vector-db-cleanup

Pass

Audited by Gen Agent Trust Hub on Apr 3, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill performs routine maintenance on a local ChromaDB instance. It does not exhibit any malicious patterns or unauthorized access behaviors.
  • [COMMAND_EXECUTION]: Shell commands are restricted to executing local Python scripts and performing local service health checks via curl. No dynamic command generation from untrusted input was observed.
  • [DATA_EXFILTRATION]: All network activity is confined to the local loopback interface (127.0.0.1) for database communication. No data is sent to external servers.
  • [PROMPT_INJECTION]: The instructions are focused on operational tasks and include clear safety guidelines for the agent, such as requiring user confirmation before destructive actions.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 3, 2026, 06:09 PM