Docs / AI Agents

MCP Server

Connect Better Fullstack to Claude Code, Cursor, VS Code, and MCP-compatible AI coding agents.

The Better Fullstack MCP server lets agents scaffold projects through structured tool calls instead of guessing CLI flags. It handles schema lookup, compatibility validation, dry-run previews, project creation, and feature additions.

This page describes the Better Fullstack MCP server. The generated-project mcp addon is separate: it writes MCP client/server configuration into a scaffolded app and may run external setup commands.

For a lighter CLI-driven agent workflow, install the Better Fullstack agent skill. The skill and MCP server both use the same generator; the skill is optimized for fast flag-based scaffolding, while MCP is optimized for structured tool calls.

Setup

Launch the server with the package manager you use:

npx -y create-better-fullstack@latest mcp

Claude Code

claude mcp add --transport stdio better-fullstack -- npx -y create-better-fullstack@latest mcp

Use the same launcher table above if you prefer pnpm, Bun, or Yarn.

Cursor

{
  "mcpServers": {
    "better-fullstack": {
      "command": "npx",
      "args": ["-y", "create-better-fullstack@latest", "mcp"]
    }
  }
}

VS Code

{
  "servers": {
    "better-fullstack": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "create-better-fullstack@latest", "mcp"]
    }
  }
}

Other MCP clients

Use the same stdio command with clients such as Claude Desktop, Zed, Cursor, Codex, OpenCode, Gemini CLI, Goose, or other MCP-compatible tools.

Package manager	`command`	`args`
npm	`npx`	`["-y", "create-better-fullstack@latest", "mcp"]`
pnpm	`pnpm`	`["dlx", "create-better-fullstack@latest", "mcp"]`
Bun	`bunx`	`["create-better-fullstack@latest", "mcp"]`
Yarn	`yarn`	`["dlx", "create-better-fullstack@latest", "mcp"]`

Tools

For full inputs, outputs, examples, and safe write behavior, see the MCP tools reference.

Tool	Purpose
`bfs_get_guidance`	Returns workflow rules and field semantics. Call this first.
`bfs_get_schema`	Returns valid options for configuration categories.
`bfs_check_compatibility`	Validates a stack and returns adjusted selections with warnings.
`bfs_plan_project`	Dry-runs project generation in memory.
`bfs_create_project`	Writes a new project to disk.
`bfs_plan_addition`	Validates addon changes for an existing project.
`bfs_add_feature`	Adds supported addons to an existing project.

Resources

The server exposes these MCP resources:

Resource	Contents
`docs://compatibility-rules`	Compatibility rules for stack combinations.
`docs://stack-options`	Available technology options by category.
`docs://getting-started`	Quick-start recipes for supported ecosystems.

Recommended tool order

For a new project:

bfs_get_guidance
bfs_get_schema
bfs_check_compatibility
bfs_plan_project
bfs_create_project

For an existing project:

bfs_plan_addition
bfs_add_feature

Key rules for agents

frontend is an array and can contain multiple frontend targets.
none disables a category.
Set ecosystem first so the agent uses the correct field family.
Array fields include frontend, addons, examples, rustLibraries, pythonAi, javaLibraries, and javaTestingLibraries.
Project planning, creation, and addition responses include graphSummary, effectiveStack, and stackPartSpecs for the graph-backed stack view.
Current MCP project schemas do not expose aiDocs as a create field; project creation uses the server's built-in AI-doc behavior.
Always call bfs_check_compatibility before creating a project.

Agent prompt

Use Better Fullstack MCP. First call bfs_get_guidance to understand
the rules, then bfs_get_schema for the relevant categories, then
bfs_check_compatibility to validate the stack, then bfs_plan_project
for a dry-run. Only call bfs_create_project after the dry-run succeeds.
Skip normal dependency installation and tell me the exact commands to
install, test, and run the project.