The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the execution of multiple shell commands to set up the environment and perform the video rendering.
It instructs the installation of the playwright package via npm install and the installation of the Chromium browser via npx playwright install chromium.
The provided render.mjs script uses the node:child_process module to spawn ffmpeg processes for image tiling, video segment concatenation, and final video encoding with audio.
[DYNAMIC_EXECUTION]: The skill facilitates the creation and execution of dynamic scripts.
The AI agent is instructed to generate a custom render.mjs script and a product-specific HTML file at runtime.
These generated files are then executed locally using Node.js and Playwright to generate the video frames and soundtrack.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, creating an attack surface for indirect prompt injection.
Ingestion points: The skill is instructed to fetch content from user-provided URLs to determine product features, colors, and fonts (SKILL.md, Step 1).
Boundary markers: There are no explicit boundary markers or instructions to isolate the fetched content from the agent's internal logic.
Capability inventory: The skill has significant capabilities, including file writing, network access (via Playwright), and local command execution (Node.js and FFmpeg).
Sanitization: The instructions do not specify any sanitization, filtering, or escaping for the text fetched from external websites before it is used in the generation process.
[EXTERNAL_DOWNLOADS]: The skill's setup and execution involve downloading and running external software.
It requires downloading the playwright Node.js package and the Chromium browser binaries.
It also mentions downloading font files (.woff2) from Google Fonts or brand-specific domains to be embedded in the HTML file.

explainer-high-energy