Streamlit Chat UI

Every app-mode project includes a production-ready chat interface with dynamic model switching.

Quick start

cd my-project-app
pip install -r requirements.txt

# Start the API server (terminal 1)
uvicorn server:app --reload --port 8080

# Start the chat UI (terminal 2)
streamlit run streamlit_app.py

The chat UI opens at http://localhost:8501.

Features

Glassmorphism dark theme
Styled with backdrop blur and gradient accents
Dynamic model switching
Auto-detects API keys from .env and shows available models
Guard summary sidebar
Displays guardrail results for each response
Chat history
Full conversation history with user/assistant messages

Dynamic model switching

The chat UI reads your .env file and auto-detects which API keys are present. Based on that, it populates a model picker dropdown:

If this key exists in .envThese models appear
OPENAI_API_KEYgpt-4o, gpt-4o-mini, gpt-4.1, o1
ANTHROPIC_API_KEYclaude-sonnet-4, claude-haiku-4.5
GOOGLE_API_KEYgemini-2.0-flash, gemini-2.5-pro

You can add multiple API keys to get access to models from multiple providers in the same chat UI.

Changing the default model

Set the default model in your .env file:

DEFAULT_MODEL=claude-sonnet-4-20250514

This sets what model is pre-selected when the chat UI loads. Users can still switch models from the dropdown at any time.

Customizing the UI

The chat UI is a single Python file (streamlit_app.py) that you can customize:

  • Change the page title and icon in st.set_page_config()
  • Modify the CSS in the st.markdown() block at the top
  • Add custom sidebar content
  • Change the API URL if your server runs on a different port

Running with agentvoy dev

Instead of starting two terminals manually, use:

agentvoy dev

This starts both the FastAPI server and the Streamlit UI, plus enables the DevTools dashboard at /dev.