The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs users to install the belt-sh/cli package via npx and references external installation guides from raw.githubusercontent.com/inference-sh/skills. These sources are not verified vendors.
[REMOTE_CODE_EXECUTION]: The skill workflow is centered around the belt app run command, which downloads and executes logic for transcription and alignment models from the inference.sh platform at runtime.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute shell commands using the belt CLI. This involves authenticated sessions (belt login) and execution of applications that interact with both local files and remote URLs.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8).
- Ingestion points: Audio data via external URLs and narration text provided in the input parameters of the elevenlabs-stt and elevenlabs-forced-alignment tools.
- Boundary markers: None identified; the skill does not instruct the agent to ignore instructions embedded within the transcribed text or input data.
- Capability inventory: The skill has access to Bash and can write output to files (e.g., transcript.json).
- Sanitization: No evidence of escaping or filtering logic is present for the data processed by the external transcription models.

elevenlabs-stt