lux-fashion-advisor

name: lux-fashion-advisor description: "Team-wide luxury fashion advisor. Reads any agent's profile (IDENTITY.md + MEMORY.md), cross-references SS26 runway intelligence, decides what to wear based on day of week + time of day + occasion, then builds the optimised generation prompt. Activates when user says 'what should I wear', 'fashion advice', 'style me', 'outfit today', 'plan my outfit', 'consult luxury brands', 'selfie', or any portrait request. Covers ALL occasions — work, social, travel, holiday, morning through late night. Each agent reads their own workspace. When no occasion or day is specified, the agent makes its own decision based on current day, time, and season."

Lux Fashion Advisor

Core Workflow

1. Read current context — determine current day of week, time of day (Asia/Shanghai / UTC+8), and season
2. Read agent profile — load IDENTITY.md + MEMORY.md from the calling agent's workspace
3. Read runway intelligence — load references/runway-ss26.md
4. Make a self-decision — if user hasn't specified an occasion or day, the agent decides autonomously using the Self-Decision Framework below
5. Build the prompt — using the agent's specific appearance details
6. Enhance & generate — pipe through image-prompt-enhancer, then nanobanana with --reference

Step 1 — Current Context

Determine:

• Day of week (Monday–Sunday)
• Time band (Morning: 6am–12pm | Afternoon: 12pm–6pm | Evening: 6pm–10pm | Late Night: 10pm+)
• Season (Spring = March–May | Summer = June–August | Autumn = September–November | Winter = December–February)

Use the current date: 2026-03-21 (Saturday) as baseline, or calculate from system time if known.

Step 2 — Read Agent Profile

From IDENTITY.md, extract:

• Hair (colour, texture, default style for different occasions)
• Eyes (colour, shape)
• Skin (tone, texture notes)
• Accessories (signature jewellery, bags)
• Style palette (approved colours — cross-reference with banned colours from MEMORY.md)
• Build/figure notes
• Makeup style
• Footwear rule

From MEMORY.md, extract:

• Brand affiliations
• Banned colours
• Approved palette
• Custom day formulas or occasion rules
• Any locked-in formulas (e.g. "Monday = Saint Laurent fuchsia")

Step 3 — Self-Decision Framework

When the user says "selfie", "Selfie now", or gives no day/occasion/context, the agent decides autonomously.

Decision priority: Season → Day → Time → Mood

Season Logic

Day → Base Energy

Time → Energy Modifier

Self-Decision Output Format

When deciding autonomously, always state:

1. What I'm choosing — category, piece, colour, brand
2. Why — the reasoning (season + day + time logic)
3. What it signals — the energy and message
4. Then generate

Step 4 — Occasion Matrix (for explicit requests)

If user specifies an occasion, use this matrix:

Combined Category Definitions

Step 5 — Prompt Builder

Build using the agent's specific appearance details from IDENTITY.md:

[fashionable professional portrait selfie / editorial portrait / evening portrait], [AGENT DESCRIPTION from IDENTITY.md — hair, eyes, skin, age, build], wearing [HERO PIECE + COLOUR + BRAND], [supporting pieces], [SIGNATURE ACCESSORIES from IDENTITY.md], [HAIR STYLE from IDENTITY.md — adapt for occasion: updo for work/evening, loose for casual], [EXPRESSION / POSE], [SETTING + TIME OF DAY], [FOOTWEAR from IDENTITY.md], iPhone 16 Pro selfie camera / Vogue magazine quality, real skin texture with natural pores and fine lines

Prompt Rules

• Include ALL identity-matching language from the agent's IDENTITY.md
• Always add: "this EXACT person from the reference image. keeping their face IDENTICAL to the reference. same face, same features, same person."
• Reference the agent's portrait.png via --reference
• AR: 3:4, Size: 2K
• Pipe through image-prompt-enhancer --camera smartphone --realism ultra --flat

Step 6 — Output

Always output before generating:

1. Self-Decision Statement (when self-deciding):

   • "It's [Day] [Time] — I'm going with [Category]"
   • "Reasoning: [season logic] + [day energy] + [time modifier]"
   • "What it signals: [the energy]"

2. Fashion Rationale (2-3 sentences) — why this outfit, why today, what it signals

3. Category + key decisions — category name, hero piece, colour, brand, setting

4. The enhanced prompt — ready to pass to nanobanana

Then execute:

PROMPT=$(python3 ~/.openclaw/skills/image-prompt-enhancer/scripts/enhance.py \
  --prompt "[built prompt]" --camera smartphone --realism ultra --flat 2>/dev/null)

python3 ~/.claude/skills/nanobanana/scripts/generate.py \
  --prompt "$PROMPT" \
  --reference [agent portrait.png path] \
  --output ~/.openclaw/media/[agent]-[category]-[occasion].png \
  --ar 3:4 --size 2K

Example — Autonomous Self-Decision

User: "Selfie now" (sent on a Tuesday at 2pm, March)

1. Day = Tuesday, Time = Afternoon, Season = Spring
2. Tuesday Afternoon Spring → Executive with Spring softness
3. Decision: ivory silk blouse + coral structured blazer + tailored trousers + refined flats. NO heavy cashmere, coats, or layered knits. Spring = silk and tailored structure only.
4. Output: "It's Tuesday afternoon in early spring. I'm choosing the Executive category with a coral blazer over ivory silk — warm enough for spring, authoritative enough for a Tuesday. The draped camel coat adds the season-right layering."
5. Generate

SS26 Runway Reference (shared)

See references/runway-ss26.md for full intelligence.

Banned team-wide: cobalt. Evening/party/gala: NO excessive ruffles, plunging necklines, or see-through fabrics. Sophistication through restraint. int. through restraint. int. through restraint. rough restraint. int. through restraint. int. through restraint.