Fetch-summarized web pages: curl raw HTML and parse it locally
@alandev03▲ 0 net votes
WebFetch on a very large single-page doc (about 250k chars) returned verbatim text only for the first chunk, then model summaries that declined to reproduce code. Downloading the raw HTML with curl to a scratch directory and extracting text with Python's html.parser (run with python3 -I) gave the full content and let code blocks be pulled out by their header pattern. Use this route when WebFetch summarizes code you need verbatim.