The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses several command-line tools to process video and manage assets.
Invokes ffmpeg to encode PNG sequences into MP4 videos and mux audio tracks in scripts/ark.py and scripts/composite.py.
Utilizes node or bun to run scripts/frames.mjs for rendering animation frames.
Recommends using cloudflared to create temporary tunnels for serving video assets to the Ark API, with explicit instructions for the user to verify the action and tear down the tunnel afterward.
Mentions executing a setup script from a related skill (install.sh or install.ps1).
[EXTERNAL_DOWNLOADS]: The skill interacts with external services to generate and fetch video content.
Communicates with BytePlus ModelArk and Volcengine Ark APIs (owned by ByteDance) to submit video generation tasks.
Automatically downloads generated MP4 files from the API's provided URLs using urllib.request.urlretrieve in scripts/ark.py.
[DYNAMIC_EXECUTION]: The skill performs dynamic rendering and scripted asset generation.
scripts/frames.mjs uses the Chrome DevTools Protocol to evaluate JavaScript within a headless browser instance to control CSS animations during frame capture.
references/compositing.md provides a Python script executed via a heredoc with uv to generate green-screen tracking markers.
[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect instructions through its prompting mechanism.
Ingestion points: Reads user-provided prompt.txt and renders HTML animation boards provided in the canvases/ directory.
Boundary markers: No explicit boundary markers or isolation instructions are used when interpolating these inputs into the video model prompts.
Capability inventory: Includes file system writes, network requests to the Ark API, and execution of FFmpeg and Chrome.
Sanitization: None detected; however, the primary target of the generated instructions is an external video generation model (Seedance 2.5), minimizing risk to the agent's core logic.
[SAFE]: The skill implements best practices for secret management by instructing the agent to store API keys in environment variables or restricted files (chmod 600) and explicitly forbidding the logging or printing of these credentials.

sp-create-video