detectkit 0.2.3

Metric monitoring with automatic anomaly detection

detectkit

Metric monitoring with automatic anomaly detection

detectkit is a Python library for data analysts and engineers to monitor time-series metrics with automatic anomaly detection and alerting.

Status

✅ Production Ready - Version 0.1.2

Published to PyPI: https://pypi.org/project/detectkit/

Complete rewrite with modern architecture and full documentation (2025).

Features

• ✅ Pure numpy arrays - No pandas dependency in core logic
• ✅ Batch processing - Efficient vectorized operations
• ✅ Multiple detectors - Statistical methods (Z-Score, MAD, IQR, Manual Bounds)
• ✅ Alert channels - Mattermost, Slack, Webhook support
• ✅ Database agnostic - ClickHouse, PostgreSQL, MySQL support
• ✅ Idempotent operations - Resume from interruptions
• 🚧 CLI interface - dbt-like commands (coming soon)

Installation

pip install detectkit

Or from source:

git clone https://github.com/alexeiveselov92/detectkit
cd detectkit
pip install -e .

Optional dependencies

# ClickHouse support
pip install detectkit[clickhouse]

# All database drivers
pip install detectkit[all-db]

# Development dependencies
pip install detectkit[dev]

Quick Start

CLI Usage (Recommended)

# Create a new project
dtk init my_monitoring_project
cd my_monitoring_project

# Configure database in profiles.yml
# Then run your metrics
dtk run --select example_cpu_usage

# Run specific pipeline steps
dtk run --select cpu_usage --steps load,detect

# Run all critical metrics
dtk run --select tag:critical

# Reload data from specific date
dtk run --select cpu_usage --from 2024-01-01

Python API Usage

import numpy as np
from detectkit.detectors.statistical import ZScoreDetector

# Your time-series data
timestamps = np.array([...], dtype='datetime64[ms]')
values = np.array([1.0, 2.0, 1.5, 10.0, 1.8])  # 10.0 is anomaly

# Create detector
detector = ZScoreDetector(threshold=3.0, window_size=100)

# Detect anomalies
data = {
    'timestamp': timestamps,
    'value': values
}
results = detector.detect(data)

# Check results
for result in results:
    if result.is_anomaly:
        print(f"Anomaly at {result.timestamp}: {result.value}")

Architecture

• Detectors - Statistical and ML-based anomaly detection
• Loaders - Metric data loading from databases with gap filling
• Alerting - Multi-channel notifications with orchestration
• Config - YAML-based configuration (dbt-like)

Testing

# Run tests
pytest tests/

# With coverage
pytest tests/ --cov=detectkit --cov-report=html

Current status: 287 tests passing, 87% coverage

Development Status

✅ Completed (Phases 1-6)

• ✅ Phase 1: Core models (Interval, TableModel, ColumnDefinition)
• ✅ Phase 2: Database managers & data loading (MetricLoader, gap filling, seasonality)
• ✅ Phase 3: Statistical detectors (Z-Score, MAD, IQR, Manual Bounds)
• ✅ Phase 4: Alerting system (Channels, Orchestrator, consecutive anomalies)
• ✅ Phase 5: Task manager (Pipeline execution, locking, idempotency)
• ✅ Phase 6: CLI commands (dtk init, dtk run with selectors)

🔄 Integration Status

• ⚠️ Full end-to-end integration pending (database connection required)
• ⚠️ Advanced detectors (Prophet, TimesFM) - optional extras
• ⚠️ Additional alert channels (Telegram, Email) - optional

Documentation

📚 Complete documentation available at: https://github.com/alexeiveselov92/detectkit/tree/main/docs

• Getting Started - 5-minute quickstart
• Configuration Guide - All configuration options
• Detectors Guide - Choosing the right detector
• Alerting Guide - Setting up alerts
• CLI Reference - Command-line documentation
• Examples - Real-world monitoring scenarios

Requirements

• Python 3.10+
• numpy >= 1.24.0
• pydantic >= 2.0.0
• click >= 8.0
• PyYAML >= 6.0
• Jinja2 >= 3.0

License

MIT License - See LICENSE file for details

Contributing

This project is currently in active development. Contributions are welcome once we reach v1.0.0.

Changelog

0.1.0 (2025-11-07)

• Initial release with complete rewrite
• ✅ Core foundation: models, database, config
• ✅ Metric loading with gap filling and seasonality extraction
• ✅ Statistical detectors (Z-Score, MAD, IQR, Manual Bounds)
• ✅ Alert channels (Webhook, Mattermost, Slack)
• ✅ Alert orchestration with consecutive anomaly logic
• ✅ Task manager for pipeline execution
• ✅ CLI commands (dtk init, dtk run)
• 📊 287 unit tests, 87% coverage