ExcaliClaw: A Collaborative Tool That Lets AI Agents Draw Diagrams With You

When AI Agents Learn to 'Hand-Draw': What ExcaliClaw Is Trying to Solve

Among tools for creating flowcharts, architecture diagrams, and mind maps, Excalidraw has won a loyal following with its hand-drawn style and lightweight collaboration. But its shortcomings are obvious too: when you want AI to help you draw a system architecture diagram, you often need to paste your description into another chat window, then manually move the generated result back.

ExcaliClaw is trying to fill exactly that gap. According to its product listing on Product Hunt, it positions itself as a "collaborative workspace for hand-drawn diagrams," with its core selling point being a built-in MCP Server—allowing AI coding tools like Claude, Cursor, or VS Code to directly create and edit diagrams, rather than working through a layer of copy-paste.

This positioning is quite pragmatic. It doesn't try to reinvent the drawing tool, but instead pulls AI agents into the act of drawing itself, making "you draw a stroke, AI adds another" possible.

Breaking Down Core Capabilities: Collaboration, Search, and AI Agents

Real-Time Collaboration and Collection Management

ExcaliClaw retains the collaborative DNA of hand-drawn diagrams. You can organize diagrams into different collections, instantly search and share them by link or name, and support multiple people drawing on the same screen in real time. Version history, trash, and backup export are all in the feature list—these seemingly unremarkable capabilities are precisely where teams most easily stumble when using drawing tools long-term.

The Relationship Between 'Collections' and Canvases

A Collection can be understood as an organizational unit one level above a single canvas. Traditional drawing tools usually manage work by files or folders, whereas a collection is closer to a shareable, searchable, collaborative project space: multiple related diagrams are grouped into the same collection, and team members can locate them by link or name without digging through layer after layer of directories. This design is especially friendly for artifacts like architecture diagrams and flowcharts that often appear in groups—a system's deployment diagram, sequence diagram, and data flow diagram naturally belong to the same collection, sharing a set of access permissions and version history, and collaboration won't lose context because they're scattered across different folders.

MCP Server: The Real Differentiator

The most noteworthy part of the product is its MCP Server. MCP (Model Context Protocol) is becoming a common way for AI tools to interact with external applications, and ExcaliClaw applies it to diagram creation.

According to the product description, Claude, Cursor, and VS Code can all use this server to generate or modify diagrams for you. This means that while writing code, you can directly have AI draw out the module relationships in your head, then manually adjust them on the same canvas. For developers who need to frequently produce architecture diagrams, this "chat while drawing" experience is much smoother than switching between multiple tools.

What Problem Does MCP Actually Solve?

MCP (Model Context Protocol) was proposed and open-sourced by Anthropic at the end of 2024. It is essentially a standardized protocol that lets large language models converse with external tools and data sources. Before MCP appeared, every time an external application was integrated, developers had to write a separate tool-calling adaptation layer for each model, and the workload grew multiplicatively. MCP abstracts this adaptation layer into a unified server-side interface: an external application only needs to implement an MCP Server once, and any client that supports the protocol (Claude Desktop, Cursor, VS Code plugins, etc.) can call it.

Applied to the drawing scenario, this means ExcaliClaw doesn't need to develop a separate plugin for each AI tool. As long as it maintains a single MCP Server, different clients can "read the canvas and modify shapes" in the same way. For users, the most direct experience is: AI is no longer an external consultant behind a chat window, but a collaborator that can read and write the same canvas data in real time.

Free Strategy and Pricing

ExcaliClaw's free plan is fairly generous: 200 drawing credits, plus unlimited AI (MCP) usage, and no credit card required. Paid plans start at €9 per month. The interface supports English, Spanish, and French.

This pricing structure reveals a judgment—the number of drawings is a cost-sensitive item, while AI calls are treated as an acquisition hook. For individual developers and small teams, using the free allowance to first get the AI drawing workflow running before deciding whether to pay is a low barrier to entry.

Where It Stands: The AI-ification Race in Hand-Drawn Tools

ExcaliClaw is not an isolated case. Around the Excalidraw ecosystem, many enhanced tools have already appeared, but most remain at the level of template libraries or style enhancements. ExcaliClaw chooses to enter from "AI agent collaboration," a direction closer to the actual rhythm of today's AI-assisted programming.

It is also fundamentally different from tools like Mermaid that generate diagrams from text. Mermaid requires you to describe diagrams in code, whereas ExcaliClaw retains the freedom of a hand-drawn canvas, with AI merely sitting beside you as a collaborator. For people who are not used to writing diagram code but still want to leverage AI for efficiency, this middle path has its appeal.

Currently, ExcaliClaw has received 73 votes on Product Hunt, ranking 14th, putting it in a state of "being discussed but not quite a breakout hit." This may indicate that it hits a relatively vertical developer need rather than the mass market.

The Essential Divide Between Mermaid and Excalidraw

Mermaid adopts the "text as diagram" paradigm, using a Markdown-like DSL to describe nodes and connections, rendering standardized graphics. Its advantages are version controllability, embeddability in documents, and clear structure, at the cost of losing the flexibility of free layout, with complex layouts often constrained by renderer rules. Excalidraw takes another path: a free hand-drawn canvas based on Canvas, where element positions are entirely determined by people, with a loose style and strong readability, suitable for quick communication and sketch expression.

ExcaliClaw's middle path is grafted precisely onto the latter: retaining Excalidraw's freedom and introducing AI capabilities as a collaborator rather than a compiler. This determines that its use cases lean more toward "exploratory diagrams that require back-and-forth human-machine modification" rather than "one-shot generated diagrams with fixed structure."

A Few Points Worth Watching

Where will the idea of pulling AI into the drawing canvas go next? There are several directions worth observing.

First, whether AI-generated diagrams can understand existing hand-drawn content. If AI can only create from scratch and cannot continue completing based on a draft you're halfway through, the sense of collaboration will be compromised.

Second, the depth of MCP support. Claude, Cursor, and VS Code each have quite different workflows. Whether ExcaliClaw can maintain a consistent experience across all three environments is key to whether it can retain developers.

Third, the sustainability of the free allowance. 200 drawings combined with unlimited AI usage is very friendly in the early customer acquisition stage, but in the long run, the cost pressure of AI calls will force the team to adjust its strategy.

For developers looking for AI-assisted drawing solutions, ExcaliClaw offers an option worth trying. It doesn't promise to overturn anything; it simply puts hand-drawn diagrams and AI agents on the same canvas. This plain idea may be closer to the real needs of daily work than many flashy AI drawing concepts.

One-line Summary

ExcaliClaw connects a hand-drawn diagram canvas with AI coding tools through an MCP Server, enabling AI to collaborate on drawing directly.

Paragraph Summary

ExcaliClaw is a collaborative tool that connects AI agents to a hand-drawn diagram canvas. Its core selling point is a built-in MCP Server that lets AI coding tools like Claude, Cursor, and VS Code directly create and edit diagrams rather than relying on cross-window copy-paste. It retains Excalidraw-style hand-drawn DNA and real-time collaboration, offering collection management, instant search and sharing, version history, and backup export. The free plan provides 200 drawing credits and unlimited AI usage, with paid plans starting at €9 per month. In positioning, it sits between Excalidraw-style hand-drawn tools and Mermaid-style code-generated diagram tools, focusing on AI collaboration rather than code description. On Product Hunt, it received 73 votes and ranked 14th, making it a product that hits a vertical developer need.