Communitygithub.com

thativanng/reels-maker

The interview before any video is built. It reads what the person already gave (source text, CV, links, brand files), then asks the rest in a few short rounds of numbered questions with a recommended answer marked, always including the format (platform, reel or video, ratio, layout, length, sound). It ends in a one-screen brief that splits "you said" from "I assumed", plus a storyboard, and builds nothing until the person says yes. Use it whenever someone asks for a reel, a video, an explainer, a career film, an animation or a "make a video about X", before film-engine or any style pack, and again when a request changes its purpose, platform or audience mid-way.

What is reels-maker?

reels-maker is a Claude Code agent skill that the interview before any video is built. It reads what the person already gave (source text, CV, links, brand files), then asks the rest in a few short rounds of numbered questions with a recommended answer marked, always including the format (platform, reel or video, ratio, layout, length, sound). It ends in a one-screen brief that splits "you said" from "I assumed", plus a storyboard, and builds nothing until the person says yes. Use it whenever someone asks for a reel, a video, an explainer, a career film, an animation or a "make a video about X", before film-engine or any style pack, and again when a request changes its purpose, platform or audience mid-way.

Works with~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/thativanng/reels-maker/tree/HEAD/skills/reel-intake
View original→Browse all skills

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

OpenAIChatGPTClaudeClaudeGeminiGeminiGrokGrokPerplexityPerplexityDeepSeekDeepSeek

Documentation

reel-intake: ask before you build

Where the kit lives: run every command in this skill from the reels-maker folder (the clone of github.com/thativanng/reels-maker that holds template/ and tools/). If the project has none, set it up first by following that repo's SETUP.md.

The rule: nothing is built (no look, no frame) until the person says yes to the brief and its storyboard. The expensive mistakes on a code-built video are rarely timing; they are answers nobody asked for. The real purpose surfaces after the script is written. The ratio turns out wrong for the platform. The look turns out to be one the person finds cheap. Each of those costs a rebuild; a question costs one line.

0 · Look before you ask

Read everything already in hand, and never ask what you can look up:

  • the request itself, any source text, links, a CV, their site or profile
  • brand files (colours, fonts, logo) in the project
  • .reels/looks-log.md, if it exists (reel-style's memory of past looks)

A field the material answers is filled and marked "from your CV", "from your site", and so on. A guess is not an answer: a guess goes in the brief under "I assumed", where the person can see it and correct it.

1 · How to ask

  • Rounds, not a form and not a drip. One round = one block below, in one message. Number every question. Mark the recommended answer with ★ and give a one-line reason.
  • Facts get options (platform, length, sound). Creative fields get a drafted suggestion to edit, never a blank: "The one thing a viewer should remember: ★ 'A forecast is a guess; safety stock covers the surprises.' Keep, edit or replace?"
  • A question that depends on an earlier answer comes pre-filled from it (the ratio follows the platform), or waits for the next round.
  • Shortcuts the person can type: defaults accepts every ★ in the round. just build it ends the interview: every open field becomes a stated decision in the brief, with its reason.
  • Four rounds is normal, five is the limit. Stop as soon as every required field is answered or defaulted.

2 · The rounds (the full question bank, with options and defaults, is in references/rounds.md)

Round 1 · Purpose (required)

  1. What is it about? (only if no source was given)
  2. The one thing a viewer should remember (drafted for them; for a vague request, offer 2–3 drafts)
  3. Who is the ONE viewer? (a recruiter, a customer, peers in their field, strangers outside it)
  4. What should that viewer do after watching? (follow · comment · tap the link · save or send · just remember)

Round 2 · Format (required, all six, in this order; facts and keep-clear zones are in references/platforms.md)

  1. Where is it posted first? The first platform is the master; the rest become cut-downs.
  2. Reel or video? Ask only where the platform has both (LinkedIn, X, YouTube, Facebook).
  3. Ratio, defaulted from the platform: Reels, TikTok and Shorts → 9:16; LinkedIn → 4:5; YouTube → 16:9.
  4. Layout: ★ let reel-style show three, or name one if they already know (it then binds all three looks). State the keep-clear zone; don't ask about it. On organic Reels, TikTok and Shorts, words stay in the middle band (house practice, not a platform number): the app covers the bottom quarter and the right-hand button rail. A boosted or paid reel keeps the bottom 35% clear (Meta's ads guide); use the -ads platform keys.
  5. Length: a hard cap, or follow the content? Show the platform's official limits next to it.
  6. Sound: works muted and better with sound (★ default) · sound-led · silent-safe. Then voiceover, music and caption style.

Round 3 · Material and story (ask what applies)

  • What already exists: screenshots, recordings, photos, footage of them, logo, data, or nothing. Missing material becomes a labelled placeholder, never a fake.
  • Every number: measured, a target, or illustrative, and where it comes from. Never invent one.
  • Who writes the words on screen? (★ I draft, your edits are final · your words exactly) Which language, and UK or US spelling?
  • One-off or a series? (a series locks one look across episodes; reel-style handles it)
  • A journey or career story adds the journey block (references/journey.md): every role and every hat, name/describe/hide each employer, the claims a colleague would call a stretch.

Round 4 · Look, guardrails, review (required)

  1. 1–3 videos they like, and what looks cheap to them. This is the costliest question to skip.
  2. Brand colours and fonts (looked up first; asked only if none were found).
  3. Anything that must NOT appear (numbers, names, faces, logos), and anything that MUST appear (handle, end card, disclaimer).
  4. The cover, the still people see before they press play (profile grid, LinkedIn thumbnail): ★ the hook line, big, over the opening look. Without one, the platform picks a frame for you (on LinkedIn reportedly the first; unverified), which is rarely the best.
  5. How they want to review: ★ stills → a 10-second strip → the full film after the independent check → a phone preview.

3 · The close

  1. Clash check. Read the answers against each other before writing the brief:
    • 9:16 for LinkedIn → suggest 4:5
    • sound-led on LinkedIn → burned-in captions are required
    • a long script in a 9:16 reel's middle band (the bottom quarter kept clear; 35% if paid) → cut words, not type size
    • over 3 minutes on Instagram → it won't be recommended to new audiences
  2. The brief, on one screen (template in references/brief.md), saved as reels/<short-name>/brief.md: every reel gets its own folder. Split it into You said and I assumed (and why). Include the format line: platform → reel or video → ratio and size → layout and keep-clear zone → length → sound.
  3. The storyboard table: time · on-screen words, exactly as they'll appear · sound and caption · why this beat. The hook lands inside 3 s, and every beat's duration adds up to the length.
  4. Ask for "yes". A correction gets a fresh read-back. Never build in the same message as the brief.

Then hand over: reel-style picks the look (3 stills), and film-engine builds in it, with the page sized by the brief's format.

Never

  • Ask what the material already answers.
  • Invent a number, a quote, a client or a result.
  • Swap the person's words for yours. Proofread them (grammar, flow), don't rewrite them.
  • Skip the format round, even when the platform seems obvious.
  • Pick the look inside the interview. That is reel-style's job, shown as stills, never named from a list.

Credits

The interview stance (the agent asks before acting, with a recommended answer for every question) comes from Matt Pocock's grill-me (mattpocock/skills, MIT). The ask-once rules (recommended option first, skip what's answered, said vs assumed, a clash check before the summary) are adapted from HyperFrames' intent interview (heygen-com/hyperframes, Apache-2.0). The brief fields follow common agency video briefs. Platform facts were checked on 2026-10-06 against the platforms' own pages where they could be read; the ones that could not (X) are marked as unverified in references/platforms.md.

Individual skills in this repo

This repo contains 4 individual skills — each has its own dedicated page.

thativanng/reels-maker

The gates every code-rendered film passes before anyone sees it. A 40 fps collision pass (no mover touches a solid or another mover), pixel probes (nothing sits over or behind a face, a head, a subtitle or the frame edge), then an independent auditor (a fresh agent) who reads frame sheets at ≥2 fps and signs off. The builder never signs off their own film. Use it before sending ANY render, after every fix round, and whenever someone says "check the film", "audit this", "is it ready", or "why does this frame look wrong".

thativanng/reels-maker

Build and render a film, reel or animation as code. One HTML page draws any frame on demand with window.seek(t), and a headless browser renders every frame at 2x into an MP4 with sound. Use it whenever someone wants a video made with code rather than a video generator, says "make a film / reel / animated explainer / career video", "render this to MP4", "turn this page into a video", or needs the finishing steps (audio cues, loudness, colour-safe encode, phone preview). It is style-agnostic and renders 9:16, 4:5, 1:1 or 16:9. It runs after reel-intake (the brief) and reel-style (the look), builds on template/starter.html or a style pack such as voxel-walk, and passes film-audit before anything is shown to anyone.

thativanng/reels-maker

Picks the look of a reel or film before it is built. It shows three genuinely different looks as stills of the same opening frame with the same words, so the person judges the look and not the content. It keeps a log of every look ever shown, so no look comes back unless asked for, and it locks one look on purpose for a series. Use it after reel-intake's brief is signed and before any test strip, whenever someone says "what should it look like", "give me options", "make it look different from last time", "it looks cheap", or starts a new reel or a new series.

thativanng/reels-maker

The voxel career-film style. A blocky 3D version of the person walks, without ever vanishing, through one scene per role. Their outfit changes as they walk, each scene rises out of the floor ahead of them, and a dark banner tracks year, company and role. Use it when someone wants their career, a journey or a "how I got here" told as a short 3D film for LinkedIn or a portfolio, says "make my career film", "turn my CV into a video", "voxel film", or wants a new chapter added to an existing one. It runs on film-engine (render, audio, finish) and must pass film-audit before anything is sent.

Related Skills

steipete/video-frames

Extract frames or short clips from videos using ffmpeg.

community

steipete/summarize

Summarize or transcribe URLs, YouTube/videos, podcasts, articles, transcripts, PDFs, and local files.

community

openclaw/openclaw

Use when testing, fixing, or extending the OpenClaw Control UI GUI with Vitest + Playwright end-to-end checks, mocked Gateway WebSocket flows, mocked dashboard runs, screenshots/videos, or agent-verifiable browser proof.

community

obra/superpowers

Obra superpowers agent skills.

community

affaan-m/ECC

See, Understand, Act on video and audio. See- ingest from local files, URLs, RTSP/live feeds, or live record desktop; return realtime context and playable stream links. Understand- extract frames, build visual/semantic/temporal indexes, and search moments with timestamps and auto-clips. Act- transcode and normalize (codec, fps, resolution, aspect ratio), perform timeline edits (subtitles, text/image overlays, branding, audio overlays, dubbing, translation), generate media assets (image, audio...

community

affaan-m/everything-claude-code

End-to-end marketing campaign planning and execution. Covers audience research, positioning, campaign angle definition, landing page copy, email sequences, social posts, ad copy, short-form video scripts, and content calendars. Use as the orchestration layer for multi-channel product launches. Use when planning or executing a multi-channel product launch, or producing landing page, email, social, or ad copy.

community