data-contracts

Data contracts prevent schema drift by versioning a datacontract.yaml that specifies schema, SodaCL quality assertions, and SLAs. You enforce these contracts in CI/CD pipelines using automated lint, test, diff, and breaking commands to guard pull requests against unexpected changes.

A data contract is a single versioned specification that prevents quality and SLA surprises by giving producers and consumers a shared agreement. It defines data structure, validation rules, and operational guarantees for datasets across Kafka, S3, and databases.

You detect breaking changes in data schemas by running the breaking command in your CI pipeline against previous data contract versions. This automated comparison uses semver changelog-driven rules to block deployments when incompatible structural modifications are introduced.

Yes, you can integrate data contracts with DataHub and OpenMetadata. The workflow supports exporting and publishing your contract assertions directly to these governance catalogs to maintain centralized schema and quality documentation.

Yes, the data contract specification fully supports embedding SodaCL-style assertions for freshness, completeness, and uniqueness, alongside custom SQL-based quality checks directly within your datacontract.yaml file.

The best way to author a data contract for Kafka or S3 datasets is to use an AI agent to generate a datacontract.yaml conforming to datacontract.com. This file defines identity, servers, models, quality checks, terms, and service levels in one specification.