The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The installation documentation includes a shell script installer command (curl | sh) that fetches and executes a setup script from the official ElevenLabs GitHub repository for their CLI tool.
[EXTERNAL_DOWNLOADS]: The skill downloads official dependencies from NPM and PyPI registries and references audio samples hosted on Google Cloud Storage for demonstration purposes.
[COMMAND_EXECUTION]: Instructions guide the user to execute commands for package managers such as npm, pip, brew, and scoop, as well as the elevenlabs CLI for performing audio conversions.
[INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect instructions by processing audio data from external sources.
Ingestion points: Untrusted audio data is ingested through local file paths (source.mp3) and remote URLs specified in SKILL.md examples.
Boundary markers: No explicit delimiters or instructions are provided to the agent to distinguish between audio processing data and its own operational instructions.
Capability inventory: The skill possesses capabilities to write files to the local system, execute shell commands via the CLI, and perform network operations to the ElevenLabs API.
Sanitization: There is no evidence of sanitization, filtering, or validation of audio content prior to processing.

elevenlabs-voice-changer