linkedin-optimizer

name

linkedin-optimizer

description

Audit LinkedIn profile alignment, generate optimized alert configurations, and track recommendation quality over time. Use this skill when the user says "my LinkedIn suggestions suck," "optimize my LinkedIn," "fix my job alerts," "what should my headline say," "am I showing up for the right roles," "audit my LinkedIn profile," "LinkedIn keywords," or any question about improving LinkedIn visibility and job match quality. Also trigger when the user forwards a LinkedIn job email digest for quality scoring. Do NOT trigger for general job search strategy (search-optimizer) or individual application evaluation (job-intake).

LinkedIn Optimizer — Profile Audit, Alert Tuning, and Match Monitoring

Intent

1. LinkedIn is an input channel, not a strategy — this skill optimizes the signal quality of one source; career strategy lives in search-optimizer
2. Profile is the lever, alerts are the filter — most "bad recommendation" problems are profile problems, not alert problems
3. Measurable improvement — track signal-to-noise ratio before and after changes so we know what worked
4. Manual execution, automated analysis — LinkedIn doesn't allow API-based profile edits; this skill generates the exact text/config to paste, then tracks results
5. Keyword alignment is a scoring problem — reuse the same pattern-matching logic from job-intake to compare profile text against target role requirements

Why this skill exists

LinkedIn's recommendation engine is a black box that infers your seniority, domain, and interests from profile text, skills, behavioral signals, and explicit preferences. When the inference is wrong, you get irrelevant job emails. This skill audits your profile against what the algorithm needs, generates precise alert configurations, and monitors whether changes improved match quality.

Prerequisites

• master/achievements.md — Source of truth for accomplishments and keywords
• master/narrative.md — Positioning and tone guide
• packages/scoring-rules/scoring-rules.yaml — For scoring LinkedIn digest jobs
• At least 3-5 target role examples (from high-scoring applications in applications/)

Workflow

Mode 1: Profile Audit

Trigger: "audit my LinkedIn," "optimize my profile," "what should my headline say"

Step 1: Load profile baseline

Read:

• master/achievements.md — all quantified accomplishments
• master/narrative.md — positioning themes and differentiators
• Top 5 highest-scoring applications (sort applications/*/metadata.json by match score)
• references/profile-audit-checklist.md — the audit framework

Step 2: Extract keyword universe

From the top-scoring applications, extract:

• Required skills that matched as "strong" — these are your proven keywords
• Job titles from those postings — these are your target title variants
• Industry terms that appear in 3+ postings — domain signal words
• Leadership terms from responsibility sections — seniority signal words

Cross-reference against master/achievements.md to confirm each keyword has backing evidence.

Step 3: Generate profile recommendations

For each LinkedIn profile section, generate specific text:

Headline (220 chars max):

• Format: [Target Title] | [Domain] | [2-3 Differentiators with Numbers]
• Pull from narrative.md differentiators and achievements.md top metrics
• Include semantic neighbors of the target title (see references/algorithm-signals.md)

About Section (2,600 chars max, first 275 visible on mobile):

• First 275 chars = quantified hook with target title and top 3 metrics
• Body = narrative.md themes expanded with specific numbers from achievements.md
• Close with what you're looking for (signals intent to the algorithm)

Skills Section (50 max, order matters):

• Generate 3 semantic clusters from references/algorithm-signals.md:
  • Leadership cluster (8-10 skills)
  • Technical cluster (8-10 skills)
  • Domain cluster (8-10 skills)
• Pin top 3 skills (most endorsed/relevant) to be visible on profile

Experience Bullets:

• For each role, ensure C-A-R format (Challenge-Action-Result)
• Front-load keywords from the keyword universe
• Include metrics from achievements.md for each role

Featured Section:

• Recommend 2-3 items: a post about AI in healthcare, a project summary, a recommendation quote

Step 4: Output the audit report

Write linkedin-audit-{date}.md to the working directory:

# LinkedIn Profile Audit — {date}

## Keyword Gap Analysis
| Keyword | In Achievements? | In Top Matches? | On LinkedIn? | Action |
|---------|------------------|-----------------|--------------|--------|
| ...     | Yes/No           | 3/5 postings    | Yes/No       | Add/OK |

## Recommended Headline
> [exact text to paste, ≤220 chars]

## Recommended About (First 275 chars)
> [exact text to paste]

## About Section (Full)
[full 2,600-char text]

## Skills to Add
### Leadership Cluster
- [skill 1], [skill 2], ...
### Technical Cluster
- [skill 1], [skill 2], ...
### Domain Cluster
- [skill 1], [skill 2], ...

## Skills to Remove
- [irrelevant skills that dilute signal]

## Experience Section Edits
### [Company] — [Title]
**Current:** [summary of what's there or "unknown — check manually"]
**Recommended bullets:**
- [C-A-R bullet with keywords]
- ...

## Featured Section
- [recommended items]

## Open to Work Settings
- **Job titles (5):** [exact titles to enter]
- **Locations:** [locations + Remote]
- **Visibility:** Recruiters Only

Mode 2: Alert Generator

Trigger: "fix my alerts," "what alerts should I set up," "my email suggestions are bad"

Step 1: Analyze target roles

From top-scoring applications and narrative.md, identify:

• Target title variants (Director of Engineering, VP Engineering, Head of Engineering, etc.)
• Target industries
• Location preferences
• Remote/hybrid/onsite preference
• Company size preferences

Step 2: Generate alert configurations

For each alert, specify:

• Search query (with boolean operators): ("Director" OR "VP" OR "Head of") AND ("Engineering" OR "Software")
• Experience level filter: Director, Executive
• Job type: Full-time
• Location: [specific or Remote]
• Frequency: Daily

Generate 3-5 focused alerts rather than 1 broad one:

# Recommended LinkedIn Job Alerts

## Alert 1: Director of Engineering — Remote
- **Query:** `("Director of Engineering" OR "Director, Engineering" OR "Director of Software Engineering")`
- **Experience:** Director, Executive
- **Location:** Remote
- **Type:** Full-time
- **Frequency:** Daily

## Alert 2: VP / Head of Engineering — Remote
- **Query:** `("VP of Engineering" OR "Head of Engineering" OR "VP, Engineering")`
- **Experience:** Director, Executive
- **Location:** Remote
- **Type:** Full-time
- **Frequency:** Daily

## Alert 3: Healthcare Engineering Leadership
- **Query:** `("Engineering" OR "Technology") AND ("Healthcare" OR "Health" OR "Clinical")`
- **Experience:** Director, Executive
- **Location:** Remote
- **Type:** Full-time
- **Frequency:** Daily

Also output the email settings to change:

• Disable "Job recommendations" emails (these use broad inference)
• Keep only "Job alerts" emails (your curated alerts)
• Enable all 4 job-seeking preference toggles

Mode 3: Digest Scorer

Trigger: User pastes or describes jobs from a LinkedIn email digest

Step 1: Parse the digest

Extract each job from the pasted content:

• Title, company, location, key requirements (if visible)

Step 2: Score each job

Use the same scoring logic as job-intake against master/achievements.md:

• Strong match / Good match / Stretch / Long shot
• Flag any that are clearly wrong seniority level

Step 3: Calculate signal-to-noise ratio

Signal-to-Noise = (Strong + Good matches) / Total jobs in digest

Track over time in linkedin-signal-log.json:

{
  "entries": [
    {"date": "2026-03-09", "total": 15, "strong": 1, "good": 3, "stretch": 4, "long_shot": 7, "snr": 0.27},
    {"date": "2026-03-16", "total": 12, "strong": 3, "good": 5, "stretch": 3, "long_shot": 1, "snr": 0.67}
  ]
}

Step 4: Report and recommend

## LinkedIn Digest Score — {date}

**Signal-to-Noise:** 27% (4 relevant / 15 total)
**Trend:** [improving/declining/stable] vs last check

### Relevant Jobs
| # | Title | Company | Score | Action |
|---|-------|---------|-------|--------|
| 1 | Dir of Eng | Acme Health | Strong | → Run job-intake |
| 2 | Head of Eng | Beta Corp | Good | → Run job-intake |

### Irrelevant Jobs (Why They Appeared)
| # | Title | Company | Issue |
|---|-------|---------|-------|
| 3 | Sr Engineer | Gamma | Wrong seniority — profile may need title keywords |
| 7 | PM Lead | Delta | Wrong function — remove "Product" from skills |

### Recommendations
- [specific profile/alert changes based on the noise pattern]

Mode 4: Keyword Optimizer

Trigger: "what keywords am I missing," "why am I not showing up for the right roles"

Step 1: Load all strong-match applications

Read all applications/*/metadata.json where match_score indicates strong or good match.

Step 2: Extract winning keywords

From those job descriptions, extract:

• Skills mentioned in 3+ postings
• Title variants used
• Industry-specific terms
• Leadership/scope descriptors

Step 3: Compare against profile

Cross-reference winning keywords against:

• Current achievements.md entries
• Recommended LinkedIn skills (from last audit)
• Headline and About section recommendations

Step 4: Output keyword gap report

## Keyword Optimization Report — {date}

### High-Value Keywords Missing from LinkedIn
| Keyword | Appears In | Evidence in Achievements? | Priority |
|---------|-----------|--------------------------|----------|
| FHIR    | 4/5 matches | Yes — ilumed integration | HIGH     |
| ...     | ...       | ...                      | ...      |

### Keywords on LinkedIn That Dilute Signal
| Keyword | Relevance to Target Roles | Recommendation |
|---------|--------------------------|----------------|
| jQuery  | 0/5 matches              | Remove         |
| ...     | ...                      | ...            |

Edge Cases

• No LinkedIn access — Skill generates all text/config offline; user applies manually
• User hasn't applied to enough roles — Use narrative.md and achievements.md as primary source; note that recommendations will improve with more application data
• User wants to optimize for a different role type — Ask which titles to target before running the audit
• Multiple career tracks — Generate separate headline/about variants for each track; user chooses which to activate

Skill Composition

Upstream	This skill	Downstream
master/achievements.md (keywords)	Profile audit	LinkedIn profile (manual edits)
master/narrative.md (positioning)	Alert generator	LinkedIn alert config (manual)
scoring-rules.yaml (scoring)	Digest scorer	search-optimizer (channel quality)
applications/*/metadata.json (patterns)	Keyword optimizer	resume-tailor (keyword sync)
		job-intake (from digest leads)

Read references/profile-audit-checklist.md for the full audit framework. Read references/alert-templates.md for boolean search patterns. Read references/algorithm-signals.md for LinkedIn algorithm behavior.