Skill Locker
All skills
v1.3.03 loop iterations

User Profile Builder

Without a profile, every conversation starts from zero — Claude guesses your tone, your expertise level, your industry jargon, your preferences. The User Profile Builder runs a guided interview and writes a structured profile so every future session begins with the colleague-who-knows-how-you-work version.

What this skill does

Without a profile, Claude has to guess. It guesses your jargon level, your tone, whether you want bullets or prose, how technical your audience is, whether you mind passive voice or hate it. Each guess might be right; often it isn't, and you correct it. The User Profile Builder replaces the guessing with a structured profile that captures the data that actually matters.

The interview is adaptive, not a wall of questions. Round 1 covers identity and role — title, company, responsibilities, primary audience. Round 2 captures working style and the most important question of all: "what annoys you most about AI output?" The negative preferences from that question often drive 80% of the daily quality improvement. Round 3 covers recurring work — the tasks you do weekly that feel repetitive, the deliverable you produce most often, the templates you already follow. Round 4 covers goals, tools, and pet peeves. Round 5 asks for a voice sample — a paragraph the user has written that represents their voice well, analysed for sentence length, formality, active-voice ratio, vocabulary complexity.

The skill skips rounds that don't apply. A solo freelancer doesn't need team-dynamics questions. A student doesn't need client-audience questions. A three-round interview that produces a usable profile beats forcing all five rounds and losing the user halfway through. Short answers don't trigger more probing — work with what's offered and fill gaps over time.

The voice fingerprint is the most under-used section in most profile setups and the most valuable. "Direct and professional" tells Claude almost nothing. "Average sentence 14 words. Contractions present. Uses em-dashes for emphasis. Active-voice ratio above 80%. Dry humour, no exclamation marks" tells Claude exactly how to reproduce the voice. The skill insists on the explicit breakdown rather than the vague label.

After the profile, three things happen. Confirmation — the profile gets read back, with the focus on Voice Fingerprint and Communication Preferences because those have the highest daily impact. Quick wins — two or three specific ways Claude can immediately help with the user's recurring work, written as concrete offers ("I can draft your weekly ELT briefings in your voice") not capability lists ("I can help with writing"). Review cadence — a quarterly refresh, or sooner if the role changes. Most of a profile survives a role change; only certain sections need updating, and the skill knows which ones.

The profile lives in CLAUDE.md, a memory file, or a dedicated profile document. It's a starting state, not a constraint. Override it where the project demands different defaults — but most of the time, the daily output gets better immediately because Claude no longer has to guess what kind of person it's writing for.

When this triggers

  • ·You're setting up Claude as a long-term working partner and want the day-one onboarding done properly
  • ·You're tired of explaining your role, your audience, and your voice every session
  • ·You want a profile that adapts the jargon level, the tone, and the depth automatically
  • ·You're handing Claude to a teammate and want them to inherit a sensible starting state
  • ·You're a non-developer and the generic AI defaults don't fit your work

Example

Trigger

User: 'I'm new to using Claude properly. Set me up a profile.'

Output

[Adaptive interview, 4 rounds — Round 5 skipped, voice sample provided in Round 2 as a LinkedIn post.] # User Profile: James *Generated: 2026-05-21 | Last updated: 2026-05-21* ## Role & Context - Title: Change Manager - Company: LCC | Public sector | 200+ staff - Main responsibilities: · Run change programmes (currently a £4M transformation) · Stakeholder briefings, weekly to ELT · Coaching middle managers through restructure - Primary audience: ELT, middle managers, programme team ## Communication Preferences - Tone: Direct, no fluff. Smart-colleague register. - Format default: Short paragraphs > bullets when sequential - Length preference: Concise. Expand on request only. - Jargon level: High in change/transformation; explain in tech. - Hard rules: No emojis. UK English. £ where natural. ## Voice Fingerprint - Sentence style: Short, mixed length, occasional fragments - Formality: 6/10 — professional but not stiff - Personality markers: Dry, direct, not afraid of bad news - Sample analysis: Average sentence 14 words. High active-voice ratio. Contractions present. Uses em-dashes for emphasis. ## Recurring Work - Weekly: ELT updates, programme RAG status, 1:1 coaching notes - Primary deliverable: Stakeholder briefings (300-600 words) - Templates: RAG-status board format, briefing template (saved) ## Goals & Priorities - Current quarter: Land Phase 2 of the £4M programme cleanly - Tools: Microsoft 365, Miro, Notion (personal), Claude - Biggest time sink: Stakeholder briefings — same questions, different audiences Save to CLAUDE.md or a memory file. Test on three different task types this week. Quick wins: stakeholder briefing drafts in your voice, weekly RAG status from notes, 1:1 prep questions.

Get this skill + 4 more

Get the full AI Memory pillar (5 skills) or the complete library.

Get the full stack — $299

What you get

  • 97-line SKILL.md, ready to drop into ~/.claude/skills/
  • Tested through 3 Karpathy-loop iterations (versions v1.0.0 → v1.3.0)
  • Triggers automatically when relevant — no command to remember
  • Lifetime updates as the skill is refined further

More from AI Memory

Browse the full library

297 skills across 31 categories. One purchase, lifetime updates.

See all bundles