The Agent Skills Directory

[SAFE]: No security issues were detected. The skill performs localized text processing on transcript files provided by the user.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external transcript files (SRT, VTT, or Whisper JSON), presenting a theoretical surface for indirect prompt injection. 1. Ingestion points: The load function in deadair.py reads content from a file path passed via command-line arguments. 2. Boundary markers: None. 3. Capability inventory: The skill is limited to reading files and printing to standard output; it cannot write files or access the network. 4. Sanitization: The script extracts content based on specific timestamp regex patterns and truncates filler text snippets to 48 characters, which limits the potential for large instruction payloads to be processed by the agent.
[DATA_EXPOSURE]: The script reads local files to extract transcript data. The load function opens files provided as arguments. While this could be used to attempt reading sensitive files, the script's strict parsing requirements for subtitle timestamps act as a functional filter, ensuring that non-transcript data is not displayed or processed.

yt-edit