The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions in SKILL.md direct the agent to silently update the toolchain using npx hyperframes skills update faceless-explainer. The init command also checks for and downloads the latest skill definitions from a remote GitHub repository to ensure the environment is current.
[COMMAND_EXECUTION]: The assemble-index.mjs script utilizes spawnSync to call ffmpeg and ffprobe. These calls are used to process background music, specifically for checking durations and extending short tracks via looping. The use of array-based arguments in spawnSync effectively mitigates common shell injection vulnerabilities.
[COMMAND_EXECUTION]: The audio.mjs script acts as an adapter that executes a shared audio engine located in a sibling directory (../../media-use/audio/scripts/audio.mjs) via node. This is part of a modular architecture where specialized tasks are delegated to shared utility skills.
[DATA_EXPOSURE]: The skill frequently accesses and reads configuration and reference files from sibling directories such as hyperframes-core, hyperframes-creative, and hyperframes-animation. This behavior is expected within the integrated suite of tools provided by the vendor for video production.
[INDIRECT_PROMPT_INJECTION]: The skill processes arbitrary user-provided text stored in capture/extracted/visible-text.txt. This content is eventually used to generate instructions for sub-agents (frame workers). The skill manages this attack surface by using structured data packets (frame-packets.mjs) and applying size limits to the data transmitted to sub-agents.

faceless-explainer