Media & Video

Code Explainer Video

code-explainer-video

Render a cinematic chaptered explainer video for any CLI or dev tool — Remotion "Terminal Noir": code that types itself, streaming terminal output, growing dependency graphs, continuous BGM.

remotionvideodevreltutorial
Install
mkdir -p ~/.claude/skills && curl -fsSL https://skill.metacog.co.kr/dist/code-explainer-video.zip \
  -o /tmp/code-explainer-video.zip && unzip -oq /tmp/code-explainer-video.zip -d ~/.claude/skills
Files28
Size74.0 KB
Bundled foldersassets/ references/ scripts/
LicenseMIT

Authored by jeonck (MIT) · Browse files on GitHub · Download zip

When Claude uses it

Produce a cinematic, chaptered explainer/tutorial VIDEO for a code, CLI, or dev-tool topic — a dark "Terminal Noir" console aesthetic (Remotion + syntax-highlighted code that types itself, streaming terminal output, dependency graphs, chapter cards, continuous BGM). Use when the user asks to make a video/tutorial/explainer/promo about a developer tool or command (Terraform, Docker, Kubernetes, Git, SQL, an SDK/API/CLI), a multi-chapter "N topics in one video" cut, or to replicate the Terraform 5-chapter manual. Builds chapters in parallel via subagents and encodes hard-won production rules (technical-accuracy validation, count reconciliation, seam fades, readability floors, deterministic render, tsc self-verify, independent review).

SKILL.md

Code Explainer Video

Overview

This skill builds a polished, chaptered explainer video for any code/CLI/dev-tool topic and renders it to MP4. The aesthetic is a dark IaC console ("Terminal Noir"): deep indigo, violet accent, monospace code that writes itself, terminal output that streams in, and dependency graphs that grow edges-before-nodes. It ships a reusable Remotion engine (src/lib), topic-agnostic scene templates, a parallel chapter-build workflow, and a distilled list of the mistakes that ruin these videos so they aren't repeated.

Default shape: a single continuous ~4–5 minute cut of N (default 5) chapters — intro → [chapter card + chapter] × N → outro — with one continuous music bed. Each chapter can also be rendered as a standalone clip.

When to use

Trigger when the request is to create a video that TEACHES or PROMOTES a developer tool, command, or codebase concept — e.g. "make a video explaining Docker Compose", "a 5-chapter Terraform tutorial video", "an animated intro for my Kubernetes course", "turn these git commands into a short explainer". Also use to reproduce or extend the Terraform 5-chapter manual. Not for live-action screen recordings, non-code marketing videos, or full-length (30-min) narrated courses with real voiceover — this produces short-per-chapter animated cuts.

Related skill: video-shotcraft (cinematic PRODUCT/webpage promos, its own shot library, and the free-commercial audio this skill reuses). Prefer THIS skill when the subject is code/CLI/dev-tooling and the format is chaptered explanation.

How to use it

Follow references/workflow.md end-to-end. Read references/pitfalls.md alongside every stage — it is the point of this skill. In short:

  1. Brief & chapter plan — pick the topic, N distinct chapters (concept → do

→ connect → operate → scale), language (English default), the REAL syntax to show (verify accuracy; all data fictional), and the usage profile (Full vs Low-usage — see references/usage-profiles.md). Default to Full for a single polished deliverable; use Low-usage (Sonnet chapters, minimal image QA, one capped review) when the user is on a Claude subscription/Max plan, wants a cheap draft, or is batching many videos.

  1. Scaffoldscripts/scaffold_project.sh <target-dir> --audio copies the

engine + templates and installs deps (and the reused SFX/BGM). It renders a silent demo out-of-box. Design tokens live in src/lib/theme.ts — reuse them; never invent a second palette.

  1. Design beats — each ~1440f chapter = 3–4 beats, each mapped to one

template: CodeEditorScene / TerminalScene / GraphScene / a concept beat. One animation technique is the star once across the whole video.

  1. Build chapters IN PARALLEL — one subagent per chapter using

references/chapter-subagent-prompt.md; each self-verifies with tsc.

  1. Assemble — wire intro + cards + chapters + outro on ONE timeline

(src/Main.tsx), add captions + flash cuts, keep boundaries seam-free.

  1. Sound — one continuous BGM bed + an SFX table pinned to relative frames.
  2. QA continuously — render stills, look, extract keyframes.
  3. Independent review — a clean-context subagent checks the finished MP4

against pitfalls.md with frame evidence; fix, then final render.

Resources

item before delivery (correctness, determinism, readability, seams, sound).

agent model/effort per stage, how many QA frames to open, review scope). On a Max/subscription plan, run subscription-auth by default and use Low-usage to stay within plan limits; fall back to an API key only for overflow.

chapter to a parallel subagent.

as-is), src/scenes/ (MasterIntro, MasterOutro, ChapterCard + CodeEditorScene/ TerminalScene/GraphScene templates), src/Main.tsx (assembly template).

Requirements: Node ≥ 18, ffmpeg (for keyframe QA), ~1 min render per minute of video on a laptop. Audio is optional but expected; the --audio flag reuses the video-shotcraft skill's free-commercial SFX/BGM.