claude-video-kit skill
The video-explainer skill turns a script into a vertical 9:16 explainer: narration, captions, Remotion and a pre-render review
- 122 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The pipeline scripts run locally, install npm and Python dependencies and call ffmpeg
- The script text goes to Fish Audio, which needs a secret API key stored in `.env`
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/claude-video-kitDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The pipeline scripts run locally, install npm and Python dependencies and call ffmpeg
- The script text goes to Fish Audio, which needs a secret API key stored in `.env`
This is third-party code. Review the repository files before installing.
What it does
The `video-explainer` skill takes an agent from a research note, a finished script or a finished MP4 to a reviewed vertical explainer. It first picks the simplest production strategy, and if that fails the checks it falls back to the repository's deterministic Remotion recipe. The skill itself uploads and publishes nothing. The `script.json` pipeline: narration, word-level caption alignment through faster-whisper, metadata build and a 1080×1920 render. Slide types are `cover`, `text`, `code` and `numberHero`. Slide length follows the audio length. Narration comes from Fish Audio or a local IndexTTS2 with a GPU, and macOS `say` works for testing the pipeline. A pre-render review is mandatory: its receipt is bound to the exact contents of `script.json`, and rendering will not start without one. The final video passes a release gate: a shorts check (format, up to 60 seconds, scene rhythm, hook) and a hash-bound evaluation of the final file.
Who it is for. Authors of technical and research explainers in Claude Code, Cursor or Codex who want a repeatable vertical video without a timeline editor.
Good fit when
- A written breakdown has to become a Shorts, TikTok or Reels video with narration and captions
- A series of videos has to share one look and you want scripts stored as JSON in the repository
- A second person or agent reviews the video and the result has to be tied to a specific version of the script
- You want to confirm on the demo that the whole pipeline runs on your Mac without API keys
Not a fit when
- You want one-click topic-to-AI-video factories: the author rules out the generative lottery and bets on auditable scripts
- You want automatic publishing: the skill uploads nothing and hands over a package after review
- The video is horizontal or long: the main pipeline targets 9:16, and long form is covered by a separate experimental document
Example request
Use video-explainer: take scripts/my-first/script.json, run the pre-render review and build a vertical 1080×1920 videoLimitations
The `v0.3.0-rc.1` release is a public candidate and there is no stable version yet. The pipeline was tested on macOS with Apple Silicon. It needs Node 20 or newer, Python 3.10 or newer, ffmpeg and, for default narration, a Fish Audio key. Narration through `say` is rough: it confuses Chinese homographs, and Whisper alignment on the `base` model can be imprecise, so use Fish Audio for anything you publish. The skill drives a local clone of the repository and does not bundle Remotion. A reviewer decides the pre-render checks, and the tool enforces the process without judging the truth of the content.
How to disable. For the plugin run `/plugin uninstall video-explainer-kit` in Claude Code. A skill installed through the Skills CLI is deleted from `.claude/skills/video-explainer`, and the local clone is removed like any folder.
FAQ
How do I install video-explainer in Claude Code?
Run `/plugin install video-explainer-kit --marketplace runesleo/x-reader` in Claude Code 2.1.275 or newer. For Codex use `codex plugin marketplace add runesleo/x-reader`, then `codex plugin add video-explainer-kit@runesleo`.
Do I need API keys for the first run?
No, the demo works without keys: it uses macOS `say` as demo-quality narration, renders 1080×1920 and runs the shorts check without uploading. For real videos set `FISH_AUDIO_API_KEY` and `FISH_AUDIO_VOICE_ID` in `.env`.
What blocks a render?
Rendering will not start if the review receipt is missing, stale, or marked `fix` or `block`. The receipt is bound to the exact `script.json`, so any script edit needs a new review.
Which slides can a script contain?
Four types: `cover` for opening and closing cards, `text` for narrated text, `code` for code blocks and `numberHero` for a large figure at the start of a short. Narration goes in a separate `voice_text` field so it can sound more natural than the on-screen text.
What does verify-shorts check?
Four conditions: a 1080×1920 canvas, a length of up to 60 seconds, scene rhythm and the hook entrance. Run `node scripts/verify-shorts.mjs <out.mp4>` after rendering.
Discussion
No questions or reviews yet.
Related
SkillsA skill that turns a topic into a narrated 4K video: research, script, local TTS, Remotion composition and an FFmpeg mix
A Claude Code and Codex skill that turns a topic into a narrated motion-graphics explainer video on Remotion with subtitles and chapters
A skill for Claude Code and Codex that builds cinematic product videos with Remotion from shot recipe cards, a ready template and audio assets
A skill that turns a coding agent into a studio for voiceover explainer videos: word-level sync, a motion library and Remotion assembly
Skills for faceless video channels: one topic becomes a video with voiceover, subtitles, covers and per-platform copy.
A director skill: it interviews you about a video idea and turns it into a second-by-second shot script for HyperFrames
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| claude-video-kit (this card) | no | 122 | |
| Remotion Agent Skills | no | 6,431 | teaches the agent to write Remotion compositions without script review or narration |
| Video Podcast Maker | no | 1,715 | makes video podcasts from text with a different recipe |
README in short
The English README, mirrored in Chinese, opens with a short audience note and the release status. It then covers plugin and Skills CLI installation, a quick first success on the demo, the release gate, the pipeline diagram, the slide list and the known limits of the first version.
README badge
Original SKILL.md
--- name: video-explainer description: Turn a research brief, finished script, or generated MP4 into a reviewed and verified vertical explainer video. Use when an agent should choose the simplest current production strategy, preserve strict final-artifact quality gates, run the deterministic Remotion fallback, diagnose the local toolchain, or verify an existing MP4 before any manual publication step. --- # Video Explainer Version note: v0.3.0-rc.1 / 2026-09-27 / Make production strategy adaptive while keeping the deterministic Remotion recipe as a strict fallback. Final-artifact acceptance is strategy-independent and bound to the exact MP4 bytes. Keep upload, account access, credentials, private voice assets, hosted rendering, and publication out of scope. Produce a local video and evidence packet. Never upload or publish from this skill. ## Core contract This Skill separates **what must be true of the final artifact** from **how the artifact is produced**. Before forcing an existing recipe, run a capability challenge: 1. Can the current model or available capability produce the final MP4 directly at the required quality? 2. Which requirements are real product, safety, brand, or evidence invariants? 3. Which existing steps only compensate for older model limitations? 4. Can intermediate stages be skipped without weakening final acceptance? 5. What is the simplest strategy likely to pass the final gates? 6. If that strategy fails, fall back to the deterministic repository-owned recipe. Choose one execution strategy: - `direct_artifact` — a current model/capability produces the final MP4 directly. Do not require `motion-storyboard.json`, `script.json`, TTS, alignment, or Remotion merely to satisfy a historical recipe. - `agent_orchestrated` — an agent may choose and combine tools as implementation details. Intermediate artifacts are optional unless they are needed to satisfy the brief or evidence requirements. - `legacy_remotion` — use the repository-owned deterministic path when direct generation cannot preserve facts, branding, layout control, narration identity, or other required qualities. All three strategies converge on objective MP4 verification plus an independent final-artifact review. Production freedom never bypasses final acceptance. ## Install The deterministic fallback orchestrates a local clone of `claude-video-kit`; it does not bundle the renderer. From the repository root, install the runtime and register the Skill with: ```bash npm ci --prefix remotion npx skills add runesleo/claude-video-kit --skill video-explainer ``` For repository development, run commands from the repository root. ## Workflow