Skills
Navigate
skills/jeremylongshore/tons-of-skills-marketplace/twinmind-performance-tuning/Gen Agent Trust Hub

twinmind-performance-tuning

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the fluent-ffmpeg Node.js library to perform audio preprocessing tasks such as noise reduction and normalization. This involves executing the ffmpeg binary with predefined filter parameters.
  • [INDIRECT_PROMPT_INJECTION]: The TranscriptCache implementation fetches data from external audioUrl links to calculate hashes for caching and to process transcriptions. This represents an ingestion point for remote data. Ingestion points: audioUrl parameter in hashAudio and transcribeWithCache functions in references/implementation.md. Boundary markers: None. Capability inventory: fetch, curl, and ffmpeg execution. Sanitization: The skill does not explicitly sanitize the source URL before fetching.
  • [SAFE]: The skill demonstrates safe handling of sensitive credentials by instructing the use of the TWINMIND_API_KEY environment variable instead of hardcoding secrets.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:48 AM