The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npx to fetch the hyperframes rendering package and initiates a download of the chrome-headless-shell binary to a local cache.
[COMMAND_EXECUTION]: Several scripts execute shell commands, including ffmpeg and ffprobe for media extraction and metadata probing, and npx for rendering. The script scripts/caption_interaction.mjs uses powershell.exe to launch default browser handlers for generated HTML review pages.
[DYNAMIC_EXECUTION]: The skill constructs HTML files at runtime that execute animations using the GSAP library based on data generated during the caption planning phase.
[INDIRECT_PROMPT_INJECTION]: The skill processes transcripts and project metadata that enter the agent's context from external files.
Ingestion points: scripts/build_captions.py reads transcript.json and scripts/caption_spatial_context.py reads project metadata.
Boundary markers: The skill implements a SHA-256 hash validation mechanism in scripts/caption_interaction_state.mjs to verify integrity across different stages of the workflow.
Capability inventory: The skill has the capability to write to the local filesystem, execute shell commands, and fetch external packages via standard registries.
Sanitization: The generator script performs HTML escaping on caption text and utilizes structured JSON parsing to mitigate malicious content in input files.

video-add-captions