● Beta — KubeEra is in active development. Profiles are live; AI-perception measurement is rolling out.
KubeEra
K

Keep

Observability
Open source · NOASSERTION

Keep is an open-source alert management and AIOps platform.

updated 2026-07-01

What it is

Keep is an open-source alert management and AIOps platform. It ingests alerts from your existing monitoring stack (Prometheus, Datadog, Grafana, PagerDuty, CloudWatch, and dozens of other providers), correlates and deduplicates them, and lets you build workflows to auto-remediate or route incidents. Think of it as a control plane that sits between "something fired an alert" and "a human or automation does something about it" — instead of replacing your monitoring tools, it unifies the noise coming out of them.

Official project description is not publicly available, but based on the codebase and docs, it positions itself against tools like Opsgenie/PagerDuty on the alerting side, with an added correlation and workflow-automation layer that leans into the "AIOps" framing.

Who builds it and why

Not publicly available in terms of founding story or backing company details, but the project is active and Python-based, with 186 contributors and commits as recent as the data snapshot shows. That contributor count relative to a single-purpose alerting tool suggests a maintained core team plus a meaningful outside contributor base, not just a corporate side project gathering dust.

Production readiness signal

  • 11,996 GitHub stars — solid traction for a category-specific tool, not mega-scale but healthy.
  • 186 contributors — indicates real distributed development, not a single-maintainer risk.
  • Latest release v0.54.1 — still pre-1.0. Treat this as a signal: the API and workflow DSL may still change in breaking ways between minor versions. Read changelogs before upgrading in production.
  • License: NOASSERTION — this is a real flag. GitHub couldn't detect a clear SPDX-recognized license from the repo metadata. Before you deploy this in a company with any legal review process, go find the actual LICENSE file and get your legal team to confirm what it says. Don't assume MIT/Apache just because it's on GitHub.
  • CNCF maturity: not publicly available — it does not appear to be a CNCF sandbox/incubating/graduated project as of this data pull. That means no CNCF governance guarantees around neutrality or long-term stewardship.

Who should use this

  • Teams drowning in alert noise from multiple tools (Prometheus + Datadog + CloudWatch, etc.) who want one place to correlate and dedupe before paging humans.
  • Platform/SRE teams who want alert workflows (auto-enrichment, auto-ticketing, conditional routing) without buying a full enterprise incident management suite.
  • Organizations that want to self-host alerting logic instead of sending alert payloads to a third-party SaaS.
  • Teams comfortable running and maintaining a Python service themselves — this isn't a managed product by default.

Who should NOT use this

  • Teams needing on-call scheduling, escalation policies, and SLA-backed incident response at enterprise scale — PagerDuty and Opsgenie are more mature here.
  • Anyone who needs a hard SLA-backed vendor relationship; pre-1.0 open source software is not that.
  • Organizations with strict OSS license compliance requirements — the NOASSERTION license status needs resolving before legal will sign off.
  • Teams wanting a "set and forget" tool. Pre-1.0 releases mean you should expect to babysit upgrades.
  • Small teams with a single alert source (e.g., just Prometheus + Alertmanager) — you likely don't need a correlation layer yet; Alertmanager's native routing may suffice.

Alternatives

  • PagerDuty / Opsgenie — mature, SLA-backed commercial incident management with on-call scheduling; the default enterprise choice if you don't want to self-host.
  • Grafana OnCall — open-source alerting/on-call tool tightly integrated with the Grafana stack, good if you're already Grafana-centric.
  • Alertmanager (native) — if your alert volume and source count is low, you may not need a dedicated correlation layer at all.

Pricing

Core project is open source (self-hosted, run it yourself, no license fee to use the code). Exact license terms are unresolved (NOASSERTION on GitHub) — confirm the actual license file before production use. If Keep offers a hosted/managed SaaS tier with paid pricing, that detail is not publicly available in the data provided — check their site directly before assuming free = the whole story.

Frequently asked

What is Keep?+
Keep is an open-source alert management and AIOps platform. It ingests alerts from your existing monitoring stack (Prometheus, Datadog, Grafana, PagerDuty, CloudWatch, and dozens of other providers), correlates and deduplicates them, and lets you build workflows to auto-remediate or route incidents.
Who builds Keep?+
Not publicly available in terms of founding story or backing company details, but the project is active and Python-based, with 186 contributors and commits as recent as the data snapshot shows.
Is Keep production ready?+
11,996 GitHub stars — solid traction for a category-specific tool, not mega-scale but healthy. 186 contributors — indicates real distributed development, not a single-maintainer risk. Latest release v0.54.1 — still pre-1.0.
Who should use Keep?+
Teams drowning in alert noise from multiple tools (Prometheus + Datadog + CloudWatch, etc.) who want one place to correlate and dedupe before paging humans. Platform/SRE teams who want alert workflows (auto-enrichment, auto-ticketing, conditional routing) without buying a full enterprise incident management suite.
Who should not use Keep?+
Teams needing on-call scheduling, escalation policies, and SLA-backed incident response at enterprise scale — PagerDuty and Opsgenie are more mature here. Anyone who needs a hard SLA-backed vendor relationship; pre-1.0 open source software is not that.
What are the alternatives to Keep?+
PagerDuty / Opsgenie — mature, SLA-backed commercial incident management with on-call scheduling; the default enterprise choice if you don't want to self-host. Grafana OnCall — open-source alerting/on-call tool tightly integrated with the Grafana stack, good if you're already Grafana-centric.
How much does Keep cost?+
Core project is open source (self-hosted, run it yourself, no license fee to use the code). Exact license terms are unresolved (NOASSERTION on GitHub) — confirm the actual license file before production use.