From brief to vertical video: Claude Video Kit

Where this fits

You've just finished a deep research brief in Claude Code — a technical explainer on, say, vector database indexing. Now you need a vertical video for LinkedIn or X. Instead of exporting the brief and re-editing in a timeline tool, you hand the agent the brief and get back a narrated 9:16 explainer with captions, TTS, and motion graphics — all reproducible from a JSON script.

Why agents benefit

• The skill produces an auditable script.json that binds to the render; you can review and edit before any pixels are generated.
• A pre-render review gate forces an independent check — the agent must confirm the script matches the brief before committing to a render.
• Caption alignment is handled automatically from the TTS output, so timing issues don't require manual frame-by-frame fixes.
• The pipeline is deterministic — same script in, same video out. No generative video lottery.
• It runs entirely locally via Remotion; the skill never uploads or publishes your content.

Practical scenarios

Technical explainer for a developer audience

You wrote a 2,000-word brief on how Postgres MVCC works. The agent converts it into a 60-second vertical explainer with captions, ready for X or LinkedIn.

Research brief → investor update

Your team publishes a weekly research digest. Instead of a slide deck, you generate a short narrated video summarizing the key findings, with the script reviewed by a teammate before render.

Reproducible content pipeline for a newsletter

You maintain a weekly newsletter with a companion video. The skill lets you regenerate the video each week from a new brief, keeping the same visual style and narration voice.

Add it to your agent workflow

Install via the skills CLI:

npx skills add runesleo/claude-video-kit --skill video-explainer

Then verify your environment and run the demo:

npm run doctor -- --output /tmp/video-explainer-first-success
npm run demo -- --output /tmp/video-explainer-first-success

After that, point the agent at a brief file and ask for a script.json review before rendering.

Compare before adopting

Before adopting, compare against:

• Quality signals: does the output look like a template or a tailored explainer? Check sample renders in the repo.
• Maintenance freshness: check commit history and release cadence (v0.3.0-rc.1 is recent).
• Alternatives: other video-generation skills that use generative models — weigh determinism vs. visual variety.
• Workflow fit: does the pre-render review gate match how your team reviews content? If you need one-click publishing, this skill adds a deliberate checkpoint.

Why it is worth tracking

Claude Video Kit sits at the intersection of two growing trends: agent skills that produce artifacts beyond text, and a demand for auditable, reproducible media pipelines. With 114 stars and an active release candidate, it's early enough that the API may shift, but mature enough to demonstrate a real workflow. Track it if you're building content pipelines where script review matters more than visual novelty. You can follow its progress on OpenAgentSkill at https://www.openagentskill.com/skills/runesleo-claude-video-kit.