music-generation

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: HIGHDATA_EXFILTRATIONCREDENTIALS_UNSAFE
Full Analysis
  • DATA_EXFILTRATION (HIGH): Potential for arbitrary local file read and exfiltration via user-controlled parameters.
  • Evidence (scripts/suno.py): The script implements a --lyrics-file (or -f) argument. When used, the script performs an unvalidated file read: with open(args.lyrics_file, "r") as f: lyrics = f.read().
  • Exfiltration Vector: The content of the read file is then assigned to the lyrics payload and sent via a network POST request to https://api.suno.ai/v1/generations.
  • Impact: An attacker could use a prompt to trick the agent into reading sensitive system files (e.g., ~/.ssh/id_rsa, ~/.aws/credentials) and sending the contents to the external API service.
  • CREDENTIALS_UNSAFE (MEDIUM): Broad and potentially unsafe searching for environment files.
  • Evidence (scripts/suno.py, scripts/udio.py): The load_env() function searches for .env files in ~/.config/skills/, ~/, and iteratively walks up 10 parent directories from the script location.
  • Impact: Automatically searching parent directories for secrets is a risky pattern that can lead to the accidental ingestion and exposure of credentials from unrelated projects on the same filesystem.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 17, 2026, 06:27 PM