The Agent Skills Directory

[OBFUSCATION]: Detected a zero-width character (U+2060 Word Joiner) embedded within a string literal in the file rules/maps.md. This character is positioned immediately before the 'mapbox' protocol in the string "⁠mapbox://styles/mapbox/standard".
[INDIRECT_PROMPT_INJECTION]: The skill exhibits vulnerability surfaces for indirect prompt injection by ingesting and processing untrusted data from external sources.
Ingestion points: Data enters the agent's context through fetch() calls for external JSON files (e.g., Lottie animations in rules/lottie.md, metadata in rules/calculate-metadata.md) and by parsing user-provided .srt subtitle files in rules/import-srt-captions.md.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are present when processing these external data sources.
Capability inventory: The skill utilizes potentially dangerous capabilities including shell command execution via remotion ffmpeg and Node.js execSync (referenced in rules/transcribe-captions.md), along with network operations (fetch) and file system writes (rules/voiceover.md).
Sanitization: There is no evidence of sanitization, validation, or escaping of the content retrieved from external URLs or files before it is processed by the agent or rendered in the composition.

remotion-best-practices