The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local Python scripts (kokoro-tts.py, elevenlabs-tts.py, compute_timings.py) and the npx hyperframes CLI. These operations are core to the skill's functionality for media processing and timing calculation.
[EXTERNAL_DOWNLOADS]: The skill fetches audio from the ElevenLabs API and may download the open-weight Kokoro TTS model (approx. 325MB) on the first run. These are legitimate downloads from well-known services associated with the skill's primary purpose.
[PROMPT_INJECTION]: There is a potential surface for indirect prompt injection as the skill processes external web content (via WebFetch) to draft scripts.
Ingestion points: External URLs and user-provided topic prompts.
Boundary markers: Absent; no explicit delimiters or ignore-instructions are used in prompt templates.
Capability inventory: Local subprocess calls (Python/Node) and file-write operations.
Sanitization: Mitigated by quality-bar instructions requiring the agent to cross-check stats and never fabricate facts.
[SAFE]: The skill correctly instructs users to manage sensitive API keys within .env files, avoiding credential hardcoding and adhering to standard security practices.

diy-yt-creator