redshift

Pass

Audited by Gen Agent Trust Hub on Apr 21, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses subprocess.run in lib/client.py and setup.py to execute AWS CLI commands. This is used for authenticating with AWS, discovering clusters, and executing SQL queries via the Redshift Data API.
  • [CREDENTIALS_UNSAFE]: The setup.py script accesses ~/.aws/credentials and ~/.aws/config. This access is used strictly to list available AWS profile names during the interactive setup process, allowing the user to select an existing IAM identity.
  • [PROMPT_INJECTION]: While the skill processes data from external databases (a surface for Indirect Prompt Injection), it implements a strict read-only validator in lib/client.py. This validator enforces an allowlist of SQL keywords (SELECT, WITH, SHOW, etc.) and blocks multi-statement queries to prevent unauthorized DDL or DML operations.
  • [SAFE]: No external data exfiltration was detected. The skill saves query results locally to ~/redshift-exports/ and uses standard AWS CLI mechanisms for all network interactions with AWS infrastructure.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 21, 2026, 10:05 AM