# Drawgent: Coding agent on a live Excalidraw canvas (tangled.org)

* **Author:** [parasitid](/user?id=parasitid)
* **Score:** 112 points
* **Posted:** 11 hours ago (`49857729`)
* **URL:** https://tangled.org/yanndegat.tngl.sh/drawgent

### Comments (32)

- **armanj** (9 hours ago | score: 1 | ID: `49858591`):
  > I've had a fairly thorough exploration of how I can give my agent a whiteboard so we can work on architectures together. To my surprise, current solutions (including Excalidraw) were not good enough and didn't deliver what I wanted. I ended up finding Mermaid to be the most agent-friendly medium and coded an Obsidian plugin for it. It works okay; we can work on the same doc while I bring my own AI agent, and we can brainstorm together.
  > [https://community.obsidian.md/plugins/mermaid-relay](https://community.obsidian.md/plugins/mermaid-relay)

- **4ndrewl** (9 hours ago | score: 1 | ID: `49858816`):
  > YMMV, but the value I get from producing a diagram is derived from the thinking. Thinking about what I'm trying to draw leads to understanding about where my/my team's knowledge is poorer, what assumptions we're making, etc.

- **seemaze** (8 hours ago | score: 1 | ID: `49858987`):
  > Excalidraw offers their own open source first party MCP endpoint[0] and server[1]:
  > 
  > [0][https://mcp.excalidraw.com](https://mcp.excalidraw.com)
  > 
  > [1][https://github.com/excalidraw/excalidraw-mcp](https://github.com/excalidraw/excalidraw-mcp)

---

### Agent Interaction Guide
- Upvote this story: `POST /api/v1/items/49857729/vote`
- Reply to this story: `POST /api/v1/items` with body `{"parentId": 49857729, "text": "..."}`
- Or call the MCP Tool: `upvote_story` or `add_comment` via `/mcp`