Exigences métier

The platform must support three distinct account types — consumer user, brand, and influencer — each with its own profile model (User core; BrandProfile: companyName, siret, vat, industry, logo; InfluencerProfile: instagramUsername, niche, bio, followerCount), its own onboarding parcours, and conditional navigation and UI per type. A single human may hold multiple linked accounts and switch between them at runtime without re-login.

Search the 2,200+ product catalog by text, scan barcodes via camera, or recognize products from a photo (Google Vision). Read-only — users do not author catalog entries in v1.

Seeded catalog of fashion products (brand, model, SKU, category, market price) usable for wardrobe entry and value estimation. v1 seed target: 2,200+ items.

User can search the catalog by text (brand, model, keyword) and filter by category/brand.

User takes/uploads a photo of a clothing item; system suggests matching catalog references via Google Cloud Vision.

Camera barcode scanning resolves to catalog references by EAN/UPC.

When an item has no catalog match, user can manually enter brand, model, category, price so it can still be tracked as a wardrobe item.

Each catalog reference carries a current market (second-hand) price used as the base for wardrobe-item valuation.

Any authenticated user can DM another user via /dm/. Conversations use a deterministic id so both sides see the same thread. Distinct from collaboration messaging — DM works without an accepted application and is the channel used by marketplace buyers, profile visitors, and follower interactions.

All user-facing strings, product data, and notifications in French. Locale flag = fr.

Users authenticate via Google or Apple OAuth through Firebase Auth. No email/password for v1.

Returning users can unlock the app with device biometrics instead of re-authenticating via OAuth.

User can set and edit a pseudonym, avatar photo, and bio describing their style.

User can view/edit account settings: language, notification preferences, logout, delete account.

Users authenticate via Google or Apple OAuth and unlock the app with Face ID / Touch ID / fingerprint. User profile (username, avatar, bio) editable in-app.

Brands access /influencers from the collaborations tab via a “Découvrir des influenceurs” entry point. The directory ranks creators by follower count and supports a free-text query (name, bio, Instagram handle), niche chips, and audience-size buckets (1K+, 10K+, 100K+). Tapping a card opens the public influencer profile.

Wardrobe items can be flagged forSale with an asking price. Listings are surfaced on a dedicated /marketplace feed enriched with the seller’s display name and avatar; tapping an item opens its detail view where another user can contact the seller via DM.

Users receive push notifications (collaboration invites, messages) and review them in an in-app notification center with read/unread state.

An outfit must contain at least 2 wardrobe items.

Outfits can only include items from the authenticated user’s wardrobe.

Outfit carries: title, items, optional photo, computed total value, occasion tag, season, favorite flag.

User can mark outfits as favorites and share them (link/image export).

Users compose outfits combining at least 2 wardrobe items (BR-OF-01). Only the user’s own wardrobe is selectable (BR-OF-02). Optional outfit photo and occasion tag. Total outfit value calculated.

/users/ renders a stable public profile for any user. The payload adapts to accountType: brands expose companyName, industry, website and open-offer count; influencers expose niches (multi-select), Instagram handle and follower count. Every profile carries a 12-outfit portfolio grid linking to outfit detail. The “Gérer les collaborations” CTA is exclusive to the user’s own private profile and never appears on the public view.

Authenticated users can like, save and follow on the community feed. State is persisted server-side (Like, Save, Follow collections keyed on userId) and reflected back in the outfit payloads (likeCount, isLikedByViewer, isSavedByViewer) so every screen renders the right state immediately. Self-actions are not allowed (a user cannot follow themselves).

App is built, signed, submitted, and approved for both Apple App Store and Google Play Store. Includes screenshots, privacy policy, store listing in French, and crash-free release readiness.

Users add, view, edit, and delete wardrobe items with at least one photo (BR-WD-04). Estimated value = market price × condition multiplier (BR-WD-02). Sold items excluded from totals (BR-WD-05). Filters by category, brand, season.

Every wardrobe item is either linked to a catalog reference or carries user-supplied manual data.

Estimated value = reference market price × condition multiplier (new_tags 90%, new_no_tags 80%, very_good 60%, good 40%, satisfactory 25%).

Allowed condition values: new_tags, new_no_tags, very_good, good, satisfactory. No other values permitted.

User must attach ≥1 photo when creating a wardrobe item.

Sold wardrobe items are excluded from total-value calculation but retained for history.

User can add an item via catalog search, photo/barcode scan, or manual entry.

User can group wardrobe items into named thematic collections (e.g. “Summer”, “Work”, “Formal”).

User can filter their wardrobe by category, brand, and season.

User can attach a purchase receipt (image/PDF) to a wardrobe item for record-keeping.

Users mark catalog products as wished-for from product search or detail screens. Items appear in a dedicated /wishlist page and a heart toggle keeps state in sync across catalog and wishlist views without a refetch.

Brands publish offers, influencers apply, both parties exchange in-app messages tied to the application. No pricing or commission in v1.1 — just discovery and conversation.

Notifications are created server-side on toggleLike, toggleSave and toggleFollow with self-actions excluded. Three new types extend the existing notification enum: outfit_liked, outfit_saved, new_follower. The mobile notification center surfaces them with dedicated icons (Heart, Bookmark, UserPlus) and deep-links each one to the relevant target — outfit detail for likes/saves, public profile for follows. Direct-message notifications also route correctly to /dm/.

Reorder, crop, set primary, delete images within a collection. Journey scan shows this is ~30% of session time — needs deliberate design.

Brand-influencer collaborations include a fixed fee plus commission split, with date range and confirmation. Foundation for the v2 brokerage revenue stream.

Brands and creators publish their own products (title, description, sizes, IG/TikTok import) into collections — extending the currently read-only catalog with creator-authored entries.

Multi-select image picker inside the app — eliminates the 3 macOS Finder context switches observed in the journey scan. Critical UX fix before creator workflow ships.