# Keep — Open-Source AIOps Alert Management Platform

> Centralize alerts from monitoring tools like Datadog, Grafana, and PagerDuty into one platform with noise reduction, correlation, and automated workflows.

## Install

Save as a script file and run:

# Keep — Open-Source AIOps Alert Management Platform

## Quick Use
```bash
docker run -d -p 8080:8080 us-docker.pkg.dev/keephq/keep/keep-api:latest
# Or install via pip
pip install keep
keep api
```

## Introduction
Keep is an open-source AIOps platform that centralizes alerts from dozens of monitoring tools into a single pane of glass. It reduces alert fatigue through noise reduction and intelligent correlation, letting SRE and DevOps teams focus on real incidents instead of drowning in notifications.

## What Keep Does
- Ingests alerts from 50+ providers including Datadog, Grafana, PagerDuty, Prometheus, and CloudWatch
- Correlates related alerts into deduplicated incidents automatically
- Provides a workflow engine to automate enrichment, routing, and remediation
- Offers a unified timeline view across all monitoring sources
- Supports bidirectional sync so actions in Keep reflect in source tools

## Architecture Overview
Keep runs as a Python-based API server backed by a database (SQLite for dev, PostgreSQL for production). It connects to monitoring providers through a plugin system of 'providers' that handle both ingest (webhooks, polling) and actions (acknowledge, resolve). A built-in workflow engine evaluates rules against incoming alerts and triggers automated responses. The frontend is a Next.js application.

## Self-Hosting & Configuration
- Deploy with Docker Compose or Helm chart on Kubernetes
- Configure providers via the web UI or YAML-based configuration files
- Set up webhook endpoints for push-based alert ingestion
- Use environment variables for database connection, auth settings, and secret keys
- Supports SSO via OIDC and SAML for enterprise authentication

## Key Features
- Alert deduplication and noise reduction cuts alert volume significantly
- Workflow automation with a visual builder for enrichment and routing rules
- Topology-aware correlation groups related alerts into incidents
- Multi-tenant architecture supports team-based access control
- Built-in AI capabilities for root cause suggestion and alert summarization

## Comparison with Similar Tools
- **PagerDuty** — commercial incident management; Keep is open-source and focuses on alert aggregation
- **Grafana OnCall** — strong Grafana integration but narrower provider support
- **Opsgenie** — Atlassian's alerting tool; Keep offers broader third-party ingestion
- **BigPanda** — enterprise AIOps with heavy ML; Keep is lighter and self-hostable
- **ilert** — commercial alerting platform; Keep provides similar features without vendor lock-in

## FAQ
**Q: Does Keep replace my existing monitoring tools?**
A: No. Keep sits on top of your monitoring stack. It aggregates and correlates alerts from your existing tools rather than generating its own.

**Q: What databases does Keep support?**
A: SQLite for development and PostgreSQL for production deployments.

**Q: Can I write custom integrations?**
A: Yes. Keep's provider system is extensible. You can add new providers by implementing the provider interface in Python.

**Q: Is there a hosted version?**
A: The Keep team offers a managed cloud version, but the platform is fully self-hostable under the MIT license.

## Sources
- https://github.com/keephq/keep
- https://docs.keephq.dev

---
Source: https://tokrepo.com/en/workflows/asset-f10159bb
Author: Script Depot