How harveyai/deep-research-starter works
Project type and purpose
• This is a lightweight single-page “full-stack” style demo application built in Python using Streamlit.
• Purpose: provide an interactive UI for sending prompts to the OpenAI Deep Research API, streaming back events, and displaying them in real time.
• It also ships sample output files (JSONL and Pydantic representations) for offline inspection.
File structure
• Top level:
• streamlit_app.py – the main application entrypoint
• pyproject.toml – Python dependency and metadata config
• assets/ – UI static assets (e.g., the screenshot)
• output_samples/ – recorded API output for reference
• README.md, LICENSE, .gitignore, .python-version
• No separate “backend” directory: the Streamlit runtime both serves the UI and performs API calls.
• No microservices; the only external dependency at runtime is the OpenAI Deep Research API.
README for additional insights
• It confirms use of Streamlit (streamlit run streamlit_app.py) and the Deep Research API.
• Mentions Pydantic for modeling the streamed events (implied by the Python repr sample).