sql
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides structured guidance for SQL development and verification without any detected malicious patterns. It explicitly advocates for security best practices like parameter binding and identifier allowlisting.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data such as SQL schemas, existing queries, and database migrations. While this represents a potential attack surface, the instructions prioritize defensive coding and thorough verification.
- Ingestion points: The skill reads external SQL files, schema definitions, and migration history as described in SKILL.md.
- Boundary markers: There are no explicit instructions for the agent to use delimiters when processing this external content.
- Capability inventory: The skill allows the agent to reason about, generate, and propose modifications to SQL queries and database schemas.
- Sanitization: The skill contains explicit instructions to bind values through the driver and use allowlists for dynamic identifiers to prevent injection attacks.
Audit Metadata