The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes system commands including yt-dlp, whisper, and ffmpeg to process downloaded media. These commands incorporate external URLs retrieved from a scraper, which is a common vector for command injection if input is manipulated.- [EXTERNAL_DOWNLOADS]: The skill downloads video files from Instagram to the local /tmp directory. It also directs the user to install several third-party software packages (yt-dlp, openai-whisper, ffmpeg) and an external MCP server to enable its core functionality.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted text from external sources.
Ingestion points: Instagram captions and transcribed audio from reels are ingested into the agent context for analysis.
Boundary markers: No delimiters or isolation instructions are provided to separate the untrusted external content from the agent's logic.
Capability inventory: The skill possesses the ability to execute shell commands and perform file operations.
Sanitization: There is no evidence of sanitization or filtering of the ingested external content before it is processed by the agent.

spy