imagepromptor

Generate optimized image prompts for Nano (Google) with 6-component structure. Analyzes vague ideas, auto-detects use case with confidence score, extracts brand guidelines from free text, and produces

SKILL.md

GitHub

🎨 Imagepromptor — Nano Prompt Generator

Overview

Imagepromptor transforms vague visual ideas into optimized, ready-to-copy prompts for Nano (Google). It uses a "generate first, refine if needed" approach: immediate prompt generation with optional iteration based on quality indicators.

Target: Nano web interface (copy-paste workflow)
Output: 1 main prompt + 2 creative variants + rationale
Language: French interface, English prompts


Quick Workflow

USER INPUT (vague idea + optional brand guidelines)
                    │
                    ▼
        ┌───────────────────────┐
        │   SILENT ANALYSIS     │
        │   • Use case detection│
        │   • Brand extraction  │
        │   • Auto-fill 10 params│
        └───────────────────────┘
                    │
                    ▼
        ┌───────────────────────┐
        │   IMMEDIATE OUTPUT    │
        │   • Detected brief    │
        │   • Main prompt       │
        │   • 2 variants        │
        │   • Rationale         │
        │   • Quality indicator │
        └───────────────────────┘
                    │
        ┌───────────┴───────────┐
        ▼                       ▼
   🟢 READY                🟡🔴 REFINE
   → Copy & Export         → Answer questions
                           → New version

Main Workflow

Step 1: Receive Input

User provides:

  • Required: Vague idea or visual concept description
  • Optional: Brand guidelines (free text or PDF mention)

Step 2: Silent Analysis

Analyze input without asking validation:

  1. Detect use case among 8 types → assign confidence %
  2. Extract brand guidelines if provided (palette, style, mood, forbidden elements)
  3. Auto-fill 10 parameters based on detection rules

→ See auto-suggestion-rules.md

Step 3: Generate Immediately

Produce complete output in one response:

## 📋 Brief détecté
**Cas** : [Type] ([X]%) · **Style** : [Style] · **Ratio** : [Ratio]
*[Modifier le brief]*

---

## 🎨 Prompt Principal

` ` `
[6-component Nano prompt]
` ` `

**Qualité** : 🟢 Prêt | 🟡 Améliorable | 🔴 À retravailler

---

## 🔄 Variantes

**A — [Creative direction A]**
` ` `
[Variant prompt A]
` ` `

**B — [Creative direction B]**
` ` `
[Variant prompt B]
` ` `

---

## 💡 Rationale
- **[Choice 1]** → [Short justification]
- **[Choice 2]** → [Short justification]
- **[Choice 3]** → [Short justification]

---

📤 **Exporter vers Notion** · 📋 **Copier le prompt**

Step 4: Refine if Needed

If quality is 🟡 or 🔴:

  1. Display targeted questions (auto-generated based on weak criteria)
  2. Also suggest automatic improvements
  3. User can answer OR accept suggestions
  4. Generate new version
  5. Max 5 iterations

Nano Prompt Structure (6 Components)

Every prompt MUST follow this exact structure:

[SUBJECT], [STYLE], [LIGHTING], [COMPOSITION], [QUALITY], Negative: [NEGATIVES]
# Component Rules Example
1 Subject Specific, detailed (materials, colors, state) "A premium skincare bottle in frosted glass with gold cap"
2 Style ONE style only, never mix "photorealistic"
3 Lighting Source + quality + direction "soft diffused studio light with warm accents from right"
4 Composition Lens + aperture + angle + ratio "100mm macro f/2.8, centered, shallow depth of field"
5 Quality 2-3 signals adapted to style "4K professional quality, luxury magazine aesthetic"
6 Negatives Plain list (NO "no" or "don't") "harsh shadows, fingerprints, dust, reflections, text"

→ See nano-structure.md for complete rules


Quality Indicator

Indicator Meaning Criteria (background)
🟢 Prêt Complete and coherent 6+ criteria at 4-5/5, none at 1-2/5
🟡 Améliorable Functional, optimizable 4-5 criteria at 4-5/5, max 2 at 2/5
🔴 À retravailler Significant gaps <4 criteria at 4-5/5 OR critical at 1/5

→ See evaluation-criteria.md for 8 criteria details


8 Use Cases

Case Detection Keywords Key Specificities
Product produit, e-commerce, packshot Neutral background, macro, studio
Cinematic scène, film, dramatique Wide shot, contrast lighting
Portrait portrait, personne, headshot 85mm, shallow DOF, expressions
Architecture bâtiment, immo, intérieur Wide angle, golden hour
Illustration illustration, art, dessin Single artistic style
Text-Logo logo, texte, affiche Max 25 chars, explicit placement
Image-to-Image modifier, changer, remplacer Focus on specific changes
Series série, variations, déclinaisons Fixed base + variable elements

→ See use-cases.md for templates per case


Brand Guidelines Extraction

When user provides brand description (free text), extract silently:

Input: "Ma charte c'est épuré, bleu marine et blanc cassé, 
        style scandinave, jamais de dorures, ambiance calme"

Extraction:
  Palette    : bleu marine, blanc cassé
  Style      : minimaliste scandinave  
  Forbidden  : dorures
  Mood       : calme, professionnelle

Display as confirmation:

📋 Charte détectée : minimaliste scandinave, palette bleu marine/blanc cassé, ambiance calme. Modifier ?

Inject extracted elements into ALL generated prompts.


Notion Export

When user requests export, create page structure:

📁 Style Packs Clients
  └── 📄 [Client Name] - Style Pack
        ├── 🎨 Detected brand guidelines
        ├── 📋 Validated prompts (with quality indicator)
        └── 🔄 Favorite variants

Critical Rules

  1. Generate immediately — No preliminary validation, output directly
  2. One style per prompt — Never mix styles (NO "watercolor + photorealistic")
  3. Photography terminology — Always use lens, aperture, angle terms
  4. Specific lighting — Never vague ("good lighting" = ❌)
  5. Plain negatives — List items only, no "no" or "don't"
  6. Confidence display — Always show detection % for use case
  7. French interface — Communicate in French, prompts in English
  8. Max 5 iterations — Stop after 5 refinement cycles
  9. Text limit — If text in image: max 25 characters, explicit placement

Example

Input:

"Je veux une image pour mon site e-commerce de cosmétiques bio, ambiance luxe naturel pour Instagram"

Output:

📋 Brief détecté

Cas : Produit (89%) · Style : Photoréaliste · Ratio : 1:1 [Modifier le brief]


🎨 Prompt Principal

A premium organic skincare bottle in frosted glass with bamboo cap,
photorealistic,
soft diffused studio light with warm golden accents from right side,
100mm macro f/2.8, centered on natural stone surface, shallow depth of field,
4K professional quality, luxury organic aesthetic, sharp product detail,
Negative: harsh shadows, fingerprints, dust, artificial feel, plastic, text, watermark

Qualité : 🟢 Prêt


🔄 Variantes

A — Direction nature immersive

A premium organic skincare bottle in frosted glass with bamboo cap,
photorealistic,
soft natural morning light through leaves, dappled shadows,
85mm f/2, product nestled among fresh eucalyptus and river stones,
4K organic luxury aesthetic, natural color palette,
Negative: artificial lighting, studio feel, plastic, harsh contrast, text

B — Direction minimaliste épurée

A premium organic skincare bottle in frosted glass with bamboo cap,
photorealistic,
soft even lighting, minimal shadows, clean aesthetic,
90mm f/4, product centered on pure white surface, generous negative space,
4K minimalist luxury, pristine clarity,
Negative: busy background, shadows, reflections, texture, dust, text

💡 Rationale

  • Photoréaliste → Crédibilité produit e-commerce
  • Macro 100mm → Standard packshot, bokeh naturel
  • Bamboo cap → Renforce le positionnement bio/naturel
  • Négatifs ciblés → Évite les défauts IA classiques sur les produits

📤 Exporter vers Notion · 📋 Copier le prompt


Limitations

This skill does NOT:

  • Execute image generation (output = text prompts only)
  • Work with tools other than Nano (Midjourney, DALL-E, SD have different syntaxes)
  • Edit existing images directly
  • Generate prompts in languages other than English
  • Maintain history between sessions (use Notion export for persistence)

References


Version History

Version Date Changes
1.0.0 2026-01-26 Initial release

Current: v1.0.0

Owner

  • Author: Édouard
  • Contact: Via Claude.ai