The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could potentially contain malicious instructions meant to influence the agent's behavior.
Ingestion points: The skill fetches metadata from user-provided URLs in rules/calculate-metadata.md, reads caption files in rules/display-captions.md, and parses SRT files in rules/import-srt-captions.md.
Boundary markers: There are no explicit instructions or delimiters mentioned to ensure the agent ignores any potential instructions embedded within these external data files.
Capability inventory: The skill has capabilities for command execution via npx and ffmpeg in rules/ffmpeg.md and rules/silence-detection.md, network operations via fetch in rules/voiceover.md and rules/compositions.md, and file system writes in rules/transcribe-captions.md.
Sanitization: There is no mention of sanitizing or validating the content of the external files before processing them.

remotion-best-practices