sql-query-generator
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the mcp__kalysta__executeQuery tool to run SQL commands. It attempts to limit risks by instructing the agent to only use read-only keywords and avoiding data-modifying statements.
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection when analyzing database records. Data ingestion occurs from tables storing user-provided prompts and logs, which may contain malicious instructions. The skill lacks programmatic sanitization and relies on the agent's adherence to natural language safety boundaries to ignore embedded commands.
Audit Metadata