pictogrammers-mcp
Enables AI assistants to search, download, and use Pictogrammers icons (Material Design Icons, MDI Light, Memory) with visual coherence by pinning a project to a single library and variant.
README
pictogrammers-mcp
An MCP server that lets an AI assistant search, download and use Pictogrammers icons (Material Design Icons, MDI Light, Memory) instead of hand-writing SVG — while keeping every icon in a project visually coherent by pinning it to a single library and variant.
Why
Hand-drawn SVG icons are inconsistent and often wrong. This server gives the model a real icon set of 8,000+ professionally designed glyphs, ranked search, and multiple output formats (raw SVG, JSX, @mdi/react, @mdi/js constants, data URIs). A pictogrammers.config.json at the project root pins one library so the model can't silently mix styles — if it tries, it gets a coherence warning.
Libraries
| id | name | grid | count | style |
|---|---|---|---|---|
mdi |
Material Design Icons | 24×24 | ~7,400 | Bold filled; most icons have an -outline twin |
mdil |
MDI Light | 24×24 | ~260 | Thin single-weight line icons |
memory |
Memory Icons | 22×22 | ~650 | Pixel-precise, for low-res/embedded displays |
Data comes from the official pictogrammers.com/data/*.json files, cached to disk on first use — after that the server works offline.
Install & build
npm install
npm run build
Register with Claude Code
Add to your .mcp.json (project) or user MCP config. See mcp.example.json:
{
"mcpServers": {
"pictogrammers": {
"command": "node",
"args": ["F:/_projects/dru/pictogrammers-mcp/dist/index.js"],
"env": {
"PICTOGRAMMERS_PROJECT_DIR": "${workspaceFolder}"
}
}
}
}
Or from the CLI:
claude mcp add pictogrammers -- node F:/_projects/dru/pictogrammers-mcp/dist/index.js
Tools
| tool | purpose |
|---|---|
init_project_icons |
Pin the project to one library + variant (writes pictogrammers.config.json). Call this first. |
project_icon_status |
Show the pinned library, defaults, and inventory of icons used. |
list_libraries |
List libraries with counts, grid sizes and npm packages. |
list_tags |
Browse the category tags of a library. |
search_icons |
Ranked keyword search; defaults to the pinned library/variant. |
get_icon |
Fetch one icon as svg / svg-path / jsx / react-mdi / js-constant / data-uri. |
download_icons |
Write .svg files (or a JSON map) to a directory. |
Typical flow
init_project_icons { "library": "mdi", "variant": "outline" }search_icons { "query": "shopping cart" }→ pickcart-outlineget_icon { "name": "cart-outline", "format": "jsx" }→ paste into a component- …or
download_icons { "names": ["cart-outline","account-outline"], "outputDir": "src/assets/icons" }
Because the project is pinned, every later search_icons/get_icon stays within the same family, and requesting another library returns a visible coherence warning.
Configuration (env)
| var | default | meaning |
|---|---|---|
PICTOGRAMMERS_PROJECT_DIR |
server cwd | where pictogrammers.config.json lives |
PICTOGRAMMERS_CACHE_DIR |
~/.pictogrammers-mcp/cache |
downloaded data cache |
PICTOGRAMMERS_BASE_URL |
https://pictogrammers.com/data |
data source |
License
MIT. Icon data is © the Pictogrammers project under their respective licenses (MDI/MDIL: Apache-2.0; see pictogrammers.com/docs).
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。