Production-grade multiservice Agent Development Kit
Project description
LiteLLM ADK (Agent Development Kit)
Highly flexible multiservice Agent Development Kit for building AI agents using LiteLLM.
Built for developers who need to swap models, API keys, and base URLs dynamically while maintaining a robust structure for tool usage, modular memory persistence, and observability.
Features
- Model Agnostic: Access 100+ LLMs (OpenAI, Anthropic, OCI Grok-3, Llama, etc.) seamlessly.
- Easy Tools: Register Python functions with the
@tooldecorator. No manual JSON schema management. - Modular Memory: Native support for conversation persistence:
InMemoryMemory: Fast, ephemeral storage.FileMemory: Simple JSON-based local persistence.MongoDBMemory: Scalable, remote persistence.
- Parallel & Sequential Execution: Built-in support for parallel tool calls with robust stream accumulation.
- Dynamic Configuration: Global defaults via
.envor per-agent/per-request overrides. - Async & Streaming: Native support for
ainvoke,stream, andastream.
Installation
pip install litellm-adk
Quick Start
Simple Conversational Agent
from litellm_adk.agents import LiteLLMAgent
from litellm_adk.memory import FileMemory
# Setup persistent memory
memory = FileMemory("chat_history.json")
agent = LiteLLMAgent(
model="gpt-4",
memory=memory,
session_id="user-123"
)
response = agent.invoke("My name is Alice.")
print(agent.invoke("What is my name?")) # Alice
Registering Tools
from litellm_adk.tools import tool
@tool
def get_weather(location: str):
"""Get the current weather for a location."""
return f"The weather in {location} is sunny."
agent = LiteLLMAgent(tools=[get_weather])
agent.invoke("What is the weather in London?")
Configuration
The ADK uses pydantic-settings. Configure via .env:
ADK_MODEL: Default model (e.g.,gpt-4o).ADK_API_KEY: Default API key.ADK_BASE_URL: Global base URL override.ADK_LOG_LEVEL: DEBUG, INFO, etc.
Documentation
License
MIT
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file litellm_adk-0.2.2.tar.gz.
File metadata
- Download URL: litellm_adk-0.2.2.tar.gz
- Upload date:
- Size: 24.3 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.13.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
ce2c33fa1f6a0cd5b0cbab0f0d02b7fc7af5c3b71bcd9973094efca6504652dc
|
|
| MD5 |
702eaaf8e61a282300f954b8f2f6e264
|
|
| BLAKE2b-256 |
468df7375e8cd07c8e29cd0624a2bfdeea8d5b111955f528fe5e61e6a933290d
|
File details
Details for the file litellm_adk-0.2.2-py3-none-any.whl.
File metadata
- Download URL: litellm_adk-0.2.2-py3-none-any.whl
- Upload date:
- Size: 29.3 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.13.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
b04547a34dc5878659d67d3392e701869ac9037cc913339a77ffba5d1f89261b
|
|
| MD5 |
9d8913d76ab39ebb5358abe45ddd52d9
|
|
| BLAKE2b-256 |
93de8edeef2e8469f08f543d6d4cfdb675240b5dc52b75c476b6d8168a7d3cef
|