The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the execution of a setup script (scripts/setup.sh) and uses local scripts (e.g., animation-map.mjs) to manage environment configuration and verify animation sequences.- [REMOTE_CODE_EXECUTION]: The setup script installs the hyperframes CLI globally from the npm registry using npm install -g, allowing for the execution of remote code at a high level of system integration.- [EXTERNAL_DOWNLOADS]: The environment setup downloads the chrome-headless-shell binary via Puppeteer and fetches animation components from the hyperframes.heygen.com catalog.- [PRIVILEGE_ESCALATION]: Documentation in scripts/setup.sh and references/troubleshooting.md recommends using sudo to resolve permission issues during the installation of global npm packages and system-level dependencies like FFmpeg.- [DYNAMIC_EXECUTION]: The skill dynamically installs and executes additional functionality, such as shader transitions and social overlays, from an external resource catalog using the npx hyperframes add command.- [INDIRECT_PROMPT_INJECTION]: The skill ingests and renders content from untrusted external URLs via the capture command, creating a potential vector for indirect prompt injection where malicious instructions embedded in a website could influence agent behavior.
Ingestion points: npx hyperframes capture in references/cli.md and references/website-to-video.md.
Boundary markers: None identified; the skill processes the raw HTML/CSS of the captured site.
Capability inventory: Access to node, npx, ffmpeg, and a headless browser for rendering and file system writes.
Sanitization: The lint and validate tools focus on accessibility and visual layout rather than security sanitization of the ingested content.

hyperframes