Select the exact scope
Choose a slide or semantic block. Every selection resolves to stable IDs—not pixels, coordinates, or guesswork.
Local-first presentation authoring
Make presentations with an agent. Keep control of every change.
Xuzora gives one person a precise visual editing loop: select the content, write an intent, and let a coding agent patch only the scope you chose.
Readable files · Stable IDs · No model API in the core
The workflow
Direct the change,
not the whole deck.
State the change
Turn your selection into a structured edit intent that keeps the instruction tied to the content you meant.
Review the patch
Your coding agent patches the readable .xuz file. The browser reflects the change without regenerating the deck.
Permanent boundaries
Built around a file you can keep.
Xuzora separates durable presentation state from editing tools and agent intelligence. The format remains readable, diffable, and independently renderable.
01
File
is the state
02
CLI
is the interface
03
Web
is the editor
04
Agent
is the intelligence
V0.1 · Active development
The core interaction is taking shape.
Xuzora currently validates readable presentation files, renders four semantic layouts, supports persisted inline text editing, and resolves visual selection to stable slide and block IDs.
The complete local agent loop, presenter mode, PDF export, and the remaining layouts are still in development.
Done
Schema and CLI validation
Done
Theme and four base layouts
Done
Inline editing and stable selection
Now
Select-to-intent workflow
Next
Local agent loop and presentation tools