Opens .fig & .pen files
Drop in a Figma export or a native .pen document. Knighn parses both formats itself and renders them with CanvasKit — pixel-accurate, with no lossy import step.
Open source, MIT licensed
Knighn is a fast, open design editor that opens .fig and .pen files natively. Edit on a CanvasKit-powered canvas, delegate the busywork to a built-in AI copilot with 90+ design tools, and keep everything in a personal cloud your agents can reach from anywhere.
Runs in your browser as a PWA — nothing to install.
Everything you expect from a modern editor, plus the plumbing that makes your files legible to software — and to agents.
Drop in a Figma export or a native .pen document. Knighn parses both formats itself and renders them with CanvasKit — pixel-accurate, with no lossy import step.
A built-in chat copilot edits the open file through 90+ purpose-built design tools. Create frames, restyle components, and refactor layer trees by describing what you want.
Files and assets sync to your own Convex deployment. Every device sees the same library, and your work never has to live on someone else's SaaS.
Your cloud exposes a remote MCP endpoint. Hand Claude — or any MCP-capable agent — an API key and it can browse, search, and inspect your design files from anywhere.
A real CLI, XPath-style queries over the scene graph, and headless rendering. Script your design files the same way you script your codebase.
One editor across every screen: an installable web PWA, a native desktop app built on Tauri, and an iOS client for reviewing on the go.
Remote MCP
Your design library becomes a first-class tool surface for AI agents. Three steps and Claude can read your files.
Generate a key from your Knighn account settings. Each key scopes an agent to your personal cloud and can be revoked at any time.
Add the MCP endpoint of your Convex deployment to Claude or any other MCP-capable client — one URL, one header.
Connected agents can list projects, search layers, inspect nodes, and fetch assets — while every request stays scoped to your account.
{
"mcpServers": {
"knighn": {
"type": "http",
"url": "https://<your-deployment>.convex.site/mcp",
"headers": {
"Authorization": "Bearer <your-api-key>"
}
}
}
}The canvas renders through CanvasKit, the same Skia engine behind professional tools — layers, vectors, and text stay crisp at any zoom.