kratos

What it is

Kratos is a Go framework for building microservices, originally built at Bilibili (the Chinese video platform) to handle their internal service architecture. It's not a full orchestration platform in the Kubernetes-operator sense — it's an application-layer framework that gives you the scaffolding for building services that run well on Kubernetes: HTTP/gRPC transport layers, service discovery integration, config management, middleware chains, logging, metrics, and code generation via protobuf. Think of it as an opinionated toolkit sitting between your business logic and the infrastructure, closer to Go-kit or Spring Boot territory than to something like Argo or Crossplane.

Who builds it and why

Kratos is maintained by Bilibili's engineering org and an open community of contributors (316 on GitHub). It was built to solve a real internal problem — standardizing how hundreds of internal services were written, deployed, and observed — and was later open-sourced. This is a pattern common to Chinese tech infra projects (similar to how Alibaba open-sourced Dubbo, Sentinel, etc.): built for internal scale first, open-sourced second. The project has no CNCF affiliation and maturity status there is not publicly available — this is an independent project, not one going through CNCF's sandbox/incubation/graduation pipeline.

Production readiness signal

25,764 GitHub stars and 316 contributors is a healthy, active community by any measure. Commit activity through mid-2026 shows it's not abandoned. It's on v3.0.0, meaning it's gone through at least two major breaking-change cycles — a sign of real-world iteration, but also a signal that upgrading major versions may require rework. MIT license is about as permissive as it gets, no legal friction for commercial use. What's missing: no public CNCF maturity assessment, no independently published SLA or adoption case studies beyond Bilibili itself, and English-language documentation/community support is thinner than the Chinese-language ecosystem around it. If you're evaluating this for a Western enterprise environment, expect to lean on translated docs and a smaller pool of engineers with hands-on experience.

Who should use this

Go shops building a microservices architecture from scratch who want an opinionated, batteries-included framework rather than assembling gRPC + service discovery + config + observability libraries themselves. Teams comfortable reading Chinese-language documentation and community discussions (a real chunk of Kratos' ecosystem, examples, and Q&A lives there) will get the most value. Good fit if you want strong protobuf-driven code generation and are already gRPC-centric.

Who should NOT use this

Skip it if you're not writing in Go — this isn't polyglot. Skip it if you're already deep into a specific service mesh (Istio, Linkerd) and don't need framework-level service discovery baked in — you'll fight duplicated concerns. Skip it if your team needs an English-first support ecosystem — Stack Overflow coverage and third-party tutorials are sparse compared to Spring Boot or even Go-kit. Skip it if you need CNCF-backed governance guarantees for procurement/compliance reasons — Kratos has no such standing. And skip it if you want a Kubernetes orchestration/operator tool — despite the "governance framework" naming, this operates at the application layer, not the cluster control plane.

Alternatives

• Go-kit — a more unopinionated, library-style toolkit for Go microservices; more flexible, more assembly required.
• Dapr — a runtime-level sidecar approach to microservice building blocks, language-agnostic and CNCF-incubated, better fit if you're polyglot or want infra-level abstraction instead of framework-level.
• Spring Boot + Spring Cloud — the default choice if your org is JVM-based rather than Go-based, with a much larger English-language enterprise support ecosystem.

Pricing

Fully open source under MIT license. No paid tier, no commercial vendor behind it — costs are engineering time to adopt and maintain, not licensing fees.