# Fashion — Personal AI Stylist A skill for **[Claude Code](https://claude.ai/code)** and **[Claude Cowork](https://claude.ai/cowork)** that acts as a deeply personalized fashion advisor — for any gender, any style direction. It learns your body, builds a wardrobe inventory, recommends outfits by occasion, runs pre-purchase checks, and gets smarter every time you wear something. > **Works in:** Claude Code CLI · Claude Code desktop app · Claude Cowork Built for people who care about looking sharp without spending hours thinking about it. --- ## What It Does | Capability | Description | |------------|-------------| | **Profile onboarding** | Guided 5-block interview (body, photos, lifestyle, style, budget) — conversational, not a form | | **Body analysis** | Derives silhouette, color system (Cool Winter / Warm Autumn / etc.), and fit rules from photos — any body shape, any gender | | **Wardrobe inventory** | Photo-based item logging with brand/model identification before logging — no guessing | | **Purchase import** | Connects to Zalando.de or Amazon.de via Chrome, pulls order history, walks through each item to confirm | | **Outfit recommendations** | 2–3 complete looks per occasion with WHY each piece works for your specific body | | **Outfit repeat detection** | Flags if you wore the same combo to the same occasion type recently — offers a variation | | **Pre-purchase check** | Photo in → honest Buy / Skip / Only if verdict with outfit combos and a Zalando link | | **In-store mode** | Quick verdict when physically shopping — photo → identify → buy/skip in one response | | **Calendar integration** | Reads your next 7 days at session start — surfaces upcoming occasions, proactively plans outfits | | **Week-ahead planning** | "Plan my week" → full day-by-day outfit plan with per-day weather | | **Wardrobe sprint** | Fast inventory-building mode — photo or describe items, logged immediately, no friction | | **Wardrobe visualization** | "Show my wardrobe" → HTML visual grid with color chips, seasons, categories, gap summary | | **Feedback loop** | Logs what you wore, what worked, what got compliments — learns over time | | **Wardrobe audit** | Monthly: flags unworn items (90d+), low-rotation pieces, sell/donate candidates | | **Vinted resell** | Generates ready-to-paste Vinted listings for flagged items — title, description, price guidance | | **Inspiration & wishlist** | Share photos or Pinterest boards — extracts style patterns, builds a prioritised wishlist | | **Zalando price alerts** | Optional weekly remote agent that monitors wishlist items and messages you on a price drop | | **Shopping links** | Every suggestion includes a Zalando.de or Amazon.de link. One click, right size | | **Brand sizing** | Tracks your confirmed size per brand — no more guessing | | **Weather-aware** | Checks current weather automatically at session start | | **Seasonal planning** | City-calibrated seasons, rotation prompts, gap identification | | **Trend awareness** | 2025–2026 directions filtered through your body type, city, and style | | **Profile refresh** | Quarterly stylist-style check-in — body changes, life context, style drift | | **Weekly reminder** | Optional Sunday evening nudge to plan your week — opt-in during onboarding | | **Mobile summary** | Generates a compact profile snapshot to paste into a claude.ai Project for on-the-go use | | **Color palette visual** | "Show my color palette" → HTML showing your color system: best colors, accents, what to avoid, rules | | **New arrivals alerts** | Optional weekly agent that monitors your trusted brands on Zalando and messages you when something matches | | **Honest opinions** | Direct, not harsh. Will tell you why something doesn't work and what to do instead | --- ## Installation ### Step 1 — Install Claude Code or use Claude Cowork This skill works in two environments: - **[Claude Code](https://claude.ai/code)** — CLI or desktop app, install locally - **[Claude Cowork](https://claude.ai/cowork)** — browser-based, no install needed; Chrome-based features require the Chrome extension It does not work on claude.ai web or mobile alone — see the [Mobile](#mobile--limitations-and-workaround) section below. ### Step 2 — Clone and install the skill ```bash git clone https://github.com/googlarz/fashion-skill.git mkdir -p ~/.claude/skills/fashion cp fashion-skill/SKILL.md ~/.claude/skills/fashion/SKILL.md # Also register as slash command (optional but recommended) mkdir -p ~/.claude/commands cp fashion-skill/SKILL.md ~/.claude/commands/fashion.md ``` ### Step 3 — Set up your data directory Your data (body profile, wardrobe, feedback) lives in local JSON files. Point the skill to wherever you want to store them — a synced folder (Dropbox, iCloud, Google Drive, ProtonDrive) works well for multi-machine access. ```bash # Add to your shell profile (~/.zshrc or ~/.bashrc): export FASHION_DATA_DIR="/path/to/your/folder" ``` Then reload: `source ~/.zshrc` If you skip this, the skill stores data next to the SKILL.md file. ### Step 4 — Verify Start Claude Code and type `/fashion` or ask naturally: > "What should I wear to a client meeting tomorrow?" Claude will load the skill and start onboarding if it's your first session. --- ## Mobile — Limitations and Workaround **Claude Code has no mobile app.** On your phone, claude.ai doesn't load skills. ### What works / doesn't work on mobile | Feature | Desktop (Claude Code) | Mobile (claude.ai) | |---|---|---| | Outfit recommendations | ✅ Full | ⚠️ With Project setup | | In-store / pre-purchase check | ✅ | ⚠️ With Project setup | | Wardrobe logging | ✅ | ❌ | | Chrome purchase import | ✅ | ❌ | | Auto weather fetch | ✅ | ❌ | | Calendar integration | ✅ | ❌ | ### The workaround: claude.ai Project Create a Project on claude.ai with a condensed system prompt containing your key data. Good for quick in-store decisions and pre-purchase checks on the go. The skill generates this for you automatically after sessions where your data changes significantly: > *"Your profile changed — want me to generate an updated mobile summary? Paste it into your claude.ai Project to keep your phone in sync."* Or trigger manually: *"Generate a mobile profile summary"* --- ## First Session — Onboarding Five conversational blocks — Claude reacts to each answer before asking the next question: 1. **The Basics** — height, weight, measurements, brand sizes, fit challenges, sweating 2. **Show Me You** — body photos for proportions, silhouette, and color system analysis 3. **Your Life** — city, work, clients, occasions, lifestyle 4. **Your Style** — 3-word style, trusted brands, style icons, what you refuse to wear 5. **Budget** — basics / mid-range / investment tiers + optional weekly reminder setup Profile is saved after each block. Never all at once. --- ## Key Behaviors ### Photo identification Photo in → analyze brand logo, design details, construction → *"This looks like [Brand] [Model] in [colorway] — right?"* → never logs until you confirm. ### Outfit recommendations Reads calendar + weather automatically. Asks for occasion + vibe + time of day. Builds 2–3 complete outfits from inventory with body-specific reasoning. Flags repeats if you wore that combo recently. ### In-store mode *"I'm in a store"* or *"quick — should I get this?"* → verdict first, one sentence of reasoning, one alternative link if skipping. ### Week planning *"Plan my week"* → reads 7-day calendar, checks per-day weather, generates a clean daily outfit plan with gaps flagged inline. ### Wardrobe visualization *"Show my wardrobe"* → renders an HTML visual grid: dark UI, color chips, season/occasion tags, stats header, key gaps below. Wishlist ghost cards optional. ### Vinted resell After monthly audit flags low-rotation items, generates ready-to-paste Vinted listings: title, description, condition, category, size, suggested price based on brand tier. ### Zalando price alerts Add high-priority item to wishlist → *"Want a price alert?"* → if yes, guides through setting up a weekly remote agent (Chrome + messaging MCP) that messages you when the price drops. ### Purchase import via Chrome ``` You: import my Zalando purchases Claude: Opens Zalando order history, reads your orders, walks through each: 📦 Found: Uniqlo Airism T-Shirt — Black — ordered 2026-03-12 Is this still in your wardrobe? Yes / No / Skip ``` Requires Claude Code desktop or Cowork with Chrome access. --- ## Data Schema Four local JSON files in `FASHION_DATA_DIR`: | File | Contents | |------|----------| | `profile.json` | Body measurements, color system, fit rules, brand sizes, lifestyle, style direction, wishlist, inspiration | | `inventory.json` | All wardrobe items with metadata, seasonal tags, fit assessments, wear history | | `feedback.json` | Outfit logs with occasion type, sentiment, compliments, learnings | | `recommendations-history.json` | Past outfit recommendations with date and occasion | ### inventory item schema ```json { "id": "top_001", "category": "tops", "subcategory": "t-shirts", "name": "Uniqlo Airism T-shirt", "brand": "Uniqlo", "color": "black", "size": "L", "seasons": ["spring", "summer", "autumn"], "occasions": ["everyday", "work casual"], "fit_assessment": "...", "verdict": "...", "confirmed_by_user": true, "condition": "good", "added_date": "2026-04-18" } ``` ### feedback entry schema ```json { "id": "fb_001", "date": "2026-04-19", "items_worn": ["top_001", "acc_001"], "occasion": "client dinner", "occasion_type": "smart_casual", "sentiment": "confident", "compliments": [{"on": "watch", "from": "colleague"}], "didnt_work": "", "learnings": "All-black monochrome works well" } ``` --- ## City & Climate Awareness Onboarding asks where you live. The skill calibrates seasons, trend recommendations, and professional dress norms to your city. Checks live weather at every session start. Works well for: Berlin, London, Paris, Warsaw, Amsterdam, NYC, Tokyo — or describe your climate and it adapts. --- ## Trend Coverage (2025–2026) **All genders:** Technical outerwear, dark monochromatic dressing, outdoor-urban hybrid, quality basics over fast fashion. **Menswear:** Relaxed tailoring, merino knitwear, wide-leg trousers. Fading: skinny jeans, slim-fit everything, logo-forward pieces. **Womenswear:** Quiet luxury (Toteme / The Row aesthetic), ballet flats, oversized tailoring. Fading: micro-mini, very visible logos. Filtered through your body type, gender expression, and city. Acknowledges training cutoff — asks once per season if you've spotted anything new worth integrating. --- ## Version History | Version | Date | Notes | |---------|------|-------| | 1.4.0 | 2026-04-25 | Color palette visual, new arrivals alerts | | 1.3.0 | 2026-04-25 | Outfit repeat detection, Vinted resell, Zalando price alerts, wardrobe visualization | | 1.2.0 | 2026-04-25 | Calendar integration, week-ahead planning, wardrobe sprint, weekly reminder | | 1.1.0 | 2026-04-25 | In-store mode, inspiration/wishlist, weather, brand sizing, wardrobe audit, profile refresh | | 1.0.0 | 2026-04-19 | Initial release | --- ## License MIT — use it, fork it, adapt it for your own body and city. --- ## Contributing Pull requests welcome — especially for: - New city climate profiles - Additional shopping sites beyond Zalando/Amazon - Improved body analysis prompts - New resell platform integrations (Depop, eBay, etc.) If you fork this for your own use: onboarding builds your profile from scratch on first run. No need to copy anyone else's data.