The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions for the installation of the belt CLI using npx and refers to installation documentation hosted on a remote GitHub repository (inference-sh/skills).
[COMMAND_EXECUTION]: The skill provides examples of shell commands used to interact with the ElevenLabs service via the belt CLI. The skill's configuration limits the agent's ability to execute commands to those starting with the belt prefix, which follows the principle of least privilege.
[INDIRECT_PROMPT_INJECTION]: The skill processes external audio URLs provided by the user.
Ingestion points: The audio parameter within the JSON input in SKILL.md serves as the entry point for untrusted data.
Boundary markers: The use of structured JSON for the CLI input provides implicit boundaries for the data.
Capability inventory: The skill is capable of executing the belt CLI tool via Bash to send data to external APIs.
Sanitization: Sanitization is not explicitly defined within the markdown instructions; security relies on the underlying CLI tool and API handling of the URL input.

elevenlabs-voice-isolator