gladia-live-transcription

Pass

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [SAFE]: No malicious patterns or security risks were identified. The skill implements official SDK usage and standard API interaction patterns.
  • [EXTERNAL_DOWNLOADS]: The skill utilizes the gladiaio_sdk Python package, which is a vendor-owned resource provided by the skill author (gladiaio) for interacting with the transcription service.
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external audio data which is converted into text transcripts, creating a surface for potential indirect prompt injection if the output is used in agent reasoning.
  • Ingestion points: Real-time audio stream processed via WebSocket (found in SKILL.md and references/websocket-events.md).
  • Boundary markers: None identified in the provided code samples to distinguish transcripts from instructions.
  • Capability inventory: Transcripts are printed or logged in examples (found in SKILL.md).
  • Sanitization: No explicit sanitization or filtering of transcribed text content is described in the documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 22, 2026, 12:13 PM