The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use yt-dlp, a well-known command-line utility, to download video files for direct inspection. This is a legitimate requirement for the skill's primary function of analyzing visual pacing and first-frame promises.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it ingests and analyzes external content (video transcripts, metadata, and visual sequences) provided by users or fetched from URLs.
Ingestion points: External media files and metadata fetched via yt-dlp (referenced in SKILL.md).
Boundary markers: None explicitly defined in the skill instructions to separate content from instructions.
Capability inventory: The skill uses file-read and yt-dlp capabilities but does not include file-write, network exfiltration, or arbitrary code execution capabilities.
Sanitization: No specific sanitization or escaping of external content is mentioned.
Assessment: While the ingestion of untrusted data is a known surface, it is fundamental to the skill's purpose (content analysis), and the risk is considered low given the lack of dangerous downstream capabilities.
[SAFE]: No obfuscation, prompt injection bypasses, hardcoded credentials, or persistence mechanisms were detected in the skill files. The repository follows standard layout and installation practices for its platform.

viral-analysis