Excalidraw Architect MCP Server
Excalidraw diagrams generated from natural language with automated graph layout algorithms.
Data last scanned today · Reviewed today
Overview
Excalidraw Architect MCP enables developers to generate structured system diagrams by abstracting coordinate math away from the LLM. It uses a graph layout algorithm to ensure that components like Kafka, PostgreSQL, and Redis are positioned and styled automatically, preventing common issues like overlapping boxes and tangled arrows. Users can maintain their system architecture as a version-controlled knowledge graph that the AI can query, edit, and re-render. You can iteratively modify existing diagrams with natural language instructions, such as adding components or changing connections, while exporting the final output to SVG or PNG formats without requiring a browser.
Our verdict
This tool carries a high risk level with a security score of 40/100. It is a community-maintained project that has not undergone editorial review, and critical components such as authentication and OAuth support remain unvetted. Due to its status as a third-party, unreviewed project, users should exercise caution before integrating it into sensitive development environments.
- Uses Sugiyama algorithm for automated, non-overlapping layouts
- Supports auto-styling for over 50 common technology stacks
- Operates fully offline without requiring external API keys
- Provides zero direct dependencies, minimizing supply-chain surface area
- Lacks official vendor maintenance and support
- Requires editorial review for security and authentication methods
- Reads as a community project without a linked homepage
Tools
| Tool | Description | Risk |
|---|---|---|
| create_diagram | Create a new diagram from structured node/connection data | medium |
| mermaid_to_excalidraw | Convert Mermaid flowchart syntax to .excalidraw | medium |
| modify_diagram | Add/remove/update nodes and connections on an existing diagram | medium |
| get_diagram_info | Read current diagram state (call before modifying) | low |
| export_diagram | Export .excalidraw to SVG or PNG image | low |
| kg_init | Create a new knowledge graph file | medium |
| kg_add_service | Add/update or remove a service (with type, domain, owner, tags, links) | medium |
| kg_remove_service | Add/update or remove a service (with type, domain, owner, tags, links) | medium |
| kg_link | Add/remove a dependency (parallel edges supported — e.g. REST and Kafka between the same pair) | medium |
| kg_unlink | Add/remove a dependency (parallel edges supported — e.g. REST and Kafka between the same pair) | medium |
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf | ||||
| ChatGPT |
Frequently asked questions
›Do I need an API key to use this MCP server?
No, the server runs fully offline in environments like Cursor or Windsurf and requires no API keys.
›How are diagrams generated without manual positioning?
The server uses a Sugiyama layout algorithm to handle pixel math and positioning, which prevents overlapping boxes and tangled arrows.
›What is the primary file format used for the architecture knowledge graph?
The knowledge graph is stored as a single markdown file, typically located at .claude/architecture.md, which is both human- and machine-readable.
›Can I export my diagrams to common image formats?
Yes, the server supports exporting any .excalidraw file to SVG or PNG format without requiring a browser.
›How does the knowledge graph ensure consistency across different diagrams?
By using a persistent model, each service maintains the same ID, styling, and metadata regardless of which view or diagram is being rendered.
Badge
Maintain this server? Add the live badge to your README.