GitHub Copilot CLI

Give Copilot CLI the HyperFrames skills, make a first request, and preview the project locally.

GitHub Copilot CLI supports both project and personal agent-skill locations. HyperFrames skills give it the framework rules and creation workflows required to build a real project.

Install the skills

Run:

npx hyperframes skills update

This installs or refreshes the global core set and links it into compatible installed agents. Copilot reads the shared ~/.agents/skills location, so the skills remain available across projects.

Open the project folder you want to edit, then start Copilot:

copilot

If you installed the skills during an active session, run:

/skills reload

Use /skills list or /skills info hyperframes to confirm discovery. GitHub’s current skills guide lists every supported location and management command.

Make the request

Using /hyperframes, make a 10-second product intro for https://example.com.

Copilot can load a relevant skill automatically, but naming /hyperframes makes the intended entry point explicit. Continue in the same session for revisions.

Preview and render

In another terminal:

npx hyperframes preview

Then ask Copilot to render or run:

npx hyperframes render --output output.mp4

Run npx hyperframes lint and npx hyperframes check before delivery.

Hosted alternative

For hosted HyperFrames tools, add the MCP server to one Copilot session:

copilot --additional-mcp-config \
  '{"mcpServers":{"hyperframes":{"url":"https://mcp.heygen.com/mcp/hyperframes/"}}}'

For a persistent connection, use Copilot’s current MCP configuration. The HyperFrames MCP guide explains the creation workflow.

Prompting guide

Vocabulary and patterns that produce better compositions.

Understand the agent flow

See what the agent does between the request and the finished project.