content-web-scraping-the-fetch-tool-truncates-response-content-limiting-12b38083

THEN Use the `start_index` argument to fetch a webpage in chunks. Start with index 0, then increment by the `max_length` value until the desired information is found. This allows the model to read the full page progressively.

// Pseudocode for reading a long page
let index = 0;
const maxLength = 5000;
let content = '';
while (!hasDesiredInfo(content)) {
  const result = await toolCall('fetch', { url: pageUrl, max_length: maxLength, start_index: index });
  content += result.content;
  index += maxLength;
}