The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill communicates exclusively with the official ElevenLabs API (api.elevenlabs.io). It correctly instructs the agent to handle sensitive credentials like ELEVENLABS_API_KEY using environment variables or .env files, avoiding hardcoding. Access to local configuration files is scoped to specific paths for voice profiles.
[INDIRECT_PROMPT_INJECTION]: The skill accepts user-provided text or script files as input for audio generation. This is a standard ingestion point for untrusted data; however, the data is passed directly to the speech API without being evaluated as instructions by the helper script.
[COMMAND_EXECUTION]: The skill uses a bundled Python script (scripts/generate_voice.py) to handle API interactions. The script uses the standard library's argparse for input and urllib for network requests, avoiding unsafe shell execution or dynamic command construction.

elevenlabs-tts