Skip to content

Jobs module

The jobs module is a complete job-search automation workflow — CV-driven, with intelligent matching, auto-generated cover letters, and an application tracker.

Jobs module

Workflow

┌────────┐    ┌─────────┐    ┌────────┐    ┌──────────┐    ┌──────────┐
│  Setup │ →  │ Profile │ →  │ Search │ →  │   Apply  │ →  │  Tracker │
│ (CV)   │    │ (skills)│    │ (match)│    │ (cover)  │    │ (status) │
└────────┘    └─────────┘    └────────┘    └──────────┘    └──────────┘

The page has 4 tabs that map to this pipeline.

1. Setup tab — CV upload

Upload your CV once, the parser extracts everything.

Supported formats: PDF, DOCX (DOC requires manual paste).

Parser extracts:

  • Contact info (email, phone, location)
  • Work history (company, title, dates, descriptions)
  • Education (school, degree, dates)
  • Skills (categorized)
  • Certifications + dates

Under the hood:

  • pdfplumber for PDF
  • python-docx for DOCX
  • LLM extraction for normalization (when linkedin-mcp-llm-keys is configured)

Tip

Re-upload anytime — the old CV is versioned, never deleted. The tracker always knows which CV version produced which application.

2. Profile tab — what you want

Field What it controls
Target roles Job title keywords (used in search query)
Skills (priority) Boosted in match scoring
Salary range Filter: jobs outside range hidden by default
Locations Filter: remote / specific cities / countries
Visa status Filter: requires-visa jobs excluded
Notice period Used in cover letter

The profile is saved per-account. Multi-account users get separate profiles.

3. Search tab — find + rank

Click Find jobs → the searcher hits LinkedIn Voyager with your filters + queries.

Match scoring

Each job gets a score from 0–100:

Component Weight What it measures
Keyword Jaccard 60% Overlap between job description tokens and your CV tokens
Skill-set bonus 30% How many of your priority skills appear in the JD
Title similarity 10% Levenshtein distance between job title and your target roles

Score thresholds:

Score Label Recommendation
80–100 Strong match Apply with confidence
60–79 Good match Apply, customise cover
40–59 Weak match Review manually
0–39 Skip Don't apply

Filters

The searcher applies your profile filters automatically. Override per-search:

  • Date posted (24h / week / month / anytime)
  • Experience level (internship / entry / mid / senior / director)
  • Job type (full-time / part-time / contract / internship)
  • Remote / hybrid / on-site
  • Industry
  • Company size
  • Under 10 applicants
  • In your network
  • Fair chance (LinkedIn's diversity filter)

4. Apply tab — generate + submit

Cover letter

When you click Apply on a job:

  1. Match score is recomputed against the live JD (not just the search snippet)
  2. The cover letter generator picks one of 4 templates (intro / story / technical / culture-fit) based on score + job category
  3. Template is filled with: company name, role, 2 of your strongest matching skills, 1 quantified achievement from your CV
  4. You see the draft → edit if you want → click Send

Eligibility gate

Before submission, the safety layer checks:

  • ✅ Your profile has the required skills (else: blocked with explanation)
  • ✅ Daily application cap not hit (else: queued for tomorrow)
  • ✅ No duplicate application to same company within 90 days
  • ✅ Job is still accepting applications
  • ✅ easy_apply_only=True AND easy_apply=True (else: opens LinkedIn in browser for manual finish)

Manual finish

If the job requires LinkedIn's manual flow (about 30% of listings), you're redirected to the LinkedIn Easy Apply form with your cover letter + profile pre-filled.

5. Tracker tab — status

Every application is logged with:

Field Source
Job ID LinkedIn Voyager
Title + company + location Search result
Match score Computed at time of apply
Cover letter used Stored version
CV version used Stored version
Applied at UTC timestamp
Status Pending → Submitted → Viewed → Replied → Interview → Offer / Rejected
LinkedIn URL For follow-up

Updates flow in via the engagement listener — when the recruiter responds, the tracker marks the row automatically.

CLI equivalent

For headless job hunting:

# Search with default profile
linkedin-mcp-jobs search --query "Senior Python Engineer" --location "Remote"

# Apply by ID (cover letter auto-generated)
linkedin-mcp-jobs apply --job-id 1234567890 --confirm

# Tracker dump
linkedin-mcp-jobs tracker --status pending --format csv

Note

linkedin-mcp-jobs is not a separate binary — it's the same linkedin-mcp-pro MCP server exposed via the jobs_* tool family. The CLI uses the same tool registry.

Privacy

  • Your CV stays on your machine in ~/.linkedin-mcp/profile/cv/
  • Never sent to LinkedIn until you click Apply
  • Never sent to any LLM unless you explicitly enable AI extraction in /llm
  • Can be deleted at any time from /settings

Next