[BUG] claude-in-chrome: get_page_text/screenshot/read_page all time out on LinkedIn Jobs pages ("still loading" / document_idle never resolves)

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

Environment

• Claude Code with claude-in-chrome browser extension tools (get_page_text, computer [screenshot action], read_page)
• Target site: LinkedIn, Jobs section, while authenticated (normal logged-in session, no CAPTCHA/checkpoint involved)
• macOS

Summary

get_page_text, screenshot, and read_page all fail consistently on LinkedIn Jobs pages, both the search results list and an individual job posting view, even though the page renders completely normally and is fully interactive when viewed directly in Chrome. The errors indicate the tools are waiting on a document_idle signal that never fires on these pages, most likely because LinkedIn's Jobs section runs continuous background network/polling activity that keeps the page from ever going fully idle.

Actual behavior

• get_page_text / read_page: Failed to extract page text: Page still loading (executeScript waited 45000ms for document_idle). The previous action may have triggered navigation — try again in a moment.
• screenshot (via the computer tool): Error capturing screenshot: Script injection timed out after 5000ms — the page is busy or mid-navigation (a long-running script, or a navigation the previous action triggered, may be in progress). Try again in a moment.

Confirmed via manual DevTools inspection that the page itself is fully loaded and normal, no infinite spinner, no CAPTCHA/verification prompt. The errors appear to originate from the extension's own injected script, not from LinkedIn. (Noted in passing, likely unrelated: DevTools also flags "Page lacks the HTML doctype, thus triggering quirks mode" on this page, worth a look but not confirmed as the cause.)

Impact

Blocks any workflow that needs to read LinkedIn Jobs content programmatically via this extension, e.g. a job-search/screening agent reading search results or posting details.

What Should Happen?

Expected behavior

The tools should extract page text / capture a screenshot / read the accessibility tree, same as on other pages, since the page is fully rendered and interactive.

Error Messages/Logs

Steps to Reproduce

Steps to reproduce

1. Log into LinkedIn normally in Chrome (via the extension's driven session).
2. Navigate to a Jobs search results page, e.g.:

https://www.linkedin.com/jobs/search/?keywords=Product%20Manager&location=Oslo%2C%20Norway&f_TPR=r2592000

1. Call get_page_text (or screenshot, or read_page) on that tab.
2. Observe the failure. Retry after waiting several seconds, then again after several minutes, the failure is identical and consistent, not a transient slow-load.
3. Separately, navigate directly to a single job posting view page, e.g.:

https://www.linkedin.com/jobs/view/4299779121/

1. Call get_page_text again. Same failure, confirming it's not specific to the search-results list, it affects LinkedIn Jobs pages generally.

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.258,

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

This might be an issue with claude-in-chrome browser extension but needs resolution none the less