The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references and encourages the installation of the belt-sh/cli tool via npx and fetches installation guidance from the vendor's GitHub repository at https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md. These resources originate from the associated service provider for the skill's functionality.
[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the belt CLI. These commands include authentication (belt login) and application execution (belt app run) to perform audio processing tasks. This behavior is confined to the scope defined in the allowed-tools frontmatter.
[INDIRECT_PROMPT_INJECTION]: The skill processes external audio and video content via URLs. This represents an attack surface where malicious instructions embedded in audio could potentially influence the agent's behavior during transcription or in downstream tasks like caption generation.
Ingestion points: The audio input parameter in belt app run commands (SKILL.md).
Boundary markers: None explicitly defined in the provided bash commands or instructions.
Capability inventory: Shell command execution via the belt CLI tool (SKILL.md).
Sanitization: No explicit sanitization of transcribed text before it is used in downstream workflows (e.g., passing to infsh/caption-videos) is documented.

elevenlabs-stt