See how AT&T cut coding costs by 56% with LiteLLM.

LiteAgents SDK

Switch agent harnesses
without rewriting your agent.

An interface modeled after the Claude Agent SDK, with your choice of harness. Keep your model, tools, and MCP. Add native controls and optional Temporal durability.

Start in Colab ↗

Python 3.11+ · Install only the harnesses you use.

import asyncio
from liteagents import (
    query, LiteAgentOptions, ProfileOptions,
)

profile = ProfileOptions(
    harness="deepagents",
    model="openai/gpt-5.4-mini",
)

async def main():
    async for message in query(
        prompt="What can you help me with?",
        options=LiteAgentOptions(profile=profile),
    ):
        print(message)

asyncio.run(main())
Set your provider API key and install the selected integration. Setup guide ↗
LiteAgents / query()
interactive demo

Check the status
of order A123.

your model
Deep Agents
harness="deepagents"
your tools + MCP

Order A123
is on its way.

Illustrative runs. Switching harnesses starts a new run.Try it in Colab ↗
Data centre server hall
08

Run it yourself. Start today.

Deploy the open-source gateway in an afternoon, with spend tracked and capped on every request. Add SSO, audit logs, and an SLA when it goes org-wide.