Every capability, built around one idea: understand the prompt behind the pixels.
All features below are real extension capabilities. Plan availability is enforced by the backend and shown live.
Three depths, three levels of fidelity
Analysis depths control how much detail the model extracts - and how long the prompt can be.
Standard - Nano Lite
Fast and concise. A clean prompt of up to 1,000 characters. Available on all plans.
Deep - Nano Spark
Detailed analysis with up to 2,000 characters. Captures composition, lighting and style cues. Pro and Ultra.
Ultra - Spark Ultra
The Visual DNA engine - 14 structured analysis sections (canvas, spatial geometry, subject, face, hair, clothing, camera, lighting, color DNA, typography + occlusion, layers, style) compiled into a six-part document: reconstruction summary, Visual DNA blueprint, spatial layout map, layer/occlusion map, master prompt up to 20,000 characters, and generator-specific versions (GPT Image / Midjourney / FLUX / Ideogram) - with a reconstruction gap audit. Ultra only.
API key management
Add your own OpenRouter keys with per-key model selection. Automatic failover between keys and models keeps extraction working when one provider is congested.
Chat editor
Refine any extracted prompt in a built-in chat. Modify specific parts or regenerate entirely - only the requested parts are rewritten.
Prompt library
Save prompts and reuse them instantly. Storage is local to your browser. Library size depends on your plan.
History & saved chats
Every extraction and chat is kept in your recent activity - local, searchable, private to you.
Format converter
Convert a prompt between output formats in one click - ideal for moving between tools and workflows.
Multi-language UI
The extension interface ships in five languages with model cards, settings and help localized.