database-schema-designer
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No patterns indicative of system prompt extraction, safety filter bypasses, or instruction overrides were found. The skill reinforces safety by providing rules for destructive database operations.
- [DATA_EXFILTRATION]: No evidence of hardcoded credentials, sensitive file path access (such as .ssh or .aws), or unauthorized network operations was detected.
- [REMOTE_CODE_EXECUTION]: The skill does not download or execute remote scripts. It provides declarative knowledge and examples for the agent to follow.
- [COMMAND_EXECUTION]: While the skill contains SQL and shell command examples (e.g., CREATE TABLE, EXPLAIN ANALYZE), these are presented as documentation for the agent's output and are not executed by the skill itself.
- [INDIRECT_PROMPT_INJECTION]: The skill represents a low-risk surface for indirect injection as it processes user-provided domain models and queries. It provides a structured methodology for processing this data into schema designs, which minimizes the likelihood of malicious instructions in the input influencing agent behavior beyond the intended design task.
Audit Metadata