LlamaIndex

Retrieved nodes are long, and a chunk cut from its page often names no one. "He said" and "last year" reach the model with nothing to resolve them. Flux adds live search to your agent and rewrites each node as the passages that answer the query, with the names and dates filled in.

Build it

1. 1

   Install

   Shell

   pip install llama-index-core llama-index-llms-openai requestsexport FLUX_API_KEY="<your key>"

2. 2

   Give your agent web search

   A plain function is a tool. Your model plans and answers, and calls Flux for anything current.

   Python

   import asyncioimport osimport requestsFLUX = "https://fluxsearch.io/api/v1"HEADERS = {"Authorization": f"Bearer {os.environ['FLUX_API_KEY']}"}from llama_index.core.agent.workflow import FunctionAgentfrom llama_index.llms.openai import OpenAIdef web_search(query: str) -> str:    """Search the live web. Returns passages with provenance and their source URLs."""    response = requests.post(        f"{FLUX}/search",        headers=HEADERS,        json={"query": query, "max_results": 5, "enrich": 3},        timeout=120,    )    response.raise_for_status()    passages = [        f"{passage['enriched_text']} ({result['url']})"        for result in response.json()["results"]        for passage in (result["lineage"] or {}).get("results", [])    ]    return "\n\n".join(passages) or "No passages found."agent = FunctionAgent(tools=[web_search], llm=OpenAI(model="gpt-5-mini"))async def main():    print(await agent.run("Who runs Barclays?"))asyncio.run(main())

   Show all 36 lines

3. 3

   Trim retrieved nodes

   Wrap Lineage as a BaseNodePostprocessor and pass it to any query engine.

   Python

   import osimport requestsFLUX = "https://fluxsearch.io/api/v1"HEADERS = {"Authorization": f"Bearer {os.environ['FLUX_API_KEY']}"}from llama_index.core.postprocessor.types import BaseNodePostprocessorfrom llama_index.core.schema import NodeWithScore, TextNodeclass FluxLineage(BaseNodePostprocessor):    def _postprocess_nodes(self, nodes, query_bundle=None):        if query_bundle is None:            return nodes        kept = []        for node in nodes:            response = requests.post(                f"{FLUX}/provenance",                headers=HEADERS,                json={"text": node.node.get_content(), "query": query_bundle.query_str},                timeout=120,            )            response.raise_for_status()            for passage in response.json()["results"]:                text = TextNode(text=passage["enriched_text"], metadata=node.node.metadata)                kept.append(NodeWithScore(node=text, score=passage["score"]))        return keptquery_engine = index.as_query_engine(node_postprocessors=[FluxLineage()])

   Show all 30 lines

Call Flux as an LLM

For a direct answer with no agent, use OpenAILike. Flux does not call tools, so leave function calling off.

import osfrom llama_index.llms.openai_like import OpenAILikeflux = OpenAILike(    model="flux-search-1",    api_base="https://fluxsearch.io/api/v1",    api_key=os.environ["FLUX_API_KEY"],    is_chat_model=True,    is_function_calling_model=False,)print(flux.complete("Who is the chief executive of Barclays?"))

Next steps