The Agent Skills Directory

[DYNAMIC_EXECUTION]: The animation runtime utilizes eval() to process dynamic coordinate expressions provided via command-line arguments.\n
Evidence: template/src/core.js contains if (at) at = at.map((v) => typeof v === 'string' ? (0, eval)(v) : v); within the window.renderSheet function.\n- [DYNAMIC_EXECUTION]: The build process includes a script that programmatically modifies the source code of a dependency to adjust rendering behavior.\n
Evidence: template/patch_brush.mjs reads the p5.brush library source and replaces constants before writing a patched version.\n- [COMMAND_EXECUTION]: The rendering utility executes external binary tools to process video frames and encode final output.\n
Evidence: template/render.mjs spawns ffmpeg processes for video encoding and searches for system-wide Chrome/Chromium executables.\n- [EXTERNAL_DOWNLOADS]: The project scaffolding script installs dependencies from the NPM registry.\n
Evidence: scripts/new_project.sh executes npm install to fetch libraries like puppeteer-core and p5.\n- [INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface where inputs provided to the renderer are evaluated as code within the browser context.\n
Ingestion points: Command-line arguments in template/render.mjs (specifically --crop-at).\n
Boundary markers: None.\n
Capability inventory: The browser environment can execute arbitrary JavaScript; the rendering script can spawn shell commands (ffmpeg).\n
Sanitization: Expression strings are evaluated without prior validation.

crayon-storybook