elevenlabs-dialogue
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install dependencies from the
inference-shGitHub organization using thenpx skills addcommand. - [EXTERNAL_DOWNLOADS]: The skill references a remote installation guide hosted at
raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md. - [COMMAND_EXECUTION]: The skill requires the
Bashtool withbeltCLI access to perform authentication and run inference tasks. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted text data within the
segmentsfield to generate audio output. This is a potential injection surface, though the risk is minimized as the output is non-executable audio. - Ingestion points: The
segmentsinput in thebelt app runcommand (found in SKILL.md). - Boundary markers: None; the text is passed as raw strings within a JSON structure.
- Capability inventory: The skill utilizes the
Bashtool to interact with the externalbeltCLI. - Sanitization: No explicit sanitization or filtering of the input text is performed before it is sent to the audio generation engine.
Audit Metadata