rag-pipeline-review
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided RAG configurations and code snippets, which constitutes an ingestion surface for untrusted data. Although it lacks dangerous capabilities like network access or command execution, it is theoretically susceptible to embedded instructions within the analyzed code.
- Ingestion points: Analyzed code and configuration files provided by the user in the agent's context (SKILL.md).
- Boundary markers: None. The prompt does not define specific delimiters for the untrusted data.
- Capability inventory: No subprocess calls, exec/eval, file-write, or network operations are requested or used.
- Sanitization: None. The skill relies on the underlying model's safety guardrails when processing input.
- [SAFE]: The skill consists entirely of natural language instructions. No code, scripts, or external dependencies are included. No obfuscation, data exfiltration patterns, or persistence mechanisms were detected.
Audit Metadata