AG-UI LangGraph Research Agent

June 18, 2025 · View on GitHub

A FastAPI-based research agent built with LangGraph that provides intelligent web research capabilities through the AG-UI protocol. This agent can perform web searches, analyze content, and generate comprehensive research reports.

Features

Intelligent Research: Automated web searching and content analysis
LangGraph Workflow: State-managed research process with clear execution flow
AG-UI Protocol: Standardized communication interface with real-time streaming
FastAPI Backend: High-performance asynchronous web API
Real-time Updates: Server-sent events for live progress tracking

Architecture

The project is organized into modular components:

src/agui/
├── main.py              # FastAPI application and AG-UI endpoint
├── langgraph/
│   ├── agent.py         # LangGraph workflow builder
│   ├── state.py         # Research state management
│   ├── research.py      # Core research logic
│   ├── report.py        # Report generation
│   └── web_search.py    # Web search functionality

Prerequisites

Python 3.10 or higher (up to 3.13)
Poetry for dependency management
OpenAI API key
SerpAPI key (for web search)

Installation

Clone the repository:

git clone git@github.com:TheGreatBonnie/ag-ui-langgraph-agent.git
cd ag-ui-langgraph-agent

Install dependencies using Poetry:
```
poetry install
```

Set up environment variables: Create a .env file in the root directory:

OPENAI_API_KEY=your_openai_api_key_here
SERPAPI_API_KEY=your_serpapi_key_here

Usage

Running the Server

Activate the virtual environment:
```
poetry shell
```
Start the FastAPI server:
```
poetry run uvicorn src.agui.main:app
```
The API will be available at:
- Main endpoint: http://localhost:8000/
- Interactive docs: http://localhost:8000/docs
- OpenAPI spec: http://localhost:8000/openapi.json

API Usage

Send a POST request to the root endpoint with a research query:

curl -X POST "http://localhost:8000/" \
  -H "Content-Type: application/json" \
  -d '{
    "thread_id": "test_thread_123",
    "run_id": "test_run_456",
    "messages": [
      {
        "id": "msg_1",
        "role": "user",
        "content": "Research lifespan of Penguins"
      }
    ],
    "tools": [],
    "context": [],
    "forwarded_props": {},
    "state": {}
  }'