demo-geocontext

Interactive demo for ignfab/geocontext based on Gradio - ChatBot and LangGraph.

Screenshot

Requirements

• uv ( Python package and project manager )
• NodeJS (npx)

Parameters

Name	Description	Default
MODEL_NAME	The name of the model (see LangGraph - create_react_agent / init_chat_model)	"anthropic:claude-sonnet-4-6"
ANTHROPIC_API_KEY	Required from anthropic:* models (https://console.anthropic.com/settings/keys)
GOOGLE_API_KEY	Required from google_genai:* models (https://aistudio.google.com/api-keys)
TEMPERATURE	Model temperature	0
DB_URI	URL of the PostgreSQL (postgresql://postgres:ChangeIt@localhost:5432/geocontext) or Redis (ex : redis://default:ChangeIt@localhost:6379/0) database.	None (use InMemorySaver)
CONTACT_EMAIL	Email for the contact button.	"dev@localhost"
GEOCONTEXT_LOG_LEVEL	Log level for Geocontext MCP.	error
LOG_LEVEL	Log level for this application.	INFO

Note that "HTTP_PROXY", "HTTPS_PROXY", "NO_PROXY" are supported if you have to use a corporate proxy.

Usage

With uv on Linux

# download repository
git clone https://github.com/ignfab/demo-geocontext
cd demo-geocontext

# configure model model and credentials
export MODEL_NAME="anthropic:claude-sonnet-4-6"
export ANTHROPIC_API_KEY="YourApiKey"

# start demo on http://localhost:8000/ :
uv run demo_gradio.py

With uv on Windows

Compared to Linux, adapt model and credentials configuration as follow with PowerShell :

#$env:MODEL_NAME="ollama:mistral:7b"
$env:MODEL_NAME="anthropic:claude-sonnet-4-6"
$env:ANTHROPIC_API_KEY="YourApiKey"

# start demo on http://localhost:8000/
uv run demo_gradio.py

With docker

See docker-compose.yaml :

# build image
docker compose build

# Use Google Gemini API
export MODEL_NAME="google_genai:gemini-2.5-flash"
export GOOGLE_API_KEY="YourApiKey"

# start demo on http://localhost:8000/
docker compose up -d

Credits

• gradio - Chatbot
• LangGraph
• langchain-mcp-adapters
• ignfab/geocontext

License

MIT