fullstack-guardian
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes natural language feature requirements (Ingestion points: triggers in SKILL.md) to generate full-stack code (Capability inventory: File writing via agent implementation). Boundary markers are explicitly defined in the constraints section, requiring the agent to design features using a three-perspective approach (Frontend, Backend, Security). Sanitization is implemented via strict instructions to use parameterized queries, output encoding, and schema-based validation (Zod/Pydantic).
- [SAFE]: Technical reference files contain industry-standard code patterns for authentication, error handling, and performance optimization. All configuration templates use placeholders for environment variables, following secure development practices.
Audit Metadata