MCPModel Context Protocol

AI Agent Integration

Name: Voltfast CLI
Author: frizzyondabeat

Voltfast ships a built-in MCP server so AI agents can scaffold frontend projects without human interaction — generating config files and install commands on demand.

Advisory mode

The MCP server never writes to disk. Every tool returns file contents and install commands for the agent to apply — keeping the agent in full control of the filesystem.

detect_project plan_setup plan_test_setup

Installation

Claude Code

RECOMMENDED

Run this command once to register the server globally — available in every project session.

terminal

$ claude mcp add --scope user volt-fast npx -- -y --package=@frizzyondabeat/volt-fast volt-fast-mcp

Claude Desktop · Trae · Cursor · Windsurf

All these clients share the same mcpServers JSON format. Add the block below to your client's config file, then restart the IDE.

mcp config

{
  "mcpServers": {
    "volt-fast": {
      "command": "npx",
      "args": ["−y", "--package=@frizzyondabeat/volt-fast", "volt-fast-mcp"]
    }
  }
}

Claude DesktopmacOS ~/Library/Application Support/Claude/claude_desktop_config.jsonWindows %APPDATA%\Claude\claude_desktop_config.json

TraeSettings → Extensions → MCP Servers → Add Server, or edit:macOS ~/Library/Application Support/Trae/mcp_settings.jsonWindows %APPDATA%\Trae\mcp_settings.json

CursorGlobal config (applies to every project):macOS ~/.cursor/mcp.jsonWindows %USERPROFILE%\.cursor\mcp.jsonProject-scoped: .cursor/mcp.json at project root

WindsurfmacOS ~/.codeium/windsurf/mcp_config.jsonWindows %USERPROFILE%\.codeium\windsurf\mcp_config.json

VS Code

GitHub Copilot

Create .vscode/mcp.json at the project root. VS Code uses a servers key (not mcpServers) and requires an explicit type field.

.vscode/mcp.json

{
  "servers": {
    "volt-fast": {
      "type": "stdio",
      "command": "npx",
      "args": ["−y", "--package=@frizzyondabeat/volt-fast", "volt-fast-mcp"]
    }
  }
}

Requires VS Code ≥ 1.99 with the GitHub Copilot extension and MCP support enabled in settings.

Tools Reference

Three tools cover the full setup lifecycle — detect, plan, apply. All inputs are validated with Zod schemas.

detect_projectDetection

Inspects a directory and returns its framework, package manager, and installed toolchain. Call this first — its output feeds directly into plan_setup and plan_test_setup.

Input

{
  projectDir: string  // absolute path to project
}

Output

{
  packageManager: "npm" | "yarn" | "pnpm" | "bun",
  detectedTools: string[],   // e.g. ["nextjs", "typescript"]
  hasTypeScript: boolean,
  hasNextjs:     boolean,
  hasVite:       boolean,
  hasTailwind:   boolean
}

plan_setupConfiguration

Generates every config file and the exact install command for any combination of ESLint, Prettier, Tailwind CSS, Husky, Commitlint, and Shadcn. Returns content — never writes to disk.

Input

{
  tools:          ("tailwind"|"eslint"|"prettier"|
                   "husky"|"commitlint"|"shadcn")[],
  projectDir?:    string,   // auto-detects framework + PM
  detectedTools?: string[], // override detection
  packageManager?: "npm"|"yarn"|"pnpm"|"bun",
  settings?: {
    filenameConvention?: "KEBAB_CASE"|"PASCAL_CASE"|...,
    tailwindCssPath?:    string,
    huskySettings?:      { ... }
  }
}

Output

{
  packageManager:  string,
  detectedTools:   string[],
  packages:        string[],
  installCommand:  string,   // e.g. "pnpm add -D eslint ..."
  configs: [
    { path: string, content: string }
  ],
  notes: string[]  // manual steps (shadcn, husky init)
}

plan_test_setupTesting

Generates a test runner scaffold for Vitest, Jest, or Cypress — including config files, package.json scripts, and the install command. Framework is auto-detected when projectDir is provided.

Input

{
  projectDir?: string,                 // auto-detects framework
  framework?:  "nextjs"|"vite"|"generic",
  runner:      "vitest"|"jest"|"cypress",
  hasTs?:      boolean  // default: true
}

Output

{
  framework:      string,
  runner:         string,
  packageManager: string,
  packages:       string[],
  installCommand: string,
  scripts: {
    "test":          string,
    "test:coverage": string,
    ...
  },
  configs: [
    { path: string, content: string }
  ],
  notes: string[]
}

Agent Workflow

A typical agent run — scaffolding a Next.js project with ESLint, Prettier, Husky, and Vitest in four tool calls.

01Detect the project

// agent calls detect_projecttool: detect_projectprojectDir: "/my-nextjs-app"

// server responds✔ packageManager: "pnpm"✔ detectedTools: ["nextjs", "typescript"]

The agent inspects the project directory to determine the framework and package manager.

02Generate the setup plan

// agent calls plan_setuptool: plan_setuptools: ["eslint", "prettier", "husky"]detectedTools: ["nextjs", "typescript"]packageManager: "pnpm"

✔ installCommand: "pnpm add -D eslint ..."✔ configs: [eslint.config.mjs, ...]

The agent requests config files and install commands for the desired tooling.

03Apply the plan

// agent writes files✔ write eslint.config.mjs✔ write prettier.config.mjs✔ write .husky/pre-commit

// agent runs install✔ pnpm add -D eslint prettier husky

The agent writes the config files and runs the install command returned by the server.

04Scaffold tests

// agent calls plan_test_setuptool: plan_test_setuprunner: "vitest"projectDir: "/my-nextjs-app"

✔ write vitest.config.ts✔ write __tests__/example.test.tsx✔ scripts → package.json

Optionally, the agent scaffolds a test runner with a single call.