframes-registrySkill
- hyperframesSkill
- key-layoutSkill
- noti-tiktok-full-textSkill
- noti-tiktok-vnSkill
- noti-youtube-editSkill
- offline-voiceSkill
- remotion-assembleSkill
- short-form-videoSkill
- skill-authoringSkill
- video-pipelineSkill
- website-to-hyperframesSkill
- webui-designSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
20 rules. 5 are common, 1 is distinctive, and 14 are very specific to this project.
5
Common
Most of the 10 edited versions we read keep these.
1
Distinctive
This author's own choices. Worth a look.
14
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 22
Routing note (AI Edit Video pipeline): when this skill runs from a project brief, the edit prompt's STYLE DESIGN and PHONG CÁCH DỰNG (video …
- Very specific: line 35
Ask one question at a time via AskUserQuestion, multiple-choice where possible.
- Very specific: line 44
Gate: all six captured. If the answer is 9:16 + face-cam, plan to hand off to /short-form-video at Gate 5.
- Very specific: line 53
If TTS: voice preference. Offer choices from npx hyperframes tts --help.
- Very specific: line 54
If face-cam: recording path · full-screen or corner placement · need transcription?
- Very specific: line 63
Before asking the user anything, inventory existing assets: Check <workspace-root>/assets/ and any project assets/ folder.
- Very specific: line 86
Read
MOTION_PHILOSOPHY.md: if it exists in the workspace root - mandatory if present. - Common: line 87
Ask where projects live: if it's not obvious:
- Very specific: line 88
If video-projects/ exists → use video-projects/<slug>/
- Common: line 98
Show the brief. WAIT for explicit approval: Don't proceed to Gate 5 without a clear "yes, build it."
- Very specific: line 106
If the brief describes a 9:16 vertical talking-head with face-cam + scene overlays, invoke /short-form-video NOW and hand off the brief.
- Very specific: line 113
If a sibling project with similar format exists, offer to copy its hyperframes.json + meta.json as a template.
- Very specific: line 116
Create <project-folder>/assets/style-profile.md from Gate 3 - single source of truth for palette/fonts/logo.
- Very specific: line 189
Wait for "Studio running" on http://localhost:3002.
- Common: line 191
WAIT for explicit "looks good, render a draft": before proceeding. Silence is not approval.
- Common: line 216
If anything's wrong: fix → re-render → re-verify.
- Very specific: line 224
Do NOT use Python's http.server - it doesn't support HTTP Range requests, so scrubbing breaks.
- Very specific: line 244
DO NOT build anywhere but inside a dedicated project folder: Never put index.html at the workspace root.
- Common: line 245
DO NOT ask the user for assets: before inventorying their workspace.
- Very specific: line 249
DO NOT impose a brand on the user: Ask first; fall back to MOTION_PHILOSOPHY defaults only when they explicitly decline.
“Common” is measured: kept by at least 6 in 10 of the 10 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.
Files
Make a Video - The Beginner-to-Finished-MP4 Skill
Two phases, eight sequential gates. Every gate produces a concrete artifact the next gate consumes. Don't skip gates.
When to use this skill - and when to hand off
Use this skill when:
- The user is new to HyperFrames and starting from a concept, script, or outline
- They want an end-to-end walkthrough, not framework reference material
- They haven't decided on format yet
Hand off when:
- The user pastes a URL and wants a video from that site → invoke
/website-to-hyperframes - The user explicitly wants a 9:16 vertical talking-head with face-cam + scene overlays → run Gates 1–4 here, then invoke
/short-form-videofrom Gate 5 onward - The user asks for framework rules, not a video → invoke
/hyperframes
Routing note (AI Edit Video pipeline): when this skill runs from a project brief, the edit prompt's STYLE DESIGN and PHONG CÁCH DỰNG (video style) sections override the interview defaults below - do not re-ask style questions the brief already answers.
The two phases
- Phase 1 - INTERVIEW (Gates 1–4): one conversational pass to gather everything before touching code. Intent, format, script, voice, style, assets, pacing. Synthesize into a
BRIEF.mdand wait for explicit approval. - Phase 2 - BUILD (Gates 5–8): scaffold → storyboard → compositions → lint → Studio preview → draft render → visual verification → MP4 preview → final render.
Gate 1 · Intent & format
Ask one question at a time via AskUserQuestion, multiple-choice where possible.
- What's this video for? (promo · social ad · launch teaser · product demo · tutorial · explainer · intro/outro card · other)
- Who's the audience? (open-ended)
- Target duration? (10–20s short · 20–45s promo · 45–90s explainer · 1.5–3 min lesson · custom)
- Aspect ratio? (16:9 1920×1080 · 9:16 1080×1920 · 1:1 1080×1080)
- Frame rate? (30 default · 60 for crisp UI · 24 cinematic)
- Platform / delivery constraints? (file size · deadline · where it'll play)
Gate: all six captured. If the answer is 9:16 + face-cam, plan to hand off to /short-form-video at Gate 5.
Full question bank: Read: references/interview-questions.md
Gate 2 · Script & voice
- Script source? (paste · outline → I'll draft · I'll record · TTS from text · no narration)
- If TTS: voice preference. Offer choices from
npx hyperframes tts --help. Also capture pace.hyperframes tts(Kokoro) is English-only - Vietnamese narration goes through the system's/api/tts(Gemini TTS or VieNeu) instead. - If face-cam: recording path · full-screen or corner placement · need transcription? (
npx hyperframes transcribe <file> --model small.en --json- English only; for Vietnamese use faster-whisper per the noti-tiktok-vn skill) - Captions? (off · hype · corporate · karaoke-word-by-word · minimal)
Gate: script captured (or drafted), audio plan captured, caption plan captured.
Gate 3 · Style intake
Before asking the user anything, inventory existing assets. Check <workspace-root>/assets/ and any project assets/ folder. Don't ask for what's already there.
Then ask progressively - they don't need answers to all of these:
- Style guide or brand doc? (paste/path · no)
- Color palette? (hex codes · none - use MOTION_PHILOSOPHY defaults)
- Fonts? (Google Fonts name(s) · file paths · none - use Inter + JetBrains Mono defaults)
- Logo file? (path · none). If a brand is named but no file is given: look in
assets/brand-logos/, or fetch it withPOST /api/brand-logos {"name":"..."}- a text wordmark is allowed ONLY after a404 BRAND_LOGO_NOT_FOUND. Never draw or generate a logo. - Reference videos for vibe? (URLs/paths · none)
- Other assets? (screenshots · product photos · b-roll · music - list paths)
- MOTION_PHILOSOPHY aesthetic (black canvas · chrome type · perspective grid · whip transitions) or a different feel?
- Pacing? (kinetic 1–2s · balanced 2–3s · relaxed 3–5s)
- Music? (none · ambient pad 0.15 · music bed 0.4 · full 0.8 - file path if they have one)
- Outro / call-to-action text?
Never impose a brand on the user. Fall back to MOTION_PHILOSOPHY defaults only when they explicitly decline to supply a style.
Full style flow + MOTION_PHILOSOPHY defaults: Read: references/style-intake.md
Gate 4 · Brief synthesis (HARD-GATE before building)
- Read
MOTION_PHILOSOPHY.mdif it exists in the workspace root - mandatory if present. If missing, proceed with the defaults inreferences/style-intake.mdand note the absence in the brief. - Ask where projects live if it's not obvious:
- If
video-projects/exists → usevideo-projects/<slug>/ - Otherwise → ask the user
- If
- Write
<project-folder>/BRIEF.md:- slug · intent · audience · dimensions · fps · duration
- script (full or outline)
- voice choice · caption plan · face-cam plan
- style profile: palette (hex), fonts, logo path, reference videos
- pacing
- asset inventory with paths
- outro / CTA text
- Show the brief. WAIT for explicit approval. Don't proceed to Gate 5 without a clear "yes, build it."
Gate 5 · Scaffold & storyboard
Handoff check first
If the brief describes a 9:16 vertical talking-head with face-cam + scene overlays, invoke /short-form-video NOW and hand off the brief. Its 4-layer scaffold is purpose-built for that format.
Otherwise continue:
Scaffold
mkdir <project-folder>- If a sibling project with similar format exists, offer to copy its
hyperframes.json+meta.jsonas a template. Otherwise from inside the folder:npx hyperframes init - Edit
meta.jsonwith the user's slug, dimensions, fps. - Copy supplied assets into
<project-folder>/assets/. - Create
<project-folder>/assets/style-profile.mdfrom Gate 3 - single source of truth for palette/fonts/logo.
Storyboard
Generate <project-folder>/STORYBOARD.md using the template in references/storyboard-template.md. Every beat gets:
Beat N - TITLE (start–end, duration) - Concept in one sentence
Visual elements: [each element, size, animation, timing]
Motion language: [kind of motion]
Eases used: [3–4 distinct GSAP eases]
Exit: [transition into next beat]
Audio: [VO line / SFX / music layer]Top of file: a timing table with scene · start · duration · composition file.
Propose a rule-of-threes structure:
- Act 1 (hook) ≈ 20% of duration
- Act 2 (body) ≈ 55%
- Act 3 (payoff + outro with 4–6 second hold) ≈ 25%
(MOTION_PHILOSOPHY §0 Law 9, §1.)
Map user intents → catalog blocks: Read: references/catalog-intent-map.md
Gate: show storyboard + timing table. Iterate until the user approves.
Gate 6 · Build compositions
Invoke /hyperframes for framework rules. This skill owns the scaffold and discipline; /hyperframes enforces the rules.
Scaffold every sub-composition
<div data-composition-id="scene-name" data-start="..." data-duration="...">
<style>[data-composition-id="scene-name"] { /* scoped */ }</style>
<!-- DOM -->
<script>
(function(){
const SLOT_DURATION = ...;
const tl = gsap.timeline({ paused: true });
// ... tweens ...
tl.to({}, { duration: SLOT_DURATION }, 0); // anchor - MOTION_PHILOSOPHY Law 11
window.__timelines["scene-name"] = tl;
})();
</script>
</div>Full boilerplate + captions pattern + ambient-bg pattern: Read: references/composition-scaffold.md
Build rules
- Ambient background on
data-track-index="0"for the full composition duration. - Kinetic-type openers: per-word stagger 0.06–0.10s.
- Captions as body-level siblings of the root composition in
index.html, each withdata-track-index ≥ 20. Never inside scene timelines (MOTION_PHILOSOPHY §3.13). - Catalog blocks installed via
npx hyperframes add <name>. Immediately scope the block's CSS to[data-composition-id="..."]- catalog blocks ship withhtml, body { ... }rules that bleed into the parent document when loaded as sub-compositions. - Vertical + face-cam: wrap native 1920×1080 face in a transform (
translate+scale) for bottom-half or full-screen mode. (If you end up here instead of/short-form-video, strongly consider the handoff.) - Apply ONLY what the user supplied. Their palette, their fonts, their logo. Don't inject anything else. If they chose MOTION_PHILOSOPHY defaults, pull the palette + font pair from
references/style-intake.md.
Determinism
No Math.random(), no Date.now(), no render-time fetch(). Use seeded PRNGs or harmonic-sin hashes (MOTION_PHILOSOPHY §3.10).
Gate 7 · Lint → Studio preview (PREVIEW GATE 1 - MANDATORY)
npx hyperframes lint- fix all errors, triage warnings.npx hyperframes previewin the background.- Wait for "Studio running" on
http://localhost:3002. - Hand the user the URL plus individual composition URLs (
http://localhost:3002/?comp=<id>). If the project has WebGL shader blocks, lead with the individual URLs - software WebGL fallback can stall the master composition. - WAIT for explicit "looks good, render a draft" before proceeding. Silence is not approval.
Hot reload is on - edits show up live.
Gate 8 · Draft render → visual verification → MP4 preview → final
Draft render
npx hyperframes render --quality draft --output renders/<slug>-draft.mp4Visual verification (MANDATORY before delivery)
Lint passing ≠ design working. Extract frames and view them.
mkdir -p renders/frames- For every beat hero moment AND every transition:
ffmpeg -y -ss <t> -i renders/<slug>-draft.mp4 -frames:v 1 -q:v 2 renders/frames/t<t>.png - Call the
Readtool on every PNG. The Read tool loads the image into context - don't just list filenames. - Confirm per frame: no cropped faces, correct face-mode per scene, text readable and on-palette, no overflow, transitions land on intended words, no blank frames.
- If anything's wrong: fix → re-render → re-verify. Never ship a broken draft.
MP4 preview (PREVIEW GATE 2 - MANDATORY)
npx serve renders -p 8080 -nDo NOT use Python's http.server - it doesn't support HTTP Range requests, so scrubbing breaks.
Hand the user http://localhost:8080/<slug>-draft.mp4. WAIT for explicit sign-off on playback and audio sync.
Final render
npx hyperframes render --quality standard --output renders/<slug>-final.mp4In the AI Edit Video pipeline this is NOT the end: continue with the video-pipeline skill (QC → assemble final via the render queue → outputs/<id>-v<N>.mp4 → update meta.json → thumbnail/publish). Standalone use outside the pipeline: report the output path. Done.
Full preflight + pre-delivery checklist: Read: references/build-checklist.md
Non-negotiables (load-bearing - do not soften)
- DO NOT skip PREVIEW GATE 1 (Studio) or PREVIEW GATE 2 (rendered MP4). Two gates per build, always.
- DO NOT claim a render is done until frames have been extracted AND Read via the Read tool.
- DO NOT build anywhere but inside a dedicated project folder. Never put
index.htmlat the workspace root. - DO NOT ask the user for assets before inventorying their workspace.
- DO NOT skip the
tl.to({}, { duration: SLOT_DURATION }, 0)anchor tween at the end of every sub-composition timeline. MOTION_PHILOSOPHY Law 11. - DO NOT use
Math.random()/Date.now()inside render logic. Seeded hashes only. - DO NOT add
class="clip"to<video>tags. It breaks them. - DO NOT impose a brand on the user. Ask first; fall back to MOTION_PHILOSOPHY defaults only when they explicitly decline.
References
references/interview-questions.md- full question bank by Gatereferences/style-intake.md- style interview + MOTION_PHILOSOPHY defaultsreferences/catalog-intent-map.md- "user says X → install Y"references/storyboard-template.md- beat-by-beat template + worked examplereferences/composition-scaffold.md- scoped-styles + IIFE GSAP boilerplatereferences/build-checklist.md- preflight + pre-delivery gates
External (workspace-level):
MOTION_PHILOSOPHY.md- the one external reference this skill assumes exists. Aesthetic baseline. Fallback lives inreferences/style-intake.mdif the file is missing.
Related skills
/hyperframes- framework rules, invoke at Gate 6/hyperframes-cli- init · lint · preview · render · transcribe · tts/hyperframes-registry- installing catalog blocks/gsap- GSAP animation reference/short-form-video- hand off at Gate 5 for 9:16 talking-head format/website-to-hyperframes- hand off at Gate 1 if the starting input is a URL
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
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.

