Breaking

lnkiai launches M3E Canvas to sketch Material 3 Expressive screens and export AI coding prompts

September 13, 2026 at 10:03 EDT

Software Dev & Coding
Open Source
AI Agents

M3E Canvas, an MIT-licensed browser editor from GitHub user lnkiai, lets developers assemble Material 3 Expressive screens visually and copy a detailed prompt for AI coding tools such as Claude Code, Codex, Gemini CLI, or Cursor. The app is a static site with no accounts and no backend; sketches stay in the browser until the user pastes the generated brief into a coding assistant.

Sept 13, 2026 · lnkiai

From sketch to coding prompt

M3E Canvas is a client-side editor that assembles Material 3 Expressive screens in the browser and writes a paste-ready prompt for Cursor, Codex, Claude Code, or Gemini CLI.

6,434

GitHub stars

654

GitHub forks

Color presets

Prompt languages

Key UI found up to 4× faster

Google research on Material 3 Expressive screens, not a benchmark of the canvas itself.

1×

Baseline

4×

Expressive

Takeaway: the design language M3E Canvas targets was built so people find controls faster — the editor’s job is to get that look into an AI prompt.

Frame width on one canvas

412

Phone

1280

Desktop

Same-named screens go into the prompt as one layout at two widths. Nav bars become rails on desktop.

How the handoff works

Sketch parts

→

Link and preview

→

Copy the prompt

→

Paste into the tool

What works

Visual layout, tap-through flows, and a prompt aimed at Claude Code, Codex, Cursor, or Gemini CLI — without an account or install.

What is thin

Look stays close to baseline M3 Expressive. Cards and nav need more variants; system UI is missing; the phone editor is buttons only.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.