Loading...
Documentation
Manage, search, and optimise every AI prompt in your library. The Prompt Library collects prompts from all your assets, and the AI Prompt Optimizer suggests concrete improvements using three proven strategies.
The Prompt Library gives you two primary workflows: browsing your collected prompts, and optimising them with AI assistance.
Optimise and choose a strategy.The gallery collects every prompt extracted from your imported assets. Each row displays the information you need to evaluate and reuse a prompt at a glance.
Prompt text
The full generation prompt, truncated in the list view with the complete text available on click.
Thumbnail
A preview of the asset generated from this prompt.
Parameter chips
Inline chips showing sampler, steps, CFG, and model name extracted from metadata.
Score indicators
Dot indicators representing quality or relevance scores for quick visual comparison.
Each row has a one-click copy button that places the full prompt text on your clipboard. The button provides visual confirmation when the copy succeeds.
Use the Arrow Up and Arrow Down keys to move between prompts without the mouse. Press Enter to open the selected prompt's detail view.
Parameter chips display key generation settings inline with each prompt. These values are extracted automatically from asset metadata during import.
euler_aThe sampling algorithm used during generation.
30Number of denoising steps applied.
7.5Classifier-free guidance scale controlling prompt adherence.
sd_xl_base_1.0The base model checkpoint used for generation.
Note: Not all assets include every parameter. Chips only appear when the corresponding metadata is present in the imported file. ComfyUI workflow JSON and Midjourney metadata are the most common sources.
Optimise to open the strategy picker.Varies key parameters (steps, CFG, sampler) systematically to find optimal settings for your prompt.
Best for: Fine-tuning a prompt you are happy with conceptually but want to improve technically.
Adds descriptive terms, style references, and quality modifiers to enrich the prompt.
Best for: Prompts that are too short or generic.
Generates targeted negative prompts to suppress common artefacts and unwanted elements.
Best for: Prompts producing unwanted visual elements.
Parameter Sweep
“I like the image but want it sharper or more detailed.”
Semantic Expansion
“My prompt is too vague; results are inconsistent.”
Negative Prompt
“I keep getting unwanted elements (extra fingers, blurry backgrounds).”
Not sure?
Start with Semantic Expansion for short prompts, or Parameter Sweep for prompts that already produce decent results.
Test how a prompt performs across different AI providers before committing to a full generation run. This helps you find the best model for a given prompt without wasting generation credits.
Preview to open the model selector.Tip: Multi-model preview is particularly useful when switching between providers (for example, comparing a Stable Diffusion checkpoint against a Flux model) to see which handles your prompt style better.
Per optimisation. Each time you run one of the three optimiser strategies on a prompt, 3 AI credits are deducted from your account balance. The cost is displayed on the confirmation step before you commit.
If your balance is below 3 credits when you attempt an optimisation, the InsufficientCreditsModal appears with a direct link to purchase more credits. You will not be charged unless you explicitly confirm a purchase.
AI credits are shared across all AI-powered features in Numonic: the AI Librarian, reflections, and prompt optimisation all draw from the same balance. Check your remaining credits on the account settings page.