knowledge-base-rag
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill implements a Retrieval-Augmented Generation (RAG) pipeline, which creates a standard attack surface where malicious instructions could be embedded in the documents being processed.
- Ingestion points: External data enters the system through the
ingestmethod inSKILL.md, which chunks and embeds text for later retrieval. - Boundary markers: The prompt template uses explicit markers such as 'Context:' and 'Question:', along with grounding instructions ('If the context does not contain the answer, say so explicitly') to help the model distinguish between data and instructions.
- Capability inventory: The implementation provided in the skill is limited to text processing and embedding; it does not include tools for file system modification, network requests beyond API calls to the embedder/LLM, or shell command execution.
- Sanitization: The code focuses on structural chunking to preserve context but does not include explicit filtering or sanitization of the input text contents.
Audit Metadata