home / mcp / scaffold mcp suite mcp server

Scaffold MCP Suite MCP Server

Toolkit for Coding Agents to work reliably with repo of any size.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "agiflow-aicode-toolkit": {
      "command": "npx",
      "args": [
        "-y",
        "@agiflowai/scaffold-mcp",
        "mcp-serve",
        "--admin-enable"
      ]
    }
  }
}

You harness the MCP (Machine Collaboration Platform) to encode your team’s conventions as reusable, enforceable patterns. With a MCP server setup, you can generate boilerplate, guide design patterns, and apply style rules across multiple AI tools to keep code consistent and compliant as you scale.

How to use

Interact with your MCP-enabled agents using natural language to plan, scaffold, and refine code. Start by selecting an MCP server configuration that matches your workflow (scaffold-mcp to generate boilerplates, architect-mcp to enforce design patterns, and style-system to align styling with your design tokens). Each server runs as a local process through the MCP command runner, and you can issue commands that the agents execute in sequence to produce project scaffolds, features, or style updates.

How to install

{
  "mcpServers": {
    "scaffold-mcp": {
      "command": "npx",
      "args": ["-y", "@agiflowai/scaffold-mcp", "mcp-serve", "--admin-enable"]
    },
    "architect-mcp": {
      "command": "npx",
      "args": [
        "-y", "@agiflowai/architect-mcp", "mcp-serve",
        "--admin-enable",
        "--design-pattern-tool", "codex",
        "--review-tool", "gemini-cli"
      ]
    },
    "style-system": {
      "command": "npx",
      "args": ["-y", "@agiflowai/style-system", "mcp-serve"]
    }
  }
}

Additional setup notes

Prerequisites include Node.js v18 or later. You will run MCP servers locally as described in the installation steps. If you use multiple MCP tools, you’ll manage them through the shared MCP configuration so that templates, design patterns, and style rules stay synchronized across tools.

Architecture

Use scaffold-mcp, architect-mcp, and style-system together to enable end-to-end project creation, pattern validation, and design-system alignment. The agents feed from templates with generation rules and coding standards, while the architecture ensures generated code follows your team’s conventions.

Workflow

Plan tasks in advance and hand them to agents to run autonomously. In copilot mode, prompt the agent with natural language or use the CLI to troubleshoot. This approach helps you create new projects, add features, and ensure code quality without repetitive manual checks.

Template structure

Templates are organized with scaffold, patterns, and rules that drive code generation and validation. You typically have scaffold.yaml to define boilerplates, architect.yaml for design patterns, and RULES.yaml for coding standards. Use templates to quickly bootstrap new apps or libraries while preserving your conventions.

Supported agents and tools

You can use Claude Code, Cursor, Gemini CLI, Codex CLI, GitHub Copilot, and Windsurf (coming soon) as MCP-enabled agents. Each tool connects to the MCP servers defined in your configuration to perform generation, review, and styling tasks.

Project types

The MCP toolkit supports monorepos and monolith projects. You can designate a template per project to apply specific scaffolds, patterns, and rules, enabling consistent behavior across diverse codebases.

Notes and troubleshooting

If an agent cannot recognize a requested action, restart the agent or verify that the MCP servers are running and reachable. Ensure you have the correct CLI arguments and that the specified tools and patterns exist in your templates.

Example tasks and commands

Create a Next.js app from a boilerplate, then add a route and style it using the design system. The MCP workflow will chain scaffold, design-pattern checks, and style resolution to produce a compliant, ready-to-run project.

Security and access

Run MCP servers locally with admin capabilities disabled or enabled as needed. When you enable admin features, template creation and pipeline adjustments become available to trusted team members.

Packages

The MCP tooling suite includes aicode-toolkit for initialization and template management, scaffold-mcp for code scaffolding, architect-mcp for pattern enforcement, style-system for design-system operations, and one-mcp for token-efficient MCP usage.

Available tools

list-boilerplates

Show available project templates that you can scaffold from

use-boilerplate

Create a new project from a selected boilerplate template

list-scaffolding-methods

List all feature scaffolding options you can add to a project

use-scaffold-method

Add a feature scaffold such as a route or service to your project

get-file-design-pattern

Retrieve design-pattern guidance before editing a file to ensure alignment with conventions

review-code-change

Review edited code and report violations or improvements after changes

list_themes

List available design themes in the design system

get_css_classes

Extract CSS classes from a theme to guide styling decisions

get_component_visual

Preview a UI component visualization without running the app

list_shared_components

Find existing shared components to reuse or extend

list_app_components

List app-specific components and package components for reuse

add-design-pattern

Add a new design pattern to the template’s architect.yaml for enforcing guidelines

add-rule

Add a new coding rule to the template’s RULES.yaml for enforcement