LlamaIndex Adapters

Two adapters for LlamaIndex: one for LLM calls and one for QueryEngine (RAG) queries.

Installation

pip install 'chengeta-ai[llamaindex]'

---

LlamaIndexLLMCacheAdapter

Drop-in replacement for any LlamaIndex LLM. Caches complete(), chat(), and their async variants.

from llama_index.llms.openai import OpenAI
from chengeta_ai import CacheManager, InMemoryBackend, CacheKeyBuilder
from chengeta_ai.adapters.llamaindex_adapter import LlamaIndexLLMCacheAdapter

llm = OpenAI(model="gpt-4o")
manager = CacheManager(
    backend=InMemoryBackend(),
    key_builder=CacheKeyBuilder(namespace="myapp"),
)
cached_llm = LlamaIndexLLMCacheAdapter(llm, manager)

# Use anywhere a LlamaIndex LLM is accepted
response = cached_llm.complete("What is retrieval-augmented generation?")
response = cached_llm.complete("What is retrieval-augmented generation?")  # cache hit

# Async
response = await cached_llm.acomplete("Explain vector embeddings")

# Chat
from llama_index.core.llms import ChatMessage
messages = [ChatMessage(role="user", content="What is RAG?")]
response = cached_llm.chat(messages)

With Settings

from llama_index.core import Settings

Settings.llm = cached_llm  # all LlamaIndex components use the cached LLM

---

LlamaIndexQueryCacheAdapter

Caches QueryEngine.query() results — ideal for RAG pipelines where the same questions recur.

from llama_index.core import VectorStoreIndex, SimpleDirectoryReader
from chengeta_ai.adapters.llamaindex_adapter import LlamaIndexQueryCacheAdapter

documents = SimpleDirectoryReader("data/").load_data()
index = VectorStoreIndex.from_documents(documents)
engine = index.as_query_engine()

cached_engine = LlamaIndexQueryCacheAdapter(engine, manager)

response = cached_engine.query("What are the key findings?")
response = cached_engine.query("What are the key findings?")  # cache hit — no retrieval, no LLM call

# Async
response = await cached_engine.aquery("Summarise section 3")

---

API Reference

LlamaIndexLLMCacheAdapter

Method	Description
complete(prompt, **kwargs)	Cached completion
acomplete(prompt, **kwargs)	Async cached completion
chat(messages, **kwargs)	Cached chat
achat(messages, **kwargs)	Async cached chat
stream_complete(...)	Passthrough (not cached)
stream_chat(...)	Passthrough (not cached)

LlamaIndexQueryCacheAdapter

Method	Description
query(query_str, **kwargs)	Cached query
aquery(query_str, **kwargs)	Async cached query