The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts references/build_hf_clips.py and scripts/scaffold.py use subprocess.run to execute system commands including ffmpeg, ffprobe, and npx hyperframes to perform video encoding and initialize project structures.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, which creates a potential vulnerability to indirect prompt injection.
Ingestion points: The skill ingests data from segments.json, word-timed SRT subtitle files, spec.json configuration files, and user-provided HTML fragments.
Boundary markers: No clear delimiters or instructions are provided to the agent to treat this data as untrusted content.
Capability inventory: The skill has the ability to write to the file system, create symbolic links, and execute shell commands.
Sanitization: Subtitle text is escaped using html.escape, but the custom overlay feature allows for the direct inclusion of raw HTML/JS fragments from external files without comprehensive validation.
[DYNAMIC_EXECUTION]: The skill dynamically generates index.html files containing complex animation logic and user-supplied text. These files are subsequently executed by the HyperFrames rendering framework.

wjs-overlaying-video