Introduction to Anomstack
Anomstack is an open-source anomaly detection platform that makes it easy to monitor and detect anomalies in your metrics data. Built on top of Dagster for orchestration and FastHTML + MonsterUI for the dashboard, Anomstack provides a complete solution for metric monitoring and anomaly detection.
Key Features
- 🔍 Powerful Anomaly Detection: Built on PyOD for robust anomaly detection
- 📊 Beautiful Dashboard: Modern UI for visualizing metrics and anomalies
- 🔌 Multiple Data Sources: Support for various databases and data platforms
- 🔔 Flexible Alerting: Email and Slack notifications with customizable templates
- 🤖 LLM Agent Integration: AI-powered anomaly analysis and reporting
- 🛠️ Easy Deployment: Multiple deployment options including Docker, Dagster Cloud, and more
How It Works
- Define Your Metrics: Configure your metrics using SQL queries or Python functions
- Automatic Processing: Anomstack handles ingestion, training, scoring, and alerting
- Monitor & Alert: Get notified when anomalies are detected
- Visualize: Use the dashboard to explore metrics and anomalies
Supported Data Sources
Anomstack supports a wide range of data sources:
- Python (direct integration)
- BigQuery
- Snowflake
- ClickHouse
- DuckDB
- SQLite
- MotherDuck
- Turso
- Redshift (coming soon)
Storage Options
Store your trained models and configurations in:
- Local filesystem
- Google Cloud Storage (GCS)
- Amazon S3
- Azure Blob Storage (coming soon)
Getting Started
Choose your preferred way to get started:
Architecture
Anomstack is built with a modular architecture that separates concerns:
- Ingestion: Pull data from various sources
- Processing: Train models and detect anomalies
- Alerting: Send notifications via multiple channels
- Dashboard: Visualize metrics and anomalies
- Storage: Store models and configurations
Contributing
We welcome contributions! Check out our Contributing Guide to get started.
License
Anomstack is open source and available under the MIT License.