Jint

Requests and state ​

Each fetch or eval invocation creates a browser, context, and page, then disposes them. Cookies supplied with --cookie and storage created by the page therefore live only for that command.

Headers and cookies can seed a one-shot load:

bash
jint-browser fetch https://example.org/account \
  --header 'Authorization: Bearer token' \
  --cookie theme=dark

Headers apply to every request unless the page or browser supplies that request's own value. A User-Agent header becomes the page's user agent, keeping navigator.userAgent and network requests consistent.

For a persistent browsing session, use:

  • serve and inspect requests, cookies, storage, and history through a CDP client.
  • mcp, whose network_requests, cookies, set_cookie, back, forward, and reload tools expose session state.

There is no separate jint-browser requests command, and fetch output contains only the requested document representation. Page errors are written to standard error.

Images are fetched so that a page can read complete, naturalWidth, currentSrc and the load event, and a srcset or <picture> selects the candidate its descriptors and media conditions choose, but only the container header is read — the browser renders nothing. References it does not follow, such as a media element, still appear in the request log with a reason. Request logs are summaries rather than traffic archives and do not retain arbitrary bodies or request headers.