The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various system commands for asset capture (e.g., adb, xcrun, screencapture) and video rendering (e.g., ffmpeg). These operations are required for the skill's primary function of generating motion graphics from real product captures.
[EXTERNAL_DOWNLOADS]: The engine setup process fetches the Playwright browser framework and its dependencies from established public registries (NPM).
[DYNAMIC_EXECUTION]: The render.mjs utility uses Playwright to dynamically load and execute local animation scripts and configurations within a headless browser to render video frames.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it ingests data from external URLs and app simulators provided by the user. While it has the capability to write files and execute commands, the data is primarily used for visual rendering.
Ingestion points: Captures from user-provided URLs or simulators, and user-supplied scripts (brief.md, beats.md).
Boundary markers: None identified in the prompt instructions for isolating external content.
Capability inventory: Shell command execution (FFmpeg, ADB, XCRun), file system access (writing renders and assets), and local network/browser execution (Playwright).
Sanitization: No specific sanitization or filtering logic is described for data extracted from external captures.

motion-graphics