Feature request: read/list access to saved jobs, plus the ability to act on existing records

[upendedshark]

Summary

The JobSync MCP server currently only supports writing jobs (add_job, add_jobs_batch, update_job) and looking up a single job if you already have its exact posting URL (find_job). There's no way for an MCP client to browse, filter, or query the set of jobs already saved in JobSync. This makes it impossible to do basic tracking/triage work — e.g. "which jobs in the Netherlands haven't been applied to yet" — without leaving the assistant and checking the app directly.

Current limitations

• No list_jobs / search_jobs tool exists. The only read path is find_job(jobUrl), which requires the caller to already know the exact posting URL.
• There's no way to query by status, country/location, company, tag, or match score.
• Jobs created through the JobSync web app (rather than via MCP) are explicitly reported as not editable via MCP (update_job is rejected for them) — even though find_job can still surface their id, title, and status.
• There's no tool to persist a cover letter against a job record — cover letters have to live outside JobSync entirely (as separate documents), so they're not visible next to the job the way a match analysis is.
• No pagination or bulk-read pattern exists at all, since there's no listing endpoint to paginate.

Requested capability

1. list_jobs (or search_jobs) — return jobs matching optional filters:
   • status (new / draft / applied / interview / offer / rejected / expired / archived)
   • location / country (free-text or structured match)
   • company
   • tags
   • hasMatchScore (boolean) / matchScoreRange
   • appliedDateRange / createdDateRange
   • Pagination (cursor/limit or offset/limit)
   • Returns at minimum: jobId, company, jobTitle, location, status, matchScore, appliedDate, jobUrl, whether it was created via MCP or the web app.
2. get_job(jobId) — fetch full detail for a single job by id (not just by URL), including the full stored description, saved match analysis, and any saved cover letter text. Useful once a job id is known from list_jobs without needing the original URL again.
3. Remove (or explain/relax) the "web-app-created jobs can't be updated via MCP" restriction. At minimum, allow update_job to work on any job the account owns regardless of how it was created — or if there's a real technical reason for the split (e.g. different storage schema), expose a way to at least append a match analysis or cover letter to those records too, since save_match_result currently isn't blocked for them but the ordinary edit path is.
4. save_cover_letter(jobId, coverLetterText) — a save/persist tool mirroring save_match_result, so a drafted cover letter is stored against the job record and visible in the app, instead of only existing as an external document. Ideally support versioning if a cover letter is redrafted.
5. A lightweight set_status(jobId, status) convenience wrapper — not strictly necessary since update_job already supports a status field, but worth considering alongside the above if a lower-friction, single-purpose tool is preferred for the common "mark this applied/interview/rejected" action.

Example use cases this unblocks

• "Which jobs in the Netherlands are still unapplied?" → list_jobs({ location: "Netherlands", status: "new" })
• "Show me every job scoring above 70 that I haven't drafted a cover letter for yet" → list_jobs({ matchScoreRange: [70, 100] }) cross-referenced against cover-letter presence from get_job
• "Write cover letters for all my open Australian applications" → list, then iterate and save_cover_letter for each
• "I edited a job in the app, can you rescan and rescore it?" → get_job(jobId) to pull current state without needing the original URL again

Priority

This is the single biggest gap in an otherwise solid MCP surface for job-search tracking — without it, any assistant using JobSync via MCP is limited to acting on jobs the user explicitly names or links, and can never answer aggregate/status questions about the pipeline as a whole.