llamaindex-lumify

LlamaIndex integration for Lumify — the agent-ready sports intelligence API. Load all Lumify tools (schedules, live scores, odds, public betting splits, and explainable AI bet confidence across 8+ sports) into any LlamaIndex agent in one call.

LlamaIndex already speaks MCP natively via llama-index-tools-mcp (BasicMCPClient + McpToolSpec), so this package is a thin convenience layer: it wires up Lumify's auth headers and turns Lumify's raw 401/402 transport failures into readable tool output instead of a crashed agent run.

Install

pip install llamaindex-lumify
# Agent examples also need:
pip install llama-index

Or install the latest from source:

pip install "git+https://github.com/lumifyai/lumify.git#subdirectory=clients/llamaindex-lumify"

Get a key

Grab a free instant key in seconds — no signup, email, or card at https://lumify.ai/docs/ai (100 credits, 14-day expiry). For a persistent account with 1,000 starter credits, use https://lumify.ai/api-keys.

Set it as LUMIFY_API_KEY, or pass it explicitly. get_lumify_tools requires a key by default so the first tool call does not fail with an opaque 401. Pass require_api_key=False only if you want to introspect the public tools/list catalog. Raw lmfy-... values and Bearer lmfy-... copy-pastes are both accepted.

Quick start

import asyncio
from llamaindex_lumify import get_lumify_tools
from llama_index.core.agent.workflow import FunctionAgent
from llama_index.llms.openai import OpenAI


async def main():
    tools = await get_lumify_tools()  # reads LUMIFY_API_KEY
    agent = FunctionAgent(tools=tools, llm=OpenAI(model="gpt-4.1"))
    result = await agent.run("What's the best MLB bet today, with the rationale?")
    print(result)


asyncio.run(main())

Pass a key explicitly or point at a self-hosted endpoint:

tools = await get_lumify_tools(api_key="lmfy-...", url="https://lumify.ai/mcp")

Tool spec

from llamaindex_lumify import LumifyToolSpec

tool_spec = LumifyToolSpec()          # sync init; reads LUMIFY_API_KEY
tools = await tool_spec.to_tool_list_async()
agent = FunctionAgent(tools=tools, llm=OpenAI(model="gpt-4.1"))

LumifyToolSpec is a McpToolSpec subclass, so allowed_tools, global_partial_params, partial_params_by_tool, and include_resources all work as documented in llama-index-tools-mcp.

Tools

Loads the full Lumify MCP tool surface (18 tools): list_sports, list_seasons, list_events, get_event, batch_get_events, query_events, get_live_score, get_odds, get_odds_history, get_stats, get_splits, get_intelligence, list_teams, get_team, search_players, get_player, get_player_events, and estimate_cost (free pre-call credit-cost estimate).

Links

• Docs / instant key: https://lumify.ai/docs/ai
• MCP guide: https://lumify.ai/docs/guides#mcp
• API reference: https://lumify.ai/docs/reference

License

MIT