The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides patterns for downloading and installing external assets and binaries necessary for its functionality.\n
rules/transcribe-captions.md utilizes the @remotion/install-whisper-cpp package to download the Whisper.cpp binary and speech-to-text models for local transcription.\n
rules/sfx.md references external sound assets hosted on a third-party GitHub repository (github.com/kapishdima/soundcn).\n- [COMMAND_EXECUTION]: Several rules document the execution of system commands and local scripts to facilitate video rendering and processing.\n
SKILL.md describes a rendering workflow that executes a local script at /root/code/render-video-modal.ts.\n
rules/ffmpeg.md and rules/transcribe-captions.md provide instructions for running ffmpeg, ffprobe, and Node.js-based scripts using bun and bunx.\n- [REMOTE_CODE_EXECUTION]: The skill includes logic for external service communication and dynamic code bundling.\n
rules/voiceover.md contains a script for generating audio by making network requests to the ElevenLabs TTS API.\n
scripts/render.mjs uses @remotion/bundler to dynamically compile and bundle the React project at runtime during the rendering process.\n- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its data-driven nature.\n
Ingestion points: The skill fetches and processes data from external URLs in calculateMetadata (rules/calculate-metadata.md), parses local caption files (rules/display-captions.md), and loads remote Lottie JSON assets (rules/lottie.md).\n
Boundary markers: No explicit boundary markers or "ignore" instructions were found in the provided processing templates.\n
Capability inventory: The skill has the ability to write files (fs.writeFileSync), perform network operations (fetch), and execute shell commands (bunx).\n
Sanitization: There is no explicit sanitization or validation logic described for external content before it is interpolated into component props or processed.

remotion