Agent Skill
Install the remocn agent skill with skills.sh so your AI generates better Remotion videos
Remocn ships an agent skill — a knowledge pack that teaches AI coding agents (Claude Code and friends) how to build videos with remocn: how the two component tiers differ, how to budget a timeline, how to compose a video out of beats, and where to read the rest.
Once installed, your agent picks the right components, budgets the timeline correctly, and styles your additions to match the library — without you pasting context every time.
The skill does not carry a copy of the component reference. It reads these docs live, starting
from /llms-components.txt and following the link to
each component's page. That way a component added today is visible to an agent whose skill was
installed months ago, and a prop that changed cannot go stale in two places at once. Your agent
needs network access to remocn.dev; without it, it will tell you rather than guess.
Install with skills.sh
The skill lives in the skills/ folder of
the remocn repo. Install it with the skills CLI:
npx skills add Remocn/remocnThis finds the remocn skill and installs it into your agent (for Claude Code:
.claude/skills/remocn/). The CLI is interactive — pick the skill and target agent when prompted.
Common flags
# Install straight to the remocn skill, no prompts
npx skills add Remocn/remocn --skill remocn --yes
# Install globally (available in every project)
npx skills add Remocn/remocn --skill remocn -g
# Target a specific agent
npx skills add Remocn/remocn -a claude-code
# Just list what's available in the repo
npx skills add Remocn/remocn --listUse without installing
To try it for a single session without writing files:
npx skills use Remocn/remocn@remocn | claudeWhat's inside
| File | Contents |
|---|---|
SKILL.md | Install flow, the two tiers and their differing APIs, canvas and timing rules, remocn-specific gotchas, and where to fetch everything else |
references/anatomy.md | Composing a full video — template vs compose vs build, the product-demo beats, the good-vs-slop bar |
references/archetypes/ | Per-archetype recipes: product demo, changelog, feature announcement, OSS showcase, CLI demo, testimonial reel, year in review, pricing reveal, logo bumper |
What the skill reads from this site instead of bundling:
| URL | Contents |
|---|---|
/llms-components.txt | Every installable component with its use / avoid signal, natural length, vibe, tier and dependencies |
/docs/<section>/<name>.md | One component's full reference — props, examples, all use / don't-use notes |
/docs/craft/ | Design defaults, motion principles, anti-patterns |
Any docs URL serves raw Markdown when you append .md, so you can read the same pages your agent
does.
Prerequisites
- An AI agent that loads skills (e.g. Claude Code).
- A Remotion project with remocn — see Installation.