Laravel AI Development Agent

An MCP-powered autonomous development toolkit for Laravel projects.

Give Claude / OpenCode / Cursor a Laravel engineering brain.

A Model Context Protocol (MCP) server that makes your AI coding agent understand, generate and debug Laravel applications. Works with any MCP-compatible client: Claude Desktop, Cursor, Codex, OpenCode, and more.

Features

✅ Understand Laravel project structure — models, routes, migrations, packages, frontend — injected automatically

✅ Generate complete features — from a natural-language request to working CRUD, Blade views, and REST APIs

✅ Debug exceptions — locate the file, diagnose common Laravel errors, suggest fixes

✅ Generate REST APIs — migration, model, API controller, Sanctum-protected routes, tests

✅ Analyze database — schema, tables, columns, migrations, relationships

✅ Safe Artisan execution — command whitelist, dangerous commands blocked, sensitive data redacted

Demo

Once connected, ask your AI agent in any MCP-compatible client:

> Add a comment feature to the blog
→ intentPlanner: parsed as create_feature/Comment
→ createFeature: migration → model → controller → request → views → test
→ workflowStatus: run persisted to .mcp/runs/

> List the routes
→ routeList: 47 routes, filterable by name

> SQLSTATE[42P01] Table not found — how to fix?
→ debugWorkflow: locate + diagnose + suggest fixes

Tools

Tool	Description
artisan	Run any php artisan command (whitelist restricted)
migrateStatus	Check migration status
envInfo	Display APP_ENV, APP_DEBUG, DB connection
cache	Clear/cache config, routes, views
configGet	Inspect config values by key
schema	List database tables, view column definitions
model	Scan Eloquent models in app/Models
log	Tail the Laravel log file
routeList	List routes with optional name/URI/method filter
runTest	Run PHPUnit tests with optional filter
envInfoSafe	Read .env filtering out sensitive values
frontendScanner	Scan resources/views, js, css structure
makeModel	Create a new Eloquent model with optional migration/factory
makeController	Create a controller with optional resource/api/model binding
makeMigration	Create a new migration file
migrationAnalyzer	Parse migrations and extract schema (columns, types, FKs)
composerAnalyzer	List project dependencies with version info
projectContext	Get comprehensive project context (version, models, routes, packages)
crudGenerator	Generate full CRUD: migration, model, controller, request, route, test
createFeature	Generate full feature: CRUD + Blade views
apiGenerator	Generate REST API: migration, model, API controller, routes, tests
debugWorkflow	Analyze an error: locate, diagnose, suggest fixes
intentPlanner	Parse a natural-language dev request into an executable plan
workflowStatus	List/inspect/resume/rollback workflow runs

Usage

Installation

# From npm (demo version)
npm install laravel-ai-agent@beta

# From GitHub (demo version)
npm install github:dok4everak47/laravel-ai-agent

With OpenCode

Add to ~/.config/opencode/opencode.jsonc:

{
  "mcpServers": {
    "laravel": {
      "type": "local",
      "command": ["node", "/path/to/laravel-ai-agent/dist/index.js"]
    }
  }
}

With Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "laravel": {
      "command": "node",
      "args": ["/path/to/laravel-ai-agent/dist/index.js"]
    }
  }
}

Environment Variables

Variable	Default	Description
LARAVEL_PROJECT_PATH	process.cwd()	Path to Laravel project
PHP_PATH	php	PHP executable path
LLM_API_KEY	(empty)	Enables the LLM semantic layer for intentPlanner (e.g. DeepSeek/OpenAI key)
LLM_BASE_URL	https://api.deepseek.com/v1	OpenAI-compatible API base URL
LLM_MODEL	deepseek-chat	LLM model used for intent analysis

Requirements

• Node.js 18+
• PHP 8.1+ (for Laravel project interaction)
• A Laravel project with artisan in its root

Development

npm install
npm run typecheck   # tsc --noEmit (type-check only)
npm test            # run all tests
npm run build       # compile to dist/
npm start           # node dist/index.js
npm run dev         # npx tsx src/index.ts (hot reload)

Security Model

✓ Local only — runs entirely on your machine, no server component ✓ No telemetry — sends zero usage data ✓ No external API by default — works fully offline; the LLM semantic layer only makes network calls when LLM_API_KEY is set ✓ Command whitelist — artisan restricted to safe commands ✓ Dangerous commands blocked — db:wipe, migrate:fresh, tinker, shell ✓ Sensitive data redaction — .env values filtered before returning

See SECURITY.md for details.

License

MIT