earthgecko/skyline

Real-time anomaly detection for time series metrics. 11 years of production anomaly detection evolution. Multi-algorithm ensembles. Pattern recognition using semi-supervised and unsupervised learning. Correlation analysis. Learns and gets better over time. No GPU required, can run on a VPS.

What it solves

Skyline is designed for environments with a large volume of high-resolution time series data where manually configuring thresholds or models for every single metric is impractical. It provides a way to passivly monitor metrics and automatically detect anomalies without manual setup for each stream.

How it works

Skyline uses Python-based algorithms to automatically determine what constitutes an anomaly for each specific metric. It integrates with common data sources like Graphite, InfluxDB, Prometheus, and VictoriaMetrics to ingest metrics. Once a stream is established, new metrics are added automatically for analysis, and users can manually train the system on a per-metric basis to refine what is considered non-anomalous.

Who it’s for

It is intended for engineers and operators who need to monitor a high volume of metrics across their infrastructure and want to avoid the manual overhead of configuring individual alerts.

Highlights

  • Automatic anomaly detection for high-resolution time series.
  • Passive monitoring that removes the need for manual threshold configuration.
  • Support for multiple ingestion sources including Prometheus, InfluxDB, and Graphite.
  • Ability to train the system on a per-metric basis to reduce false positives.

Related

  • Project
    grafana/promql-anomaly-detection

    A PromQL-based framework for detecting anomalies in time series data using recording and alerting rules within Prometheus, requiring no external systems.

  • Project
    sintel-dev/Orion

    A machine learning library for unsupervised time series anomaly detection that provides verified ML pipelines to identify rare patterns in signals.

  • Project
    netdata/netdata

    An open-source, real-time infrastructure monitoring platform that provides per-second metrics and ML-powered anomaly detection with zero configuration.

  • Project
    plasma-umass/scalene

    Scalene is a fast, line‑level Python profiler that measures CPU, GPU (NVIDIA) and memory usage, separates Python from native code, visualizes stitched stacks, and can ask LLMs (OpenAI, Azure, Bedrock, Ollama, etc.) for optimization suggestions directly from the UI.

  • Project
    stumpy-dev/stumpy

    A scalable Python library for computing matrix profiles to identify repeated patterns and anomalies in time series data.