The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill correctly handles sensitive information by instructing users to store the ELEVENLABS_API_KEY in environment variables rather than hardcoding it. The worker scripts use os.environ to retrieve the key and explicitly avoid logging or printing it.
[COMMAND_EXECUTION]: Uses subprocess.run to invoke ffmpeg for audio conversion. The implementation is secure as it uses a fixed argument list with shell=False. Furthermore, output paths are strictly validated to ensure they are absolute, regular files, and not symlinks, mitigating path traversal or overwrite attacks.
[EXTERNAL_DOWNLOADS]: Connectivity is limited to the official ElevenLabs API (api.elevenlabs.io). The skill utilizes Python's standard library (urllib.request) for network operations and requires no external third-party packages, reducing the supply chain attack surface.
[DATA_EXFILTRATION]: No evidence of unauthorized data transmission was found. Network requests are strictly functional (TTS synthesis and voice listing) and do not access or transmit sensitive system files or local configuration data.
[PROMPT_INJECTION]: The skill's instructions are focused on configuration and operation. There are no attempts to override agent behavior, bypass safety guardrails, or solicit sensitive information through instructional patterns.

narova-elevenlabs