LinkedIn Profile Optimizer

Audit a LinkedIn profile against real recruiter-search signals and fix the gaps via Chrome MCP browser automation.

Prerequisites

• LinkedIn MCP — for pulling the full profile and analytics
• Chrome MCP — for making edits. Auto-starts Chrome on first tool call (no manual launch flags needed); use the session where you're logged into LinkedIn
• For multi-step edit flows, prefer the pipeline tool (navigate → snapshot → click/fill in one call) over individual tool round-trips

Workflow

1. Pull Profile & Audit

get_person_profile(username, sections: experience,education,skills,certifications,projects,honors,languages,posts)

Analyze these signals against recruiter-search behavior:

Present ranked pitfalls with specific, actionable fixes.

2. Optimize via Chrome MCP (Priority Order)

A. Open to Work (Highest Impact)

• Path: Profile → "Open to" button → "Finding a new job"
• Set: multiple titles (5 max), all location types (On-site/Hybrid/Remote), target city, "Flexible/casually looking", Full-time, Recruiters only visibility
• ⚠️ Enabling Remote re-renders the form (refs shift, adds required "Locations (remote)" field)

B. Location

• Path: /in/{user}/edit/intro/
• Change City field to actual work location

C. Headline

• Path: /in/{user}/edit/intro/ → headline textbox (max 220 chars)
• Template: {Role} @ {Company} | {Specialty} | {Tech1} · {Tech2} · {Tech3} · {Tech4} · {Tech5}

D. Top Skills

• Path: /in/{user}/add-edit/SUMMARY/ (the About editor contains the skills picker)
• Remove soft skills, add 5 technical skills matching target roles
• Use typeahead: type skill name → ArrowDown → Enter to select
• ⚠️ This is NOT in the Skills detail page — it's inside the About editor

E. Languages

• Path: /in/{user}/add-edit/LANGUAGE/
• Typeahead for language name + proficiency level dropdown
• Add native language + English with correct CEFR mapping:
  • B2 = "Professional working proficiency"
  • C1 = "Full professional proficiency"
  • Native = "Native or bilingual proficiency"

F. About Section

• Path: /in/{user}/add-edit/SUMMARY/
• Rich text editor (tiptap). Use .fill() to replace content.
• Lead with: role + company + city + specialty + stack (first 2 lines show in search previews)
• Include: what you work on (bullet points), education, what you're open to, links
• Keep handle/alias at the end, not the opening

G. Projects

• Path: /in/{user}/add-edit/PROJECT/
• Fields: name (255 max), description (2000 max), start/end dates, "Associated with" dropdown, skills, media
• Prioritize: open-source with stars, thesis/research, portfolio
• Associate with relevant experience entry

H. Experience Descriptions

• Path: /in/{user}/details/experience/edit/forms/{formId}/
• Find formId from the experience detail page's edit links
• Description field: 2000 char max
• ⚠️ Toggle "Notify network" OFF before saving (unless user wants to broadcast)

I. Remove Internal Mobility Card

• Path: /in/{user}/opportunities/internal-mobility/edit/ → Delete → Confirm
• Removes the "Interested in jobs at {Company}" card that signals "not looking externally"

3. Verification

• Navigate to profile and screenshot final state
• Check analytics after 1-2 weeks for lift in search appearances

4. Persist to Mercury

Capture the profile metrics + computed score so the dashboard can chart progress over time. Run this whenever you pull fresh analytics (this is the only point profile metrics enter the system — the dashboard can't scrape them itself):

mercury metric record \
  --search-appearances {N} --profile-views {N} \
  --post-impressions {N} --connections {N} \
  --score {0-100} \
  --breakdown '{"openToWork":"recruiters-only","headline":"strong","location":"São Paulo","topSkills":"strong","languages":"set","projects":"4 added","about":"strong","connections":"weak (<500)"}'

The --breakdown is a JSON object of {signal: status} pairs — one per audited signal from step 1's table. The dashboard renders each as a labeled pill (Overview

• Profile sections), color-coded: greenish for strong/good/set/done values, red for weak/missing/empty/none, neutral otherwise. Use concise, human-readable status strings so the pills read well.

Log the optimization session:

mercury activity log --skill profile-optimizer --summary "Audited + fixed {what}; score {old}->{new}"

LinkedIn Deep Links Reference

/in/{user}/edit/intro/                              — Headline, location, name
/in/{user}/add-edit/SUMMARY/                        — About text + Top Skills picker
/in/{user}/add-edit/LANGUAGE/                       — Add language
/in/{user}/add-edit/PROJECT/                        — Add project
/in/{user}/details/experience/edit/forms/{id}/      — Edit experience entry
/in/{user}/details/skills/edit/forms/{id}/          — Edit skill associations
/in/{user}/opportunities/internal-mobility/edit/    — Internal mobility card

Gotchas

1. Typeahead fields (Language, Skills): must use ArrowDown + Enter after typing — plain text won't register (use press_key after fill)
2. Multiple "Save" buttons: use the dialog-scoped one, not global nav — target by the uid from the latest take_snapshot
3. Refs shift on re-render: after enabling Remote, adding/removing skills, etc. — take a fresh take_snapshot (stale uids won't resolve)
4. Long option lists bloat snapshots: year/skill dropdowns are large — prefer acting on the field uid directly rather than snapshotting the whole open list
5. "Done" button conflicts: video player has one too — use press_key Escape to close post-save modals
6. Premium "next action" modals: appear after saving (suggest connections, ask about skills) — dismiss with Escape or skip
7. Use pipeline for edit flows: chain navigate → wait_for → take_snapshot → click/fill to avoid stale refs and reduce round-trips