AgentSkillsCN

CDC Pattern Implementer

实现变更数据捕获模式,以实现实时数据集成。

SKILL.md
--- frontmatter
name: CDC Pattern Implementer
description: Implements Change Data Capture patterns for real-time data integration
version: 1.0.0
category: Data Integration
skillId: SK-DEA-013
allowed-tools:
  - Read
  - Write
  - Edit
  - Glob
  - Grep
  - Bash

CDC Pattern Implementer

Overview

Implements Change Data Capture patterns for real-time data integration. This skill provides expertise in CDC configuration and implementation across various database and streaming platforms.

Capabilities

  • Debezium connector configuration
  • CDC pattern selection (log-based, trigger-based, timestamp-based)
  • Initial snapshot strategy
  • Schema change handling
  • Exactly-once delivery configuration
  • Sink connector setup
  • Tombstone handling
  • CDC monitoring setup

Input Schema

json
{
  "sourceDatabase": {
    "type": "postgres|mysql|oracle|sqlserver",
    "connection": "object"
  },
  "tables": ["string"],
  "targetSystem": "kafka|kinesis|pubsub",
  "requirements": {
    "latencyMs": "number",
    "exactlyOnce": "boolean"
  }
}

Output Schema

json
{
  "connectorConfig": "object",
  "snapshotStrategy": "object",
  "schemaConfig": "object",
  "monitoringConfig": "object",
  "documentation": "string"
}

Target Processes

  • ETL/ELT Pipeline
  • Streaming Pipeline
  • Data Warehouse Setup

Usage Guidelines

  1. Identify source database and tables for CDC
  2. Define target streaming system
  3. Specify latency and delivery guarantees
  4. Configure appropriate snapshot strategy for initial load

Best Practices

  • Use log-based CDC when possible for minimal source impact
  • Plan initial snapshot strategy carefully for large tables
  • Implement proper error handling and dead letter queues
  • Monitor replication lag and connector health
  • Test schema evolution handling before production