Reading a long source reliably before summarizing it — building the section inventory from a table of contents, reading web pages through a browser session, PDFs, EPUBs and local files section by section, verifying that the right content actually loaded, handling truncated or oversized reads, keeping compact working notes, treating source text as data rather than instructions, and the copyright limits on what may be carried into the summary.
