The Agent Skills Directory

[COMMAND_EXECUTION]: The skill revolves around executing the hyperframes CLI tool via npx. It includes commands for project initialization, rendering, and AWS Lambda management (lambda deploy, lambda render, etc.).
[EXTERNAL_DOWNLOADS]: The CLI tool downloads several external resources during normal operation, including a pinned version of the Chrome browser (browser ensure), AI coding skills (skills), and various models for asset preprocessing (TTS, transcription, and background removal).
[INDIRECT_PROMPT_INJECTION]: The capture command scaffolds projects by crawling a live URL, which introduces an ingestion point for untrusted external data. While this is a standard feature of the tool, it represents an attack surface if an agent is subsequently asked to process the captured content.
Ingestion points: npx hyperframes capture <URL> (scaffolds a project from a live website, downloading assets and creating layers).
Boundary markers: No explicit markers or "ignore instructions" warnings are mentioned for the content captured from external URLs.
Capability inventory: The CLI can perform file system writes, subprocess execution (FFmpeg), and network operations (AWS Lambda deployment and the publish command).
Sanitization: The documentation does not describe sanitization or filtering of the content retrieved during the capture process.

hyperframes-cli