A full toolkit for running an AI agent service built with LangGraph, FastAPI and Streamlit.
It includes a LangGraph agent, a FastAPI service to serve it, a client to interact with the service, and a Streamlit app that uses the client to provide a chat interface. Data structures and settings are built with Pydantic.
This project offers a template for you to easily build and run your own agents using the LangGraph framework. It demonstrates a complete setup from agent definition to user interface, making it easier to get started with LangGraph-based projects by providing a full, robust toolkit.
🎥 Watch a video walkthrough of the repo and app
Run directly in python
# At least one LLM API key is required
echo 'OPENAI_API_KEY=your_openai_api_key' >> .env
# uv is recommended but "pip install ." also works
pip install uv
uv sync --frozen
# "uv sync" creates .venv automatically
source .venv/bin/activate
python src/run_service.py
# In another shell
source .venv/bin/activate
streamlit run src/streamlit_app.py
Run with docker
echo 'OPENAI_API_KEY=your_openai_api_key' >> .env
docker compose watch
- LangGraph Agent: A customizable agent built using the LangGraph framework.
- FastAPI Service: Serves the agent with both streaming and non-streaming endpoints.
- Advanced Streaming: A novel approach to support both token-based and message-based streaming.
- Content Moderation: Implements LlamaGuard for content moderation (requires Groq API key).
- Streamlit Interface: Provides a user-friendly chat interface for interacting with the agent.
- Multiple Agent Support: Run multiple agents in the service and call by URL path
- Asynchronous Design: Utilizes async/await for efficient handling of concurrent requests.
- Feedback Mechanism: Includes a star-based feedback system integrated with LangSmith.
- Dynamic Metadata:
/info
endpoint provides dynamically configured metadata about the service and available agents and models. - Docker Support: Includes Dockerfiles and a docker compose file for easy development and deployment.
- Testing: Includes robust unit and integration tests for the full repo.
The repository is structured as follows:
src/agents/
: Defines several agents with different capabilitiessrc/schema/
: Defines the protocol schemasrc/core/
: Core modules including LLM definition and settingssrc/service/service.py
: FastAPI service to serve the agentssrc/client/client.py
: Client to interact with the agent servicesrc/streamlit_app.py
: Streamlit app providing a chat interfacetests/
: Unit and integration tests
AI agents are increasingly being built with more explicitly structured and tightly controlled Compound AI Systems, with careful attention to the cognitive architecture. At the time of this repo's creation, LangGraph seems like the most advanced open source framework for building such systems, with a high degree of control as well as support for features like concurrent execution, cycles in the graph, streaming results, built-in observability, and the rich ecosystem around LangChain.
I've spent a decent amount of time building with LangChain over the past year and experienced some of the commonly cited pain points. In building this out with LangGraph I found a few similar issues, but overall I like the direction and I'm happy with my choice to use it.
With that said, there are several other interesting projects in this space that are worth calling out, and I hope to spend more time building with them soon:
- LlamaIndex Workflows and llama-agents: LlamaIndex Workflows launched the day I started working on this. I've generally really liked the experience building with LlamaIndex and this looks very promising.
- DSPy: The DSPy optimizer and approach also seems super interesting and promising. But the creator has stated they aren't focusing on agents yet. I will probably experiment with building some of the specific nodes in more complex agents using DSPy in the future.
- I know there are more springing up regularly, such as I recently came across Prefect ControlFlow.
-
Clone the repository:
git clone https://github.com/JoshuaC215/agent-service-toolkit.git cd agent-service-toolkit
-
Set up environment variables: Create a
.env
file in the root directory. At least one LLM API key or configuration is required. See the.env.example
file for a full list of available environment variables, including a variety of model provider API keys, header-based authentication, LangSmith tracing, testing and development modes, and OpenWeatherMap API key. -
You can now run the agent service and the Streamlit app locally, either with Docker or just using Python. The Docker setup is recommended for simpler environment setup and immediate reloading of the services when you make changes to your code.
This project includes a Docker setup for easy development and deployment. The compose.yaml
file defines two services: agent_service
and streamlit_app
. The Dockerfile
for each is in their respective directories.
For local development, we recommend using docker compose watch. This feature allows for a smoother development experience by automatically updating your containers when changes are detected in your source code.
-
Make sure you have Docker and Docker Compose (>=2.23.0) installed on your system.
-
Build and launch the services in watch mode:
docker compose watch
-
The services will now automatically update when you make changes to your code:
- Changes in the relevant python files and directories will trigger updates for the relevantservices.
- NOTE: If you make changes to the
pyproject.toml
oruv.lock
files, you will need to rebuild the services by runningdocker compose up --build
.
-
Access the Streamlit app by navigating to
http://localhost:8501
in your web browser. -
The agent service API will be available at
http://localhost:80
. You can also use the OpenAPI docs athttp://localhost:80/redoc
. -
Use
docker compose down
to stop the services.
This setup allows you to develop and test your changes in real-time without manually restarting the services.
You can also run the agent service and the Streamlit app locally without Docker, just using a Python virtual environment.
-
Create a virtual environment and install dependencies:
pip install uv uv sync --frozen source .venv/bin/activate
-
Run the FastAPI server:
python src/run_service.py
-
In a separate terminal, run the Streamlit app:
streamlit run src/streamlit_app.py
-
Open your browser and navigate to the URL provided by Streamlit (usually
http://localhost:8501
).
The agent supports LangGraph Studio, a new IDE for developing agents in LangGraph.
You can simply install LangGraph Studio, add your .env
file to the root directory as described above, and then launch LangGraph studio pointed at the root directory. Customize langgraph.json
as needed.
Currently the tests need to be run using the local development without Docker setup. To run the tests for the agent service:
-
Ensure you're in the project root directory and have activated your virtual environment.
-
Install the development dependencies and pre-commit hooks:
pip install uv uv sync --frozen pre-commit install
-
Run the tests using pytest:
pytest
To customize the agent for your own use case:
- Add your new agent to the
src/agents
directory. You can copyresearch_assistant.py
orchatbot.py
and modify it to change the agent's behavior and tools. - Import and add your new agent to the
agents
dictionary insrc/agents/agents.py
. Your agent can be called by/<your_agent_name>/invoke
or/<your_agent_name>/stream
. - Adjust the Streamlit interface in
src/streamlit_app.py
to match your agent's capabilities.
The repo includes a generic src/client/client.AgentClient
that can be used to interact with the agent service. This client is designed to be flexible and can be used to build other apps on top of the agent. It supports both synchronous and asynchronous invocations, and streaming and non-streaming requests.
See the src/run_client.py
file for full examples of how to use the AgentClient
. A quick example:
from client import AgentClient
client = AgentClient()
response = client.invoke("Tell me a brief joke?")
response.pretty_print()
# ================================== Ai Message ==================================
#
# A man walked into a library and asked the librarian, "Do you have any books on Pavlov's dogs and Schrödinger's cat?"
# The librarian replied, "It rings a bell, but I'm not sure if it's here or not."
Contributions are welcome! Please feel free to submit a Pull Request.
- Get LlamaGuard working for content moderation (anyone know a reliable and fast hosted version?)
- Add more sophisticated tools for the research assistant
- Increase test coverage and add CI pipeline
- Add support for multiple agents running on the same service, including non-chat agent
- Service metadata endpoint
/info
and dynamic app configuration - More ideas? File an issue or create a discussion!
This project is licensed under the MIT License - see the LICENSE file for details.