monosketch-mcp
An MCP server that enables AI agents to create beautiful ASCII diagrams using Unicode box-drawing characters, with support for stateful canvas operations, multiple shape types, and style customization.
README
monosketch-mcp
An MCP server that gives AI agents the ability to create beautiful ASCII diagrams using Unicode box-drawing characters. Built on the rendering engine from MonoSketch.
┌──────────────────┐ ┌──────────────────┐
│ │ │ │
│ Frontend │──────▶│ API │
│ │ │ │
└──────────────────┘ └────────┬─────────┘
│
│
┌────────▼─────────┐
│ │
│ Database │
│ │
└──────────────────┘
Setup
Requires Bun runtime.
git clone https://github.com/supersterling/monosketch-mcp.git
cd monosketch-mcp
bun install
Add to Claude Code
claude mcp add monosketch -- bun run /path/to/monosketch-mcp/index.ts
Add to Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"monosketch": {
"command": "bun",
"args": ["run", "/path/to/monosketch-mcp/index.ts"]
}
}
}
Tools
render_scene
Stateless one-shot rendering. Pass an array of shapes, get back ASCII art.
{
"shapes": [
{
"type": "rectangle",
"x": 0, "y": 0, "width": 20, "height": 5,
"border": { "enabled": true }
},
{
"type": "text",
"x": 0, "y": 0, "width": 20, "height": 5,
"content": "Hello World",
"horizontalAlign": "center",
"verticalAlign": "middle",
"border": { "enabled": true }
}
]
}
Output:
┌──────────────────┐
│ │
│ Hello World │
│ │
└──────────────────┘
Canvas Tools (Stateful)
Build diagrams incrementally across multiple tool calls:
| Tool | Description |
|---|---|
create_canvas |
Create a named canvas session |
add_shape |
Add a shape to a canvas |
modify_shape |
Update shape properties |
remove_shape |
Remove a shape |
render_canvas |
Render current state to ASCII |
list_canvases |
List active sessions |
clear_canvas |
Remove all shapes |
Utilities
| Tool | Description |
|---|---|
list_styles |
List all available border, fill, and anchor styles |
measure_text |
Compute bounding box for word-wrapped text |
Shape Types
Rectangle
{ "type": "rectangle", "x": 0, "y": 0, "width": 10, "height": 4,
"border": { "enabled": true, "style": "S1", "corner": "rounded" },
"fill": { "enabled": true, "style": "F1" } }
Text
{ "type": "text", "x": 0, "y": 0, "width": 20, "height": 3,
"content": "Centered text", "horizontalAlign": "center", "verticalAlign": "middle",
"border": { "enabled": true } }
Line
{ "type": "line",
"points": [{"x": 0, "y": 2}, {"x": 10, "y": 2}, {"x": 10, "y": 5}],
"stroke": { "enabled": true },
"endAnchor": "A1" }
Group
{ "type": "group", "children": [
{ "type": "rectangle", "x": 0, "y": 0, "width": 10, "height": 3, "border": { "enabled": true } },
{ "type": "rectangle", "x": 15, "y": 0, "width": 10, "height": 3, "border": { "enabled": true } }
]}
Style Reference
Border / Stroke Styles
| ID | Name | Characters |
|---|---|---|
| S1 | Thin (default) | ─ │ ┌ ┐ └ ┘ |
| S2 | Bold | ━ ┃ ┏ ┓ ┗ ┛ |
| S3 | Double | ═ ║ ╔ ╗ ╚ ╝ |
| S4 | Rounded | ─ │ ╭ ╮ ╰ ╯ |
Fill Styles
| ID | Name | Character |
|---|---|---|
| F1 | Space | (opaque blank) |
| F2 | Solid | █ |
| F3 | Medium shade | ▒ |
| F4 | Light shade | ░ |
| F5 | Diagonal | ▚ |
Anchor Styles (Line Endpoints)
| ID | Name | L R U D |
|---|---|---|
| A1 | Filled arrow | ◀ ▶ ▲ ▼ |
| A12 | Hollow arrow | ◁ ▷ △ ▽ |
| A2 | Filled square | ■ ■ ■ ■ |
| A21 | Hollow square | □ □ □ □ |
| A220 | Filled diamond | ◆ ◆ ◆ ◆ |
| A221 | Hollow diamond | ◇ ◇ ◇ ◇ |
| A3 | Hollow circle | ○ ○ ○ ○ |
| A5 | Filled circle | ● ● ● ● |
Dash Patterns
Add "dash": { "dash": 3, "gap": 1 } to any shape for dashed borders/strokes.
Line Crossing
When lines cross, junction characters are automatically resolved:
│
────┼────
│
The engine handles all combinations of single, bold, and double line styles with correct Unicode junction characters (T-junctions, crosses, corners).
Tests
bun test
208 tests covering geometry primitives, bitmap rendering, board compositing, crossing resolution, shape hierarchy, renderer pipeline, canvas management, and end-to-end integration scenarios.
Architecture
The rendering engine was extracted from MonoSketch, an open-source ASCII diagram editor.
Shape Description (JSON)
→ BitmapFactory (per shape type)
→ MonoBitmap (sparse character matrix)
→ MonoBoard (tiled canvas with crossing resolution)
→ ASCII string output
Engine modules:
engine/geo/— Point, Rect, Size geometry primitivesengine/shape/— Shape hierarchy (Rectangle, Text, Line, Group)engine/style/— Stroke, fill, anchor, and alignment stylesengine/bitmap/— Sparse bitmap matrix and per-shape rendering factoriesengine/board/— Tiled canvas compositing with Unicode line crossing resolution
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。