PluginBench
MCP Server
Active
MIT

Iconoteka MCP Server

io.github.turbaba/iconoteka

Search and retrieve 1,300 open-source icons in 7 weights and 2 styles as paste-ready SVG.

What is the Iconoteka MCP server?

The Iconoteka MCP server provides access to a library of 1,300 precisely designed, open-source pictograms across 23 categories. It enables searching icons by name and retrieving them in any of 7 weights (thin to bold) and 2 styles (stroke or fill) as ready-to-use SVG paths. Icons are designed on a 24-point grid for crisp rendering and optical balance across all variants.

Iconoteka is a universal icon library with 1,300 icons available in 7 weights and 2 styles. Use this MCP server to search for icons by keyword and retrieve SVG code that you can paste directly into your projects. It's useful for designers and developers who need quick access to a consistent, professionally designed icon set without managing multiple files.

How to install Iconoteka

Copy-paste configuration for popular MCP clients.

transport: stdio
Config generated by PluginBench — verify against the source before use.
~/Library/Application Support/Claude/claude_desktop_config.json
{
  "mcpServers": {
    "iconoteka": {
      "command": "npx",
      "args": [
        "-y",
        "iconoteka-mcp"
      ]
    }
  }
}

Tools & capabilities

Tools this server exposes to the agent.

  • search_icons — Search the Iconoteka library by name, keyword, or category to find matching icons.
  • get_icon_svg — Retrieve the SVG path data for a specific icon in a chosen weight (thin, ultralight, light, regular, medium, semibold, bold) and style (stroke or fill).

Use cases

  • Find and copy SVG code for icons to use in web projects or design files
  • Search icons by category (e.g., arrows, interface, nature, transportation) to discover relevant pictograms
  • Retrieve icons in different weights to match your design system's typography hierarchy
  • Get both stroke and fill variants of the same icon for flexible styling options
  • Build icon components by fetching SVG paths programmatically for design systems or component libraries

Iconoteka MCP server FAQ

What is the Iconoteka MCP server?

It's an MCP server that gives you access to search and retrieve SVGs from the Iconoteka icon library—1,300 open-source icons in 7 weights and 2 styles.

Is Iconoteka free to use?

Yes, Iconoteka is free and open-source under the MIT license. It's free for personal and commercial use, including in closed-source products, with no permission or fee required.

How do I install the Iconoteka MCP server?

Install via npm with `npm install iconoteka-mcp`, then configure it in your MCP client (Cursor, Claude, etc.) to enable the search and icon retrieval tools.

What icon formats and weights are available?

Icons are available as SVG paths in 7 weights (thin, ultralight, light, regular, medium, semibold, bold) and 2 styles (stroke and fill), all designed on a 24-point grid.

Can I use these icons in commercial projects?

Yes, Iconoteka is free for both personal and commercial use, including closed-source products. No credit is required, though it's appreciated.

How are the icons organized?

Icons are organized into 23 categories (arrows, interface, hardware, nature, transportation, etc.) and can be searched by name or keyword.

README (reference)

Source of truth, from the repository.

Iconoteka

A universal, open-source library of precisely designed pictograms.

Website · Figma Plugin · License

1,300 iconsfrom rapid prototyping to real-world wayfinding systems
7 weightsThin, Ultralight, Light, Regular, Medium, Semibold, Bold
2 stylesstroke and fill
14,970 shapescomplete coverage for almost any use case
23 categoriesArrows, Interface, Hardware, Nature, Transportation, …

Adding icons to your project

Take the icons you need — there's no need to install or clone the whole library.

In Figma — install the plugin, search, drag the icon onto the canvas.

One icon at a time — every icon is a standalone SVG on the CDN:

https://cdn.jsdelivr.net/gh/iconoteka/iconoteka@02.00.15/Icons/Interface/bell-notification-notify-reminder-ring-sound-r-s.svg

The path is Icons/{Category}/{name}-{weight}-{style}.svg, where weight is one of t u l r m s b (thin → bold) and style is s for stroke or f for fill. Categories with spaces need URL-encoding: Files%20%26%20Data.

Inline in your markup — both styles are outlined paths, so an icon is a single <path> with no stroke attributes. Every SVG ships with fill="currentColor", so it inherits the surrounding text colour:

<svg width="24" height="24" viewBox="0 0 24 24" aria-hidden="true">
  <path d="M3 15.55C4.675 13.225 5 11.45 5 8.5…" fill="currentColor"/>
</svg>

Swap the weight or style by changing one letter in the filename — the geometry is designed to stay optically balanced across all seven weights.

Framework packages

Icons as components, generated from the same SVGs:

npm install iconoteka-react    # or iconoteka-vue, iconoteka-svelte
import { Bell } from "iconoteka-react";

<Bell />
<Bell weight="bold" variant="fill" size={32} />

weight takes any of the seven, variant is stroke or fill, and anything else is spread onto the <svg>. Each icon is its own module, so bundlers drop what you don't import.

Using the full set

icons.json carries every icon and variant in one file — useful for bundling the set into a design system, generating components, or pulling icons at build time.

npm install iconoteka
import icons from "iconoteka" with { type: "json" };  // ESM
const icons = require("iconoteka");                   // CommonJS

Or fetch it from the CDN:

https://cdn.jsdelivr.net/gh/iconoteka/iconoteka@02.00.15/icons.json

Pin the tag rather than @main — jsDelivr caches branch URLs for up to 7 days. The npm package contains icons.json only; the raw SVGs stay in the repo.

{
  "meta": {
    "version": "02.00.09",
    "count": 1300,
    "categories": 23,
    "popular": ["ai", "alert", "bank_card", "bell", "…"]
  },
  "icons": [
    {
      "name": "leaf-ecology-nature-natural-leaves-eco-green-organic-plant-wind",
      "displayName": "Leaf",
      "searchTerms": ["leaf", "ecology", "nature", "natural", "leaves"],
      "category": "Nature",
      "variants": { "regular": { "stroke": "M12 2.5…", "fill": "M12 2.5…" } }
    }
  ]
}

The first segment of name is the icon's identity; the rest are search-only aliases, so match on the identity rather than the full string. Weights are keyed thin, ultralight, light, regular, medium, semibold, bold, each holding a stroke and/or fill path.

A curated selection carries "popular": true, listed by identity in meta.popular. Popular is a flag, not a category — no icon is stored twice.

Design

24-point grid — every icon is built on a 24px grid, so straight lines stay crisp at 24, 48, 96px and beyond, and the set sits comfortably alongside other popular libraries.

Advanced optics — deliberate overshoots, density-based weight modularity and optical stroke compensation throughout.

Bracketing system — preserves structural integrity across all seven weights. Unlike stroke-based sets that simply scale outlines, legibility and visual balance hold at both extremes.

System architecture — a structured metaphor hierarchy, a 5° geometric angle rule, tiered corner rounding and modular components keep the library consistent at production scale.

Universality — geometric construction and high visual clarity keep the icons legible across products and industries that demand contrast and structural consistency.

Support

Iconoteka is a free, open-source library created and maintained by one person. Support its future development through a monthly Patreon subscription or a one-time Ko-fi donation.

License

MIT — see LICENSE. Free for personal and commercial use, including in closed-source products. No permission needed, no fee. Credit isn't required, though it's always appreciated.

Author

Built by Oleg Turbaba, Brand Design Director at Clay Global — a multidisciplinary designer working at the intersection of brand identity, digital products and visual experiences.

For enquiries, collaborations, interviews or sponsorships: olegturbaba@gmail.com

Instagram · LinkedIn · Behance

Sharing your work? Mention @iconoteka on Instagram, LinkedIn or X.

Related MCP servers

Website performance monitoring: scans, Core Web Vitals, RUM data and alerts.

0
JavaScript
MIT
View repository →

Build-aware code intelligence for embedded C/C++ firmware — libclang, FTS5, call graphs, 33 tools

7
Python
MIT
View repository →

Temporal memory MCP server with LanceDB vector search, weight-decay scoring, and LLM consolidation

View repository →

Patent search and prior art discovery via Google Patents on BigQuery

5
Go
MIT
View repository →

Vietnam public-tender search (VNEPS) for AI agents, pay per use.

0
View repository →

Live web search for AI agents. $0.001/call, x402 on Base, no API key.

0
TypeScript
MIT
View repository →