prisma-cli
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for commands that process external natural language or raw SQL, creating potential injection surfaces if the agent uses untrusted data to populate these fields.
- Ingestion points: The
--promptflag inprisma init(referenced in SKILL.md and references/init.md) and SQL input via file or stdin inprisma db execute(referenced in references/db-execute.md). - Boundary markers: None defined within the skill instructions; the agent should treat input to these flags as untrusted.
- Capability inventory: The Prisma CLI has broad capabilities for file system access, database schema modification, and network operations.
- Sanitization: Not addressed in the documentation; requires manual sanitization by the agent or user.
Audit Metadata