cdc-debezium

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill includes plaintext passwords in example SQL and connector JSON (e.g., "secret" in CREATE ROLE/USER and "database.password"), which requires the LLM to reproduce secret values verbatim when emitting those examples or configs.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The required workflow consumes Debezium CDC events from Kafka topics (e.g., consumer.subscribe(["pgserver1.public.customers"]) / Flink reading topic = 'pgserver1.public.customers'), where message values can include outsider-authored free text originating from posted data into the source database and thus be ingested by the LLM at runtime.