YANA STUDIO 1.5

One workspace.
The whole AI loop.

From goal to code, terminal, review and evidence — every step lives on one clear surface, with authority never hidden behind a chat box.

Plan · 4 steps
Analyze → Fix → Test → Review
Yana wants to run
npm test · Allow once
Hand a goal to Yana…
+ Files · Code · Terminal · Governed

REAL APP SCREENSHOTS

Not a mockup — this is Studio 1.5.1 actually running.

Two screenshots taken directly from the latest build, crossfading every few seconds.

Yana Studio — Chat screen, Vietnamese locale Yana Studio — Inspector, Korean locale

WORKSPACE

Real code and terminal

CodeMirror, project-wide file search, multi-tab PTY, split panes, search and scrollback.

workspace/ ├─ app ├─ components └─ tests terminal 3 sessions

CONTEXT

The AI knows what it's looking at

Attached files, project memory, Git, terminal and token budget are all visible, not hidden.

CONTEXT 3 files 1 terminal 34 project facts 18.2K / 128K

REVIEW

Changes you can actually inspect

Inline diff comments, an activity timeline and the permission surface before you accept anything.

AI CHANGES 4 files M App.tsx +42 -13 A Login.tsx +87 [Review] [Revert]

MODEL PLANE

Pick a provider, see the real models.

The official catalog is split by cloud and local provider so the UI never guesses a model name. Changing model never changes the runtime's authority.

OpenAIavailable models
Anthropicavailable models
Groqavailable models
Ollamalocal models

DESIGN CANVAS

Visual design is a Studio capability.

Canvas is experimental today; the direction is Design → Code → Run → Inspect, not a disconnected editor bolted on the side.