LlamaIndex
Native binder for LlamaIndex LLMs and embedders.
The LlamaIndex binder supplies LLM and embedder objects that send their calls through DeepintShield.
What works
- Binder
llamaindex(aliasesllama-index,llama_index) with buildersllmandembedder - One-line
guard_retriever()to check retrieved chunks before they reach the model - Automatic Agentic enforcement for tool calls on governed workflows
Limits to know: There is no dedicated integration guide; the binder is documented in the SDK reference.
How to connect
Install deepintshield[llamaindex]==3.2.3, create DeepintShield.from_env(), then call shield.bind("llamaindex").llm(...) or .embedder(...).
Read the SDK providers and frameworks docs. For retrieval checks, see RAG Security.