Cerbos is an open-source authorization engine that decouples access control into Git-managed YAML policies with sub-millisecond local evaluation.
Cerbos is an open-source authorization engine that decouples access control logic from application code into declarative, version-controlled policies. Operating as a stateless Policy Decision Point sidecar or service, it evaluates contextual access requests against rules written in YAML and Google's Common Expression Language. This architecture lets teams update permissions across distributed services without modifying codebase logic or restarting workloads.
In most applications, authorization starts as role checks hardcoded into API endpoints or database queries. As systems grow, those checks turn into brittle conditional branches scattered across repositories. General-purpose engines like Open Policy Agent decouple policy across infrastructure, but they require learning specialized domain languages like Rego. Cerbos focuses strictly on application permissions, offering structured schemas and expression syntax that developers and product owners can read and maintain together.
From a workflow perspective, Cerbos treats access policies as standard code artifacts. Rules live in Git alongside test suites, enabling teams to validate changes in CI using the built-in test runner before deployment. The engine loads policies from local disks, Git repositories, object stores, or databases, refreshing in-memory rulesets without service restarts. By compiling policies into optimized internal evaluation trees, Cerbos avoids external network lookups during checks, keeping decisions fast and predictable.
PlanResources that applications map to SQL WHERE clauses to filter large datasets before retrieval.cerbos run-tests for automated verification in continuous integration pipelines.go-git, which lacks sparse checkouts, fails on Azure DevOps repositories, and requires manual SSH host key configuration in containers.PlanResources into SQL requires custom adapters, with prebuilt support currently limited to Prisma and SQLAlchemy.Is Cerbos free for commercial use? Yes. The Policy Decision Point is open-source under the Apache-2.0 license, while Cerbos Hub provides a commercial managed platform for collaborative policy editing and distribution.
What is the difference between Cerbos and Open Policy Agent (OPA)? OPA is a general-purpose engine using Rego for infrastructure and Kubernetes admission, while Cerbos specializes in application access control using YAML and Common Expression Language.
How does Cerbos handle filtering large lists of records?
Applications call the PlanResources API to generate an abstract syntax tree of authorization conditions, which the application maps to database SQL filters.
Cerbos is an effective choice for engineering teams that need structured, context-aware authorization without the burden of bespoke policy DSLs or scattered permission checks. Active maintenance with commits through late 2026 and an Apache-2.0 license make it straightforward to adopt in production microservices or embed within commercial platforms. Teams that only need simple static roles or an out-of-the-box user login interface, however, will find a dedicated policy engine unnecessary.
Go, gRPC and Protobuf done right, LLM agents in production, and the hard parts of shipping a first product.
Show that Cerbos is listed as an open-source alternative. Paste into your README: