DataQuarantine

DataQuarantine is a web app (with optional CLI) that sits between your ingestion and your warehouse/lakehouse to automatically quarantine suspicious records and files instead of letting them silently corrupt downstream tables. It applies lightweight, configurable rules (range checks, null spikes, referential integrity hints, duplicate bursts, schema drift detection) and produces a “quarantine bucket” with clear reasons, sample rows, and suggested fixes. Teams can approve, auto-fix, or reject quarantined data, then replay it back into the pipeline with an audit trail. It’s not a full data quality platform; it’s a focused gatekeeper that reduces firefighting and restores trust in dashboards. The MVP targets common stacks (S3/GCS + Snowflake/BigQuery/Databricks) and integrates with Slack for incident routing. Expect some false positives early; the value is catching the worst issues fast and making remediation repeatable.