Skip to content

ai config

Jump to bottom
Marina Stoyanova edited this page May 13, 2026 · 3 revisions

AI Configuration for Ignite UI Projects

The ai-config command sets up AI coding assistant integration for your Ignite UI project. In a single command, it:

  • Configures MCP servers — writes the MCP config file for your chosen coding assistant with the Ignite UI MCP server entries
  • Copies skill files — adds framework-specific skill guides into your agent directories
  • Sets up instruction files — populates each agent's instruction file with project-specific guidance Most common usage: 
ig ai-config --assistants generic --agents claude

Flags Reference

Flag Values Default
--assistants generic, vscode, cursor, gemini, junie, none Prompted interactively; generic in non-interactive mode
--agents generic, claude, copilot, cursor, codex, windsurf, gemini, junie, none Prompted interactively; generic + claude in non-interactive mode

Supported Coding Assistants

Coding Assistant Choice value Config Path Root Key
Generic (Claude Code, VS Code, and others) generic .mcp.json mcpServers
VS Code (GitHub Copilot) vscode .vscode/mcp.json servers
Cursor cursor .cursor/mcp.json mcpServers
Gemini gemini .gemini/settings.json mcpServers
JetBrains Junie junie .junie/mcp/mcp.json mcpServers

Supported AI Agents

Agent Skills Directory Instruction File
Generic .agents/skills AGENTS.md
Claude .claude/skills .claude/CLAUDE.md
Copilot .github/skills .github/copilot-instructions.md
Cursor .cursor/skills .cursor/rules/cursor.mdc
Codex .codex/skills .codex/instructions.md
Windsurf .windsurf/skills .windsurf/rules/guidelines.md
Gemini .gemini/skills .gemini/GEMINI.md
Junie .junie/skills .junie/guidelines.md

Ways to Use AI Config

1. Standalone command (ig ai-config)

Run inside an existing Ignite UI project to add or update your AI configuration at any time:

# Interactive — prompts for coding assistants, then agents
ig ai-config
 
# Non-interactive — specify both assistants and agents
ig ai-config --assistants cursor --agents claude copilot
ig ai-config --assistants generic vscode --agents claude generic cursor
 
# Skip MCP configuration only
ig ai-config --assistants none --agents claude generic
 
# Skip skill files and instructions only (MCP servers are still configured)
ig ai-config --assistants vscode --agents none

2. During project creation (ig new)

AI configuration runs after the project structure is generated but before dependencies are installed.

# Interactive — prompts for AI assistants and agents after scaffolding
ig new my-app --framework=angular --type=igx-ts
 
# Non-interactive — specify assistants and agents upfront
ig new my-app --framework=angular --type=igx-ts --agents claude copilot --assistants generic
 
# Skip AI configuration entirely
ig new my-app --framework=angular --type=igx-ts --agents none --assistants none

Sequence during ig new <name>:

  1. Project files are generated
  2. AI configuration runs inside the new project directory
  3. Git is initialized (unless --skip-git)
  4. Dependencies are installed (unless --skip-install) Because AI configuration runs before packages are installed, skill files fall back to the CLI's built-in templates. Re-run ig ai-config after installing packages to pick up the latest skill files.

Interactive wizard (ig new with no arguments)

Running ig new without a project name launches a step-by-step wizard. After choosing your framework, project type, and theme, you'll be prompted for coding assistants and agents. The wizard mirrors the options in the tables above, with Generic and Claude pre-selected by default.

3. Angular Schematic (ng add / ng generate)

For Angular CLI projects, AI configuration is available as a schematic:

# Interactive — prompts for coding assistants and agents
ng generate @igniteui/angular-schematics:ai-config
 
# With specific assistants and agents
ng generate @igniteui/angular-schematics:ai-config --assistants cursor --agents claude copilot
 
# Runs automatically as part of ng add (defaults: agents ["claude", "generic"], assistants ["generic"])
ng add igniteui-angular

When run via the Angular schematic, an additional angular-cli MCP server entry is included automatically alongside the Ignite UI servers.

Skill Files

Skill files are framework-specific guides copied into each agent's skills directory. They are sourced from your installed Ignite UI package and kept in sync each time you run ig ai-config — existing files are only updated if their content has changed.

Tip: If you run ig ai-config before installing packages (e.g. with --skip-install), the CLI falls back to built-in templates. Re-run the command after installing to pick up the skill files from your installed version. If you see a warning that no skill files were found, check that your project has an ignite-ui-cli.json or a package.json with Ignite UI dependencies.

MCP Server Configuration

The command writes (or merges into) the config file for your chosen coding assistant. Existing third-party MCP server entries are always preserved — the command merges, never overwrites.

Generic, Cursor, Gemini, and Junie (.mcp.json and equivalents, root key mcpServers):

{
  "mcpServers": {
    "igniteui-cli": {
      "command": "npx",
      "args": ["-y", "igniteui-cli", "mcp"]
    },
    "igniteui-theming": {
      "command": "npx",
      "args": ["-y", "igniteui-theming", "igniteui-theming-mcp"]
    }
  }
}

VS Code / GitHub Copilot (.vscode/mcp.json, root key servers):

{
  "servers": {
    "igniteui-cli": {
      "command": "npx",
      "args": ["-y", "igniteui-cli", "mcp"]
    },
    "igniteui-theming": {
      "command": "npx",
      "args": ["-y", "igniteui-theming", "igniteui-theming-mcp"]
    }
  }
}

Angular schematic — also adds an angular-cli server:

{
  "mcpServers": {
    "angular-cli": {
      "command": "npx",
      "args": ["-y", "@angular/cli", "mcp"]
    }
  }
}

Clone this wiki locally