aurora dsql
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes externally-sourced SQL content (such as database migrations, schema dumps, and user-provided queries) to perform database operations. This establishes a surface for indirect prompt injection where instructions could theoretically be embedded in the data, though the skill implements robust sanitization.
- Ingestion points: User-provided SQL statements and externally sourced DDL files processed in Workflow 2 and Workflow 7.
- Boundary markers: The skill does not explicitly define specific text delimiters or 'ignore' instructions for the LLM when processing SQL strings in the top-level SKILL.md.
- Capability inventory: High-privilege database operations are available via the
transactMCP tool and cluster management scripts usingpsql, as detailed in the MCP Tools and CLI Scripts sections. - Sanitization: The skill mandates validation of all external DDL with
dsql_lint(fix=true)and requires application queries to be built usingsafe_query.build(), providing significant protection against malicious input.
Audit Metadata