Ascii Canvas provides a native visual interface for LLMs, enabling the creation of diagrams and UIs that are both human-inspectable and machine-readable as text. It offers an infinite canvas, multi-byte character support, and a copy-pasteable architecture, aiming to facilitate the construction of AI-interpretable visual assets.

How It Works

The project renders editable Unicode grids, moving beyond opaque pixels to support freeform drawing, structured UI composition, and frame-based ASCII animation. Its core innovation lies in structured nodes (text, backgrounds, boxes, split boxes, lines) that maintain semantic meaning, allowing for direct manipulation and styling, unlike flattened text representations. This approach ensures visual elements remain parseable by LLMs while offering a rich editing experience.

Quick Start & Requirements

Installation involves cloning the repository and running npm install. Development can be initiated with npm run dev, and builds are generated via npm run build. Standard Node.js development environment is assumed.

Highlighted Details

• Structured Editing: Compose scenes using semantic nodes, drag-and-drop components/templates, and style shapes and text ranges directly. Supports layer-aware backgrounds and resizable split layouts.
• Freeform Drawing: Features multi-layer rendering, grid-aware Unicode handling (including CJK, Emoji, Nerd Fonts), smart text flow, and precision selection tools.
• Animation Workflow: Offers fixed canvas presets, a frame timeline with onion skinning, and export options for JSON, GIF, asciinema .cast, and ANSI text.
• ANSI & Clipboard: Robust import/export for ANSI-styled terminal text, supporting various color modes and attributes. Paste operations handle plain text, rich clipboard data, and ANSI sequences.

Licensing & Compatibility

This project is licensed under the MIT License, which is permissive for commercial use and integration into closed-source projects.

Limitations & Caveats

Future development outlined in the roadmap includes NES (Next Edit Suggestion) for predictive character placement, AI Chat Integration for natural language canvas generation, and full ANSI terminal sequence workspace with SVG export support, indicating these features are not yet implemented.