Database Schema Reviewer
Pass
Audited by Gen Agent Trust Hub on Mar 29, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted external data in the form of database schemas and SQL files. Ingestion points: SQL DDL, Prisma schemas, Drizzle ORM definitions, and plain-text descriptions provided in the input. Boundary markers: There are no explicit boundary markers or instructions to the agent to ignore hidden instructions within comments or text of the processed data. Capability inventory: The skill's capabilities are limited to providing textual analysis and feedback to the user. It does not execute shell commands, perform network requests, or modify system files. Sanitization: No sanitization or validation of the input content is performed prior to processing.
- [NO_CODE]: The skill consists entirely of natural language instructions and markdown documentation. No executable scripts, binaries, or configuration files are provided.
Audit Metadata