excalidraw▌

Two modes are available. Try MCP first — it has more capabilities.

Generate Excalidraw diagrams from natural language descriptions in multiple formats. \n \n Supports nine diagram types: flowcharts, relationship diagrams, mind maps, architecture diagrams, data flow diagrams, swimlane business flows, class diagrams, sequence diagrams, and ER diagrams \n Outputs valid .excalidraw JSON files that open directly in Excalidraw or the VS Code extension \n Includes layout guidelines, element count recommendations, and color schemes for consistent visual design \n Optio

Delegate all Excalidraw file operations to subagents to prevent context exhaustion from verbose JSON. \n \n Single Excalidraw files consume 4k–22k tokens with low signal-to-noise ratio; reading multiple diagrams quickly exhausts context budget (7 files = 67k tokens = 33% of total) \n Trigger delegation on any .excalidraw or .excalidraw.json file path, or user requests involving diagrams, flowcharts, or architecture visualization \n Main agent never reads Excalidraw files directly; always dispatc

$23

Core principle: Main agents NEVER read Excalidraw files directly. Always delegate to subagents to isolate context consumption.

Generate Excalidraw diagram JSON files that argue visually, not just display information. \n \n Supports simple conceptual diagrams (abstract shapes, mental models) and comprehensive technical diagrams (concrete examples, code snippets, real data formats) \n Includes a research mandate for technical diagrams: look up actual specs, event names, APIs, and data formats before designing \n Provides a visual pattern library (fan-out, convergence, timelines, trees, cycles) where shape structure mirror