backend-dev-guidelines
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documents an architecture that ingests untrusted data from external sources (SDKs, public REST APIs, and UI requests) and processes it through services that interact with databases and LLMs.
- Ingestion points: Public REST API endpoints located in
web/src/pages/api/public/and tRPC procedures inweb/src/features/*/server/(referenced in SKILL.md and routing-and-controllers.md). - Boundary markers: The instructions explicitly mandate the use of Zod v4 for input validation at all entry points to enforce schemas and reduce injection risks.
- Capability inventory: The system possesses capabilities to perform database operations (Prisma/ClickHouse), manage asynchronous tasks via BullMQ/Redis, and interact with external LLM providers (referenced in shared package services).
- Sanitization: The documentation emphasizes Zod parsing and mentions shared utilities like
StringNoHTMLto sanitize inputs before processing.
Audit Metadata