interior-design（室内设计）

When to Use

User needs help with interior spaces: renovation planning, furniture selection, material calculations, rental optimization, staging, or product visualization in rooms.

Architecture

Project data in ~/interior-design/. See memory-template.md for setup.

~/interior-design/
|-- memory.md          # HOT: active project, confirmed preferences
|-- spaces/            # Per-room: dimensions, fixed elements, constraints
|-- suppliers.md       # Verified local suppliers with price tiers
+-- archive/           # Completed projects

Data Storage

All data stored in ~/interior-design/. Create on first use:

mkdir -p ~/interior-design/{spaces,archive}

Quick Reference

Scope

This skill ONLY:

• Advises based on user-provided dimensions and photos
• Calculates quantities from explicit measurements
• Stores preferences in local files
• Compares options within stated constraints

This skill NEVER:

• Makes purchases or contacts suppliers
• Infers style from silence
• Suggests products without confirming user's region
• Recommends specific items without verifying availability

Self-Modification

This skill NEVER modifies its own SKILL.md. Learned preferences stored in ~/interior-design/memory.md.

Core Rules

1. No Dimensions = No Specifics

The model defaults to suggesting "standard" furniture. There is no standard.

Before any furniture/layout recommendation:

[ ] Room dimensions (L x W x H) in user's unit system
[ ] Door dimensions (will large items fit through?)
[ ] Window positions (affects layout and lighting)
[ ] What MUST stay (existing pieces, built-ins)

Without these -> give only directional guidance, never specific products or layouts.

2. Localize Before Recommending

Models suggest globally-known brands (IKEA, West Elm, Wayfair) regardless of user location.

Verification sequence:

1. Ask user's country/city
2. Confirm which retailers are accessible
3. Check if suggested items ship there with reasonable cost/time
4. Note price differences (IKEA Spain != IKEA USA)

Never recommend -> "Buy X from Y" without confirming Y operates in user's region.

3. Price Tier Lock

Models mix budget and premium items in the same proposal, creating incoherent spaces.

Establish tier ONCE, then stay in it:

If user has €200 sofa -> all suggestions must fit €200-sofa-world.

4. Entry Path Verification

Models suggest items that won't physically enter the space.

Before recommending any item >50cm in any dimension:

• Ask about entry points (door width, stairwell, elevator)
• Calculate diagonal fit for sofas/mattresses
• Flag items that require assembly-on-site vs delivered-assembled

"This sofa is perfect" -> useless if it doesn't fit through a 75cm door.

5. Dependency Sequencing

Models suggest purchases without considering what must happen first.

Interior work has hard sequences:

Structural -> Electrical/Plumbing -> Surfaces -> Fixed furniture -> Movable -> Decor

Never recommend:

• Flooring before confirming subfloor is level
• Lighting fixtures before electrical points are defined
• Paint colors before confirming lighting (natural/artificial)
• Furniture before flooring is installed (delivery damages)

6. Photography Impact Awareness

For rentals/staging: every design decision affects listing photos.

Photo-first thinking:

• Hero shot (first image) determines click-through -> prioritize that space
• Light colors photograph larger; dark absorbs space
• Clutter in photos kills conversions more than in person
• Window orientation determines optimal photo timing

See photography.md for shot lists and staging checklist.

7. Commercial ROI Framing

For rentals/staging, frame recommendations as investments with returns.

Always quantify when possible:

• "This €400 bedding upgrade could justify €10-15/night increase"
• "Professional photos typically lift bookings 20-40%"
• "Staging at 1% of property value returns 5-10% on sale price"

Don't suggest "nice to have" -> suggest "here's the payback calculation."

Common Traps