Drawgent: Coding agent on a live Excalidraw canvas
Drawgent is a new tool that connects local CLI coding agents like Claude Code directly to a live Excalidraw canvas for visual architecture editing.
summary is enough It provides a clear overview of a novel AI-whiteboard integration tool, but the release page details everything without requiring a deep dive.
summary
Drawgent is a single binary tool bridging local coding agents like Claude Code or Codex with a live Excalidraw whiteboard using MCP. It allows users to prompt architecture updates, use laser pointer gestures for specific tasks, and link diagram boxes directly to code repositories. The tool monitors layouts using a headless Chrome renderer and ensures that code changes and visual diagrams stay synchronized.
key points
- Connects local CLI coding agents directly to Excalidraw whiteboards using the Model Context Protocol.
- Supports visual gestures like laser pointing and AGENT canvas notes to trigger code refactoring or impact analysis.
- Maintains bidirectional code links from diagram elements to specific lines and files, allowing CI checks for broken links.
who should read it
Software engineers and developers experimenting with local AI coding workflows and visual architecture tools.
discussion positive
consensus
Commenters are intrigued by the visual-to-code workflow, comparing it to alternative tools like Mermaid, TLDraw, and custom MCP setups.
pushback
Some push back that forcing AI into diagramming defeats the cognitive benefits of manual sketching and risks producing visually cluttered output.
I found a big part of the appeal with tools such as Excalidraw is that people are still more willing to engage with your content because they think that its made by hand.
Compare a piss-tinted LLM-generated diagram with 80 pipes, flows and generic names everywhere to a seemingly hand-drawn sketch.
I have had my Exaclidraw connected to an agent through their MCP and I have gotten way more attention/time/reads on both blogs and papers because of the illusion of being human-made/hand-drawne making it more palatable.
It notes that Excalidraw's hand-drawn aesthetic bypasses anti-AI bias because people are more willing to engage with content that looks human-made.
Generated by an LLM from the article and thread. Check the sources before relying on it.