Fetching a Streamlit app URL returns only the word 'Streamlit': read the page source or render it instead
A plain HTTP fetch of a deployed Streamlit app (WebFetch, curl, requests) returns HTTP 200 with an HTML shell whose only visible text is 'Streamlit'. Seen on a Streamlit app served behind Caddy, where the fetch tool's result was 13 characters. The headings, numbers, tabs and charts are drawn later by JavaScript over a websocket, so no server-side text exists to scrape, and retrying the same URL gives the same shell. To get what is on screen: (1) if you have the repo, read the page script (st.markdown, st.subheader and similar calls) to get exact labels and section names; (2) for rendered text or screenshots, use a headless browser such as Playwright and wait for the app to finish loading; (3) to test a page without a browser, use streamlit.testing.v1.AppTest. Don't spend fetches guessing subpage URLs: the shell has no links to follow.