hughdbrown hdb-linkedin-profile-fixer
Rewrite a LinkedIn profile so it reads specific, scoped, and numbers-driven — the kind of profile that pulls inbound recruiter DMs without applying to anything.
Usage
/hdb:linkedin-profile-fixer
Optionally pass the profile export path and/or target-role notes inline.
Core principles (do not violate)
- Sequence beats stacking. Rewrite one section per step. Never audit + rewrite headline, About, experience, skills, and featured in a single response — output quality collapses and every section comes out in the same generic ghostwriter voice. One brilliant section beats six shallow ones.
- A target role is mandatory. You cannot optimize for "a better job." You need the actual text of 3–5 real job descriptions the user wants. Without them, refuse to proceed and ask for them.
- Never invent metrics. Numbers ("4M+ creators monthly", "2.7B users") are what make a profile evaluable. You do not have the user's real numbers — ask for them. If the user has none for a claim, rewrite without a fabricated figure; do not guess, estimate, or insert a placeholder that reads as fact.
- Kill generic filler. Ban "passionate", "delight users", "results-driven", "cross-functional initiatives", "leverage", "synergy", and verbs-and-vibes bullets with zero substance. Every line must be something a recruiter can actually evaluate.
- PDF export, not scraping. Tell the user to export their profile (LinkedIn → "Resources" panel on the right → "Save to PDF") and upload it. Do not suggest Apify or live-scraping connectors — they break on private profiles, mobile sessions, and rate limits.
Setup gate
Before writing anything, confirm you have both inputs:
- The profile. A LinkedIn PDF export (preferred), or pasted profile text. If missing, give the export instructions above and stop.
- Targets. 3–5 full job descriptions for roles the user actually wants, as raw text. For emerging roles (AI PM, agentic-workflows lead) the closest real JDs are fine. If missing, stop and ask for them. Vague targeting → vague output.
Once both are in hand, extract from the JDs: the recurring keywords, required skills, seniority signals, and the 2–3 outcomes these roles most reward. You will anchor every rewrite to these.
The sequence — run these steps in order, one response each
Run each step, present the rewrite, and wait for the user's reaction / real numbers before moving to the next. Do not batch.
Step 1 — Tear it apart (diagnosis only)
Read the whole profile against the target JDs. Output a blunt critique, not a rewrite: what the headline buries, where the About section opens generically, which experience bullets are "verbs and vibes" with no numbers, and which JD keywords are missing entirely. End by naming the single section that will move the needle most — usually the headline or the About opener — and start there.
Step 2 — The headline
Rewrite the headline. It must lead with the most interesting, specific thing the user does, scoped and quantified. Model the form: "I help [org] ship [specific thing] that [number/scale] [audience]. Previously [specific prior scope]." Offer 2–3 variants. Ask the user for any scale/number you don't have rather than inventing one.
Step 3 — The About section
Rewrite the About. The first sentence is everything — it must be specific and scoped, never "I'm a [title] passionate about…". Weave in JD keywords naturally, front-load concrete outcomes with real numbers (ask for them), and keep it in the user's voice, not LinkedIn-poetry. Surface every place you need a real metric as an explicit [NEED: …] question.
Step 4 — Experience bullets
Rewrite experience bullets for the most recent / most relevant 1–2 roles. Every bullet = action + scope + measurable outcome. Convert "Led cross-functional initiatives" into "Shipped X, now used by Y, cutting Z by N%". Mark each missing figure as [NEED: …]. Do not fabricate.
Step 5 — Skills, Featured, and consistency pass
Align the Skills section to the JD keywords (drop stale ones, add the ones recruiters filter on). Recommend what to pin in Featured. Then do a final read-through for voice consistency and any remaining filler or unverified numbers.
Closing checklist
Before declaring done, verify:
- Every section was rewritten in its own step (no stacking)
- Every number in the final copy is one the user supplied — zero invented figures, zero unresolved
[NEED: …]left as fact - Headline and About both open specific and scoped — no banned filler words remain
- Rewrites are anchored to the supplied JDs' keywords and rewarded outcomes
- The copy still sounds like the user, not a ghostwriter