romansh/laravel-creem-agent

Autonomous AI agent for Creem.io store monitoring — heartbeat, proactive workflows, notifications, chat interface

1.2.8 2026-03-28 11:17 UTC

Requires

• php: ^8.2
• illuminate/cache: ^11.0|^12.0
• illuminate/console: ^11.0|^12.0
• illuminate/notifications: ^11.0|^12.0
• illuminate/queue: ^11.0|^12.0
• illuminate/support: ^11.0|^12.0
• laravel/ai: ^0.4
• romansh/laravel-creem: ^1.0
• romansh/laravel-creem-cli: ^1.0 || dev-main

Requires (Dev)

• orchestra/testbench: ^9.0|^10.0
• pestphp/pest: ^3.0

MIT

• monitoring
• laravel
• saas
• Agent
• heartbeat
• ai
• creem

README

Autonomous AI agent for Creem.io store monitoring — heartbeat engine, proactive workflows, multi-store support, chat interface, and notifications.

Installation

composer require romansh/laravel-creem-agent

Setup

php artisan vendor:publish --tag=creem-agent-config
php artisan vendor:publish --tag=creem-agent-migrations
php artisan migrate
php artisan creem-agent:install

Quick Start

# Run first heartbeat
php artisan creem-agent:heartbeat

# Chat with agent
php artisan creem-agent:chat "how many active subscriptions?"

# Check agent status
php artisan creem-agent:status

# Run for all stores
php artisan creem-agent:heartbeat --all-stores

Configuration

• See config/creem-agent.php for:
• Multi-store setup
• Notification channels (Telegram)
• Telegram ownership mode: laravel or openclaw
• Natural-language routing via Laravel AI SDK
• Heartbeat frequency per store
• Quiet hours
• OpenClaw integration

AI / Natural Language Routing

The chat interface now uses laravel/ai to classify natural-language requests into agent intents, with the legacy regex parser kept as a fallback.

Example .env settings:

CREEM_AGENT_LLM_ENABLED=true
CREEM_AGENT_LLM_PROVIDER=openai
CREEM_AGENT_LLM_MODEL=gpt-5.4
CREEM_AGENT_LLM_TIMEOUT=30

OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=
GEMINI_API_KEY=
OPENROUTER_API_KEY=
OLLAMA_BASE_URL=http://localhost:11434

Provider switching is done through CREEM_AGENT_LLM_PROVIDER, while model switching is done through CREEM_AGENT_LLM_MODEL.

When CREEM_AGENT_TELEGRAM_MODE=openclaw, Telegram should be configured in OpenClaw gateway using the native channel model from https://docs.openclaw.ai/channels/telegram. The package can render a starter snippet with:

php artisan creem-agent:openclaw-telegram-config

Architecture

Uses a CLI Facade (CreemCli) that auto-detects:

1. Native creem CLI (brew) → fast shell exec with --json
2. laravel-creem-cli → in-process SDK calls (fallback)

Transparent failover between backends.

Proactive Workflows

• FailedPaymentRecovery — alerts on past_due subscriptions
• ChurnDetection — detects cancellation spikes
• RevenueDigest — summarizes new transactions
• NewCustomerWelcome — celebrates growth
• AnomalyDetection — flags unusual metric drops

License

MIT