database-schema-impl

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes user-defined entity and field descriptions to generate implementation code, which presents a surface for indirect prompt injection.\n
  • Ingestion points: User-provided database entities and field definitions collected via templates.\n
  • Boundary markers: There are no explicit delimiters or instructions to prevent the agent from following commands embedded within user data.\n
  • Capability inventory: The skill involves generating and executing shell commands (e.g., npx prisma, python manage.py) and creating/modifying schema and migration files.\n
  • Sanitization: No specific validation or escaping steps are defined for user-provided identifiers before their use in shell commands or generated source code.\n- [COMMAND_EXECUTION]: The skill uses standard development commands such as npx prisma migrate dev and python manage.py migrate. These are appropriate for the skill's stated purpose of database management.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 10:38 PM
Security Audit — agent-trust-hub — database-schema-impl