What problem does it solve? Editing a Kafka Connect connector JSON is risky: a renamed property, a wrong topic route, or a mismatched SMT chain silently breaks a customer data pipeline. This Skill enforces a disciplined change loop — edit, validate, diff, deploy, and prove with row-count checksums — so every connector change ends with a RUNNING connector and verified data in the sink, not just a saved file. ## Core Features & Use Cases - Versioned config management: Enforces the house naming convention where every change is a new timestamped file, keeping the previous version as an instant rollback. - Safe deploy loop: Validates configs via the Connect config/validate endpoint, diffs against the live connector, deploys with create-or-update, and checks every task reaches RUNNING. - Data proof with IVT: Runs source and sink checksum SQL per table and compares COUNT/CHECKSUM_AGG so mismatches are caught before handoff. - Debezium 1.x to 3.x migration: Documents observed property renames (e.g. database.server.name to topic.prefix) and topic layout changes for upgrade projects. - Use Case: When asked to add a table to a Debezium SQL Server source connector, the Skill updates the include list, predicate, sink topics, and IVT SQL, then deploys to uat and proves row counts before touching prd. ## Quick Start Ask the agent to add a new table to the uat Debezium source connector for the HRMI pipeline and verify the data lands in the staging sink.
