make-video
/make-video
Produkuje wideo recytacji Koranu, krótkie lub długie, dla podanego zakresu wersetów, z wybranym lektorem, stylem i formatem przez potok renderowania.
- Autor
- msrshahrukh100 ↗
- Repozytorium
- quran-content-creator-claude-skill ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 06.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 1
Brak sygnałów. Repozytorium zmienione 06.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Produkuje wideo recytacji Koranu, krótkie lub długie, dla podanego zakresu wersetów, z wybranym lektorem, stylem i formatem przez potok renderowania.
Oryginalny opis (EN)
Produce a Quran recitation short/reel or long-form video for a given surah:ayah range, reciter, style and format with this repo's pipeline (packet → audio → background → compose → validate → render → outbox). Use when asked to make, render, re-render or preview a video for a Quranic reference.
Jak działa
- The non-negotiable rule
- First run
- Religious guardrails
- 1. Write an episode spec (references only)
- Scene plans (for minimalistic-graphics)
- 2. Run
- 3. Outputs
- 4. When validation fails
Podgląd SKILL.md
--- name: make-video description: Produce a Quran recitation short/reel or long-form video for a given surah:ayah range, reciter, style and format with this repo's pipeline (packet → audio → background → compose → validate → render → outbox). Use when asked to make, render, re-render or preview a video for a Quranic reference. --- # make-video ## The non-negotiable rule **Never type, paste, paraphrase, "fix", transliterate or hardcode Quranic text** — Arabic, translations, word glosses or surah names — anywhere: episode YAML, templates, code, tests, captions, metadata, commit messages, or your replies used as content. All of it comes from the Quran.com API via `src/quran`, frozen in a `VersePacket` (sha256), and is injected by code. Episode specs contain **references only**. `src/validate` blocks any render whose on-screen text differs from the packet, or that shows Arabic-script text outside a `data-slot` element. Never bypass or weaken it to "get a render out"; if it fails, fix the template/compose code, not the check. ## First run If `assets/fonts/UthmanicHafs.woff2` or `node_modules` is missing, run `npm install && npm run setup`. Needs ffmpeg. ## Religious guardrails Never depict Allah, prophets, messengers or angels in any form. No music, no comedy. Keep the Arabic readable. ## 1. Write an episode spec (references only) `episodes/<id>.yaml`: ```yaml id: baqarah-285-286 # slug; output goes to work/<id>/ and outbox/<id>/ surah: 2 # 1–114 ayahFrom: 285 ayahTo: 286 # ayahFrom ≤ ayahTo reciterId: 7 # default 7 (Mishary Alafasy, chapter catalog) reciterSource: chapter # chapter (full-surah mp3) | ayah (per-ayah mp3s); default chapter translationId: 20 # default 20 (Saheeh International) format: short # short = 1080x1920 reel/short, long = 1920x1080 style: minimal-mushaf # folder under templates/ background: # optional, only used by styles with needsBackground query: [clouds, night sky, ocean] # disabled: true ``` Unknown keys are rejected. List reciters (and whether they have word segments) with `npx tsx src/quran/reciters.ts` (`--no-probe` for a quick list). Styles: `ls templates/` and each `templates/<style>/style.json` (`formats`, `needsBackground`, `needsWordTimings`). `minimalistic-graphics` is adapted from [claude-animation-skills](https://github.com/artemnovitckii/claude-animation-skills) (MIT, see its `CREDITS.md`). It takes a scene plan: hairline motifs chosen per passage (`episodes/<id>.scenes.minimalistic-graphics.json`); without a plan it falls back to the staircase. ## Scene plans (for `minimalistic-graphics`) Without a plan this style shows a neutral staircase. To make the drawings follow the meaning, write `episodes/<id>.scenes.minimalistic-graphics.json`. It holds drawing ids and word anchors only, never verse text. Read the word glosses from the fetched packet (`work/<id>/<style>-<format>/composition-data.json`) to find where each meaning starts, and tell the user why you chose each drawing so they can review it. Drawings (`templates/minimalistic-graphics/style.json`): revelation, books, arches, listen, return, capacity, scales, wander, burden (`large`), rain, shelter (`payoffWord`). If nothing fits a passage, use a neutral drawing like `listen` rather than force a meaning. Example: `episodes/baqarah-285-286.scenes.minimalistic-graphics.json`. ## 2. Run ```bash npm run make -- episodes/<id>.yaml # spec.style + spec.format, full render npm run make -- episodes/<id>.yaml --style a,b # several styles npm run make -- episodes/<id>.yaml --style all --format long npm run make -- episodes/<id>.yaml --skip-render # compose + validate + metadata only (fast) npm run make -- episodes/<id>.yaml --reciter 3 --reciter-source ayah npm run make -- episodes/<id>.yaml --background ~/clips/sea.mp4 # local background clip npm run make -- episodes/<id>.yaml --dev-fixture-bg # dev only: work/fixtures/background.<format>.mp4 when no PEXELS_API_KEY ``` Other flags: `--quality draft|standard|high` (default high), `--no-check` (skip `hyperframes check`, lint still runs), `--refresh-audio`, `--scenes <plan.json>` (scene plan for animated styles; default `episodes/<id>.scenes.<style>.json`, else `episodes/<id>.scenes.json`, else none — motif ids + verse/word anchors only, never Quranic text; validated against the packet and style.json `"scenes"`). Renders above ~12 Mbps video are re-encoded (libx264 crf 20, audio copied) before packaging. Background-needing styles use Pexels (`PEXELS_API_KEY` in `.env`), else `--background`, else (dev only) `--dev-fixture-bg`; otherwise that style is reported FAILED with a clear message. Styles needing word timings are skipped when the reciter has no word segments. Processed backgrounds are cached in `work/<id>/_background-<format>-*/`, audio in `work/_audio/`. ## 3. Outputs - `work/<id>/<style>-<format>/` — composed HyperFrames project (`index.html` with injected `window.__QURAN_DATA`, `audio.mp3`, `background.mp4`, `fonts/`, `composition-data.json`, `validation.json`, `renders/`). Preview: `cd` there and `npx hyperframes preview` / `snapshot --at 3,20`. - `outbox/<id>/<style>-<format>.mp4` + `.metadata.json` — title/description/hashtags generated by code from packet fields, full translation text, reciter + translation credit, Pexels attribution, packet sha256, validation status. Do not hand-edit Quranic text in metadata. A summary table is printed at the end; exit code 1 if any style FAILED. ## 4. When validation fails Read `work/<id>/<style>-<format>/validation.json`. Categories: `static` (Arabic codepoint in template source), `runtime-text` (slot text ≠ packet), `coverage` (missing/duplicated words or translation), `outside-slot` (Arabic text not in a `[data-slot]`), `safe-area` (short: text outside x 80–1000, y 260–1500), `lint`/`check` (hyperframes). Fix the template (see `templates/CONTRACT.md`). Validator tests: `npx tsx --test src/validate/*.test.ts`.
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · msrshahrukh100/quran-content-creator-claude-skill | 95 | MIT |
| mvrks-inc/nolang-plugin | 100 | Apache-2.0 |
| murathany7/reel-studio | 100 | MIT |
| TheNoumanDev/zero-dollar-studio | 100 | MIT |
| EyesofInfinity/claude-resolve-editing-suite | 100 | MIT |
| jutsuhq/content-factory | 100 | MIT |
| theunisdk/video-studio | 100 | MIT |
| GoldLegendW80/llm-video-maker | 100 | MIT |
| av/skills | 95 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki 237b008eef02 (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100quran-ayah-verification
Sprawdza tekst wersetu Koranu z zaufanego źródła, porównuje go znak po znaku i podaje numer sury oraz wersetu. - 100paw-cra-agent-video-producer
Planuje i produkuje wideo krótkie, długie, odcinkowe i motion graphics, z napisami i lektorem. - 90swift-render-make
Tworzy film MP4 z scen SwiftUI przez swift-render: zwiastun, reklamę lub grafikę ruchomą aplikacji albo pakietu Swift. - 100mirroring-reference-video
Dopasowuje własne wideo do struktury, tempa i stylu wizualnego podanego filmu referencyjnego. - 100video-marketing
Pomaga planować marketing wideo, pisać scenariusze i optymalizować krótkie oraz długie formy: skrypty TikTok, Reels i YouTube, haki i strategię treści. - 90make-faceless-video
Tworzy krótki film bez twarzy z lektorem i napisami na koncie AutoShort lub StoryShort, a render uruchamia po wyraźnej zgodzie.