hifi-download-skill

Pass

Audited by Gen Agent Trust Hub on Feb 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONCREDENTIALS_UNSAFEDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION] (LOW): The run.sh script executes Python files using a positional parameter (scripts/$1.py). This design lacks validation and could theoretically be exploited for path traversal or execution of unintended scripts if an agent handles unvalidated input.
  • [CREDENTIALS_UNSAFE] (LOW): The setup_config.py script accepts sensitive information, including Qobuz passwords and Spotify secrets, as plaintext command-line arguments. This may result in credentials being stored in shell history or captured in agent execution logs.
  • [DATA_EXFILTRATION] (LOW): The skill performs network operations to non-whitelisted domains such as Last.fm and Qobuz. Notably, communication with the Last.fm API (http://ws.audioscrobbler.com) is performed over unencrypted HTTP, exposing user metadata to potential interception.
  • [INDIRECT_PROMPT_INJECTION] (LOW): An attack surface exists in scripts/lastfm_taste.py, which processes music metadata (artist and track names) from external APIs. Ingestion Point: scripts/lastfm_taste.py; Boundary Markers: Absent; Capability Inventory: run.sh (exec), platform_download.py (file-write); Sanitization: Absent. While unlikely to lead to execution, malicious metadata could be used to influence the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 20, 2026, 04:52 AM