sql-optimization-patterns
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze SQL queries. Because the
EXPLAIN ANALYZEcommand executes the provided statement to collect statistics, it could trigger side effects if the query invokes database functions or triggers. - Ingestion points: User-provided SQL queries and table schemas analyzed for performance (SKILL.md).
- Boundary markers: Instructions mandate confirming the permitted environment and stopping for clarification if safety boundaries are missing (SKILL.md).
- Capability inventory: SQL execution capabilities, specifically
EXPLAIN ANALYZEand database maintenance commands (implementation-playbook.md). - Sanitization: The skill provides clear instructions to use bound parameters ("Values are bound, never interpolated") and to use authorized test databases with representative data rather than production environments.
- [SAFE]: The skill incorporates multiple security-best-practice instructions. It requires starting with read-only access and obtaining explicit authorization before performing DDL, data changes, or maintenance. It also warns that monitoring tools like
pg_stat_statementsmay expose sensitive values in query text, advising against unauthorized export. External documentation links target official PostgreSQL resources.
Audit Metadata