framework

SiftQ + LlamaIndex

BaseRetriever subclass for LlamaIndex workflows.

Use SiftQ as the retrieval source in LlamaIndex query engines, agent workflows, and ReAct agents. Implements `BaseRetriever` so it composes with rerankers, postprocessors, and response synthesizers.

Install

pip

pip install llama-index-retrievers-siftq

Example

python

from llama_index.retrievers.siftq import SiftQRetriever
from llama_index.core.query_engine import RetrieverQueryEngine

retriever = SiftQRetriever(scope="scholar", size=15)
engine = RetrieverQueryEngine.from_args(retriever)
response = engine.query("Transformer architecture innovations 2025")
print(response)

Why use it

✓Native BaseRetriever — works with every LlamaIndex pattern
✓Combine with LlamaIndex rerankers (Cohere, BGE) for two-stage retrieval
✓Supports streaming retrieval in agent workflows

Source

github.com/siftq/llama-index-retrievers-siftq ↗

Try the SiftQ API playground →