The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses python -c to execute scraping and transcription logic via the command line. This involves the agent inserting user-provided profile URLs and limits into a predefined code template.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it processes external content (titles and transcripts) from TikTok. Ingestion points: Video metadata and transcript text are read from the transcripts/ directory during the summarization step. Boundary markers: None are present to help the agent distinguish between data to be summarized and potential instructions within that data. Capability inventory: The agent can read/write files and execute project-related shell commands. Sanitization: No validation or sanitization of the scraped transcript content is performed before processing.

bulk