skills/prunaai/pruna-skills/whisperx/Gen Agent Trust Hub

whisperx

Pass

Audited by Gen Agent Trust Hub on Oct 25, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [NO_CODE]: The skill consists entirely of Markdown documentation and a manifest file. It provides instructions, command-line examples, and workflow guidance but contains no scripts or executable logic within the package itself.- [EXTERNAL_DOWNLOADS]: The documentation references standard installation commands for other skills using 'npx skills add'. These installations target the '@prunaai/pruna-skills' package on NPM, which is a resource owned by the skill's author ('prunaai') and is consistent with the skill's stated purpose.- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill mentions the 'REPLICATE_API_TOKEN' environment variable and provides a placeholder ('r8_...') for user configuration. It correctly instructs the agent to handle these credentials using a dedicated helper skill ('pruna-api'), which is a standard safety practice for API interactions.- [INDIRECT_PROMPT_INJECTION]: The skill documents how to pass external audio files and lyric prompts to an AI model. However, it lacks an execution environment where this could be exploited, and the documentation focuses on standard parameters for the WhisperX model hosted on Replicate.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 25, 2026, 09:19 AM