rag-architect
Pass
Audited by Gen Agent Trust Hub on Mar 27, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: Potential surface for Indirect Prompt Injection.
- Ingestion points: External document content is ingested and processed by LLMs in code templates for chunking and context compression (e.g., in chunking-strategies.md and retrieval-optimization.md).
- Boundary markers: The provided code snippets do not demonstrate the use of delimiters or specific instructions for the agent to ignore instructions embedded within the retrieved documents.
- Capability inventory: The skill provides patterns for passing untrusted text to LLM APIs for summarization and transformation.
- Sanitization: Architectural examples for document ingestion do not include input validation or sanitization logic.
- [EXTERNAL_DOWNLOADS]: Fetches tokenizer data from the official NLTK repository using the nltk.download utility in references/chunking-strategies.md.
- [REMOTE_CODE_EXECUTION]: Loads specialized embedding model architectures from Jina AI using the trust_remote_code parameter in the transformers library within references/chunking-strategies.md.
Audit Metadata