deepgram-performance-tuning

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes external audio files and URLs to generate transcripts, creating a surface for indirect prompt injection where malicious instructions embedded in audio could influence the agent's behavior.\n
  • Ingestion points: Processes audio data from remote URLs (audioUrls in lib/parallel-transcription.ts) and local files (filePath in lib/streaming-transcription.ts).\n
  • Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the provided implementation examples.\n
  • Capability inventory: The skill frontmatter defines high-privilege tool access, including Bash, Write, and Edit, which could be exploited if an injection attack succeeds.\n
  • Sanitization: No logic is included for sanitizing or filtering transcripts before they are returned to the agent context.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 11:46 AM