VideoSDK Agent Skill

The VideoSDK agent skill teaches your AI coding agent to build on VideoSDK: calling, live streaming, and the auth work each needs. One skill, videosdk, covering every supported SDK.

Install​

npx skills add videosdk-live/skills

npx skills is agent-agnostic. It asks which agents to install into, so pick the ones you want from the list (Claude Code, Cursor, Codex, and the others it supports) and it installs the skill there. Restart your agent afterwards.

To place it by hand instead, copy the videosdk folder into your agent's skills directory, for example ~/.claude/skills/videosdk.

What happens on first use​

The skill activates on intent, so you describe the app rather than name the skill. Then it asks before it writes anything.

It works out your stack, whether you want a call, an interactive live stream, or an HLS broadcast, and what the app is for. Then it restates the plan in a single sentence. If it guessed wrong, correcting it costs you one word instead of a rebuild.

It also settles auth on the way through. Tokens get minted on a server route, and the API secret stays out of client code.

Example prompts​

Add video calling to this app.

Build a telehealth consultation, doctor and patient, with a waiting room.

Build a webinar where a host presents and the audience can ask questions.

Build a live shopping stream where viewers can be brought up on stage.

Name the platform and the token backend up front and it has fewer questions to ask.

Pair it with the MCP server​

The skill decides what to build. When it needs a real method name, it asks the VideoSDK MCP server for the page. Install both.