rag-engineer
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- PROMPT_INJECTION (SAFE): The skill uses identity and behavioral guidelines to steer the AI agent toward RAG expertise. Phrases like 'Ignore generic approaches' are used strictly within the context of prioritizing the provided architectural patterns over general knowledge, which is standard for specialized agent skills.
- EXTERNAL_DOWNLOADS (SAFE): No external URLs, scripts, or package manager references were detected in the skill files.
- DATA_EXFILTRATION (SAFE): There are no network operations, hardcoded credentials, or attempts to access sensitive system files.
- COMMAND_EXECUTION (SAFE): The skill does not contain any shell commands or subprocess execution logic.
- INDIRECT_PROMPT_INJECTION (LOW): As a RAG architectural assistant, the skill is designed to analyze user-provided code and documents. While this creates an ingestion surface, the skill lacks the capabilities (network/file-write) required to facilitate an exploit. The provided reference files actually improve safety by encouraging structured data handling and sanitization (e.g., metadata filtering and contextual compression).
Audit Metadata