rag-design-doc
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill consists entirely of Markdown instructions and documentation templates. It does not contain any executable scripts, shell commands, binary files, or external tool dependencies.
- [INDIRECT_PROMPT_INJECTION]: The skill guides the creation of systems that process external corpora and user queries, which inherently involves an ingestion surface for untrusted data. However, the skill explicitly mandates mitigation strategies such as strict grounding instructions ('answer only from context') and 'I don't know' failure modes.
- Ingestion points: SKILL.md (Corpus and Queries inputs identified for the RAG system).
- Boundary markers: Present (Section 6 includes instructions for formatting retrieved context and enforcing grounding constraints).
- Capability inventory: None (the skill itself possesses no active tools or code execution capabilities).
- Sanitization: Not applicable (the skill provides architectural guidance rather than implementation code).
Audit Metadata