Feature Description
Feature
Add support for reading a user’s saved jobs directly from LinkedIn’s Jobs Tracker page through a new get_saved_jobs MCP tool. The tool should scrape the saved-jobs view, return the raw page text in the
standard sections payload, extract LinkedIn job IDs from the saved-job links, and continue through multiple pages of saved jobs up to a configurable max_pages limit.
The implementation should handle LinkedIn’s paginated saved-jobs UI reliably by detecting the highest visible numbered page as pagination advances, deduplicating job IDs across DOM polling, and reporting
progress while pages are being fetched. This gives agents a consistent way to retrieve a user’s saved jobs without requiring manual page-by-page navigation.
Implementation
Here is the PR that implemented this feature #338. This is the second PR for this. I did it once, then the code changed so I've re-implemented this again. Would be great to get this merged please @stickerdaniel
Use Case
Users often save jobs on LinkedIn as a shortlist for later review, but those jobs were not previously accessible through the MCP server. Adding saved-jobs support lets an agent read that shortlist, summarize opportunities, compare saved roles, prioritize follow-up, and use the extracted job IDs to fetch more detailed information for specific postings.
Pagination support matters because saved jobs can span multiple tracker pages. Without multi-page scraping, the tool would only expose a partial shortlist and become less useful for users with an active or long-running job search.
Suggested Approach
Already implemented as #338
Feature Description
Feature
Add support for reading a user’s saved jobs directly from LinkedIn’s Jobs Tracker page through a new get_saved_jobs MCP tool. The tool should scrape the saved-jobs view, return the raw page text in the
standard sections payload, extract LinkedIn job IDs from the saved-job links, and continue through multiple pages of saved jobs up to a configurable max_pages limit.
The implementation should handle LinkedIn’s paginated saved-jobs UI reliably by detecting the highest visible numbered page as pagination advances, deduplicating job IDs across DOM polling, and reporting
progress while pages are being fetched. This gives agents a consistent way to retrieve a user’s saved jobs without requiring manual page-by-page navigation.
Implementation
Here is the PR that implemented this feature #338. This is the second PR for this. I did it once, then the code changed so I've re-implemented this again. Would be great to get this merged please @stickerdaniel
Use Case
Users often save jobs on LinkedIn as a shortlist for later review, but those jobs were not previously accessible through the MCP server. Adding saved-jobs support lets an agent read that shortlist, summarize opportunities, compare saved roles, prioritize follow-up, and use the extracted job IDs to fetch more detailed information for specific postings.
Pagination support matters because saved jobs can span multiple tracker pages. Without multi-page scraping, the tool would only expose a partial shortlist and become less useful for users with an active or long-running job search.
Suggested Approach
Already implemented as #338