haystack
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The RAG pipeline in scripts/basic_rag_pipeline.py presents an attack surface where malicious instructions embedded in documents could potentially influence the agent.
- Ingestion points: Documents are written to the InMemoryDocumentStore in scripts/basic_rag_pipeline.py.
- Boundary markers: The prompt template used by ChatPromptBuilder does not utilize distinct delimiters or XML tags to separate document content from instructions.
- Capability inventory: The pipeline enables the OpenAIChatGenerator to process and generate responses based on the retrieved context.
- Sanitization: No sanitization or validation of the retrieved document content is performed before its interpolation into the prompt template.
Audit Metadata