Skip to content
Infragistics

CLI MCP for AI-Assisted Development

AI coding assistants can generate code quickly, but the results don’t always match your project, framework, or component library. Developers often spend time fixing imports, correcting components, and filling in missing setup steps. Ignite UI CLI MCP helps solve this problem by giving AI assistants the current Ignite UI docs, API reference and CLI setup steps, so they scaffold projects, add components and answer documentation questions with the Ignite UI CLI instead of guessing. Available for Ignite UI for Angular, React, Web Components and Blazor.

YouAI assistantCLI MCPTerminalCreate an Angular appget_project_setup_guide · angularig new … --template=baseig new MyApp --framework=angularget_doc · grid-pagingigx-paginator usage✓ App ready, paged grid
  1. You → AI assistantCreate an Angular app
  2. AI assistant → CLI MCPget_project_setup_guide · angular
  3. CLI MCP → AI assistantig new … --template=base
  4. AI assistant → Terminalig new MyApp --framework=angular
  5. AI assistant → CLI MCPget_doc · grid-paging
  6. CLI MCP → AI assistantigx-paginator usage
  7. AI assistant → YouApp ready, paged grid

The Problem

AI Without Context Creates Bad Output

Most AI coding assistants generate code from patterns, not from real awareness of your application, creating problems in front-end development. The code may look correct at first, but it often doesn’t match your project structure, dependencies, routing, or component library. Developers end up fixing imports, adjusting patterns, and rewriting UI that breaks when it runs.

The real issue is context. Without access to the actual development workflow, AI fills in missing details with guesses.

Pick a prompt to see which CLI MCP tools the assistant calls and what it gets back.

Create a new Angular app with a data grid

get_project_setup_guide(framework: "angular")CLI steps, templates: base · side-nav · side-nav-auth

The assistant runs the Ignite UI CLI in your terminal, with the base template for a single view

The MCP supplies the steps; the CLI creates the project

How do I pin columns in the React grid?

search_docs(query: "column pinning", framework: "react")grid-column-pinning

get_doc(framework: "react", name: "grid-column-pinning")pinning position, toolbar, API, styling

get_api_reference(platform: "react", component: "IgrColumn", member: "pinned")pinned: boolean

Answered from the React docs, with React APIs only

Which options control filtering in the Angular combo?

search_api(query: "combo filtering", platform: "angular")IComboFilteringOptions, IgxSimpleComboComponent, …

get_api_reference(platform: "angular", component: "IComboFilteringOptions")caseSensitive, filteringKey

Real member names and types for the installed version

Set up a new Blazor app with Ignite UI

get_project_setup_guide(framework: "blazor")dotnet new and NuGet steps

Blazor gets .NET steps, not npm ones

The Solution

Turning AI Guesswork into Guided Workflows

Ignite UI CLI MCP gives your AI assistant the current Ignite UI docs, API reference and framework-specific setup steps, so it scaffolds apps and adds components with the Ignite UI CLI — the same commands your team runs — instead of from memory.

It exposes six read-only tools: list_components, get_doc, search_docs, get_project_setup_guide, search_api and get_api_reference. They look things up; your assistant still writes the code and runs the commands, in the open.

What You Can Do With Ignite UI CLI MCP

From a first project to the hundredth component, the assistant works from the Ignite UI sources rather than a guess.

Start a project

Ask for a new app and the assistant gets the framework’s setup guide: which CLI command to run, which template to pick, and what to do next.

  • base, side-nav and side-nav-auth templates, chosen by what the app needs
  • Angular Schematics and ng add for existing Angular projects
  • dotnet new and NuGet steps for Blazor
get_project_setup_guide(framework: "angular") → ig new MyApp --framework=angular --type=igx-ts --template=base

Answer from the docs

Instead of reading a whole documentation site, the assistant searches it and fetches the one topic it needs, with its code samples.

  • Full-text search with prefix matching
  • Whole topics with their code samples
  • A table of contents per framework
search_docs(query: "column pinning", framework: "react") → get_doc(framework: "react", name: "grid-column-pinning")

Look up exact APIs

When a question comes down to one property or event, the assistant looks it up rather than guessing the name.

  • A component’s properties, methods and events, or a single member
  • Search by keyword when the component name is unknown
  • One framework, or all four at once
search_api(query: "combo filtering", platform: "angular") → get_api_reference(platform: "angular", component: "IComboFilteringOptions")

Stay in one framework

Every call names its framework, and the server tells the assistant never to mix APIs between them — Igx, Igr, Igc and Igb stay apart.

  • Framework detected from component prefixes, file types and packages
  • Pairs with the Theming MCP for palettes and component themes
About the Theming MCP
.vscode/mcp.json
{
  "servers": {
    "igniteui-cli": {
      "command": "npx",
      "args": ["-y", "igniteui-cli", "mcp"]
    },
    "igniteui-theming": {
      "command": "npx",
      "args": ["-y", "igniteui-theming", "igniteui-theming-mcp"]
    }
  }
}

Setup

Get Started in a Minute

The CLI MCP runs from the igniteui-cli package with npx, so there is nothing to install first. It needs Node.js 18 or later; the first run downloads the package.

In VS Code, create .vscode/mcp.json as shown. Cursor, Claude Desktop, Claude Code and JetBrains AI Assistant take the same servers under mcpServers. Or run ig ai-config in an existing project: it writes this for your assistants and adds the Agent Skills too.

Read the Setup Guide

Why Ignite UI

Ignite UI combines a complete component system with the structured context AI agents need to generate accurate UI — in three layers, each independently configurable, and all three set up by ig ai-config.

AI Agent Skills

Pre-built skills teach an assistant how Ignite UI components are actually used — the props, the patterns and the pitfalls — so the code it writes matches the library rather than a generic approximation of it.

  • Ship inside igniteui-angular, igniteui-react and igniteui-webcomponents; Blazor skills install from GitHub
  • Component, grid and theming patterns, plus design-to-code from an image
About AI Agent Skills

CLI MCP Server

You are here: the layer that gives an assistant the current docs, API reference and setup steps, so it builds with the Ignite UI CLI instead of from memory.

  • Six read-only tools for docs, API reference and setup guides
  • Angular, React, Web Components and Blazor
npx -y igniteui-cli mcp

Theming MCP Server

The Theming MCP generates palettes, themes and component styles through the Ignite UI Theming Engine, so an assistant themes with real tokens instead of from memory.

  • Ships inside the igniteui-theming package
  • Material, Bootstrap, Fluent and Indigo, light or dark
About the Theming MCP Server

Frequently Asked Questions

What is Ignite UI CLI MCP?

Ignite UI CLI MCP is a Model Context Protocol (MCP) server that gives AI coding assistants the official Ignite UI docs, API reference and CLI setup guides, so they create and extend applications with the Ignite UI CLI instead of guessing code patterns. It covers Ignite UI for Angular, React, Web Components and Blazor. The result is code that better matches your project from the start.

Which tools does Ignite UI CLI MCP expose?

Six read-only tools. Documentation: list_components, get_doc and search_docs. API reference: search_api and get_api_reference. Project setup: get_project_setup_guide, which returns the Ignite UI CLI (or, for Blazor, dotnet new and NuGet) steps for a framework.

Does Ignite UI CLI MCP change my files or run commands?

No. Its tools only look things up. Your AI assistant writes the code and runs commands such as ig new in your terminal, the same way it runs any other command — so you can review every step.

How do I set up Ignite UI CLI MCP?

Register npx -y igniteui-cli mcp as an MCP server in your AI client — in VS Code, under servers in .vscode/mcp.json; in Cursor, Claude Desktop, Claude Code and JetBrains AI Assistant, under mcpServers. Or run ig ai-config in your project to write the configuration for you, together with the Theming MCP and the Agent Skills.

Is Ignite UI CLI MCP free?

Yes. The MCP server ships inside the igniteui-cli package, which is free and MIT-licensed. It works with the open-source Ignite UI components and with the licensed suite; the licensed components need their own license.

Which AI assistants work with Ignite UI CLI MCP?

Ignite UI CLI MCP works with AI clients that support the Model Context Protocol over STDIO transport — including VS Code with GitHub Copilot, GitHub Copilot cloud agent, Cursor, Claude Desktop, Claude Code, and JetBrains AI Assistant. Any other MCP-compatible client that supports STDIO is also supported.

Do I need to change my existing development workflow to use Ignite UI CLI MCP?

No major workflow changes are required. Ignite UI CLI MCP builds on existing CLI-based workflows, allowing teams to continue using familiar tools while adding AI assistance in a controlled way. This makes it easier to adopt without disrupting established development practices.

Free for every developer

AI Toolchain, Free to Use

Extend the AI tools you and your team already use. The toolchain is free and ships with the library, because AI agents are already part of your team.

  • MIT-licensed on npm
  • No usage-based AI tier or separate billing
  • Works with open-source and licensed Ignite UI

Select a framework to get started for free: