Prompt Enhancer
Take any existing prompt and supercharge it using techniques from 2,500+ curated prompts.
Enhancement Workflow
Step 1: Analyze the Input Prompt
Evaluate the user's prompt on these dimensions:
- Specificity (1-5): How detailed is the subject description?
- Technical quality (1-5): Camera, lighting, composition details?
- Style clarity (1-5): Is the aesthetic clearly defined?
- Model optimization (1-5): Uses model-specific syntax correctly?
- Length appropriateness (1-5): Right length for the target model?
Present a brief score card before enhancing.
Step 2: Find Similar Top Prompts
Search for high-quality reference prompts:
python3 {PROMPT_ENGINE_DIR}/scripts/search_prompts.py "KEY_TERMS" --limit 5
Step 3: Apply Enhancement Techniques
Choose from these enhancement strategies based on what's missing:
Detail Injection (for low specificity):
- Add material textures ("brushed aluminum", "weathered leather")
- Add environmental details ("dust particles in light", "morning dew")
- Add character details ("freckled skin", "calloused hands")
Technical Elevation (for missing camera/lighting):
- Add camera specs ("shot on Canon R5, 85mm f/1.2")
- Add lighting ("golden hour backlighting", "Rembrandt lighting")
- Add film stocks ("Kodak Portra 400 colors", "Fuji Velvia saturation")
Style Anchoring (for unclear aesthetic):
- Add photographer/artist references ("in the style of Annie Leibovitz")
- Add film/era references ("Y2K aesthetic", "1970s Kodachrome")
- Add mood keywords ("moody", "ethereal", "gritty")
Negative Refinement (for models that support it):
- Add negative prompts to exclude unwanted elements
- Specify what NOT to include ("no text", "no watermark")
Structure Optimization (for poor organization):
- Reorder elements: subject first, then environment, then style
- Group related modifiers together
- Remove redundant or conflicting terms
Step 4: Present Enhanced Version
Output format:
- Original prompt (for comparison)
- Enhanced prompt (the improved version)
- What changed (bullet list of specific improvements)
- Score improvement (before -> after on 5-point scale)
- Variations (2-3 alternative enhanced versions)
Enhancement Rules
- Never remove the user's core intent/subject
- Preserve the user's preferred style if stated
- Keep enhancements relevant to the target model
- Don't over-engineer: some prompts are intentionally minimal
- Always explain WHY each change was made
Source: AgriciDaniel/claude-prompts — distributed by TomeVault.