The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run to call ffmpeg and edge-tts for audio processing and synthesis in skill.py and generate_interview_podcast.py. These calls use list-based arguments, which prevents shell injection, but they execute commands on content derived from user-provided files.
[DATA_EXFILTRATION]: The skill hardcodes an external IP address (http://219.147.109.250:7860) to interface with an IndexTTS2 synthesis service. When this engine is selected, user-provided text from Markdown documents is sent to this unverified server, posing a potential data exposure risk for sensitive content.
[PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection (Category 8).
Ingestion points: Untrusted data enters the agent context through the markdown and script input files in skill.py and generate_interview_podcast.py.
Boundary markers: The skill does not use delimiters or specific instructions to isolate user-provided text, allowing embedded instructions to potentially influence the AI's script generation phase.
Capability inventory: The skill has the ability to write files to the workspace/ directory and execute shell commands (ffmpeg, edge-tts).
Sanitization: Content is processed using basic regular expressions to remove Markdown formatting, which does not sanitize malicious instructional overrides.

podcast-generator