🌊 Flink SQL Skill for Claude

A Claude skill for Apache Flink SQL, Table API, and UDF development — covering both OSS Flink and Confluent Cloud.

🎯 What This Skill Enables

When Claude has this skill loaded, it can help you:

• Write Flink SQL queries — windows, joins, aggregations, MATCH_RECOGNIZE
• Build Table API pipelines — Java and Python
• Create UDFs — Scalar functions, Table functions, Process Table Functions (PTFs)
• Deploy to Confluent Cloud — CLI commands, compute pools, artifacts
• Troubleshoot errors — Common mistakes and solutions
• Design streaming architectures — Event-time, watermarks, state management

📚 Skill Contents

flink-sql-skill/
├── SKILL.md                          # Core patterns, quick reference (~330 lines)
└── references/
    ├── udf-guide.md                  # UDF development templates (~400 lines)
    ├── sql-patterns.md               # Advanced SQL patterns (~500 lines)
    ├── confluent-cloud.md            # Confluent-specific features (~480 lines)
    ├── ptf-guide.md                  # Process Table Functions (~725 lines)
    ├── kafka-patterns.md             # Kafka connector patterns (~450 lines)
    └── troubleshooting.md            # Common errors & fixes (~600 lines)

Total: ~3,500 lines of curated Flink knowledge

🚀 How to Use

Option 1: Claude.ai Projects

1. Create a new Project in Claude.ai
2. Go to Project Knowledge
3. Upload SKILL.md and the references/ folder contents
4. Start chatting about Flink!

Option 2: Claude Code CLI

# Clone this repo
git clone https://github.com/gAmUssA/flink-sql-skill.git

# Reference in Claude Code
claude --skill ./flink-sql-skill/

Option 3: Raw GitHub URL

Reference the skill directly:

https://raw.githubusercontent.com/gAmUssA/flink-sql-skill/main/SKILL.md

💡 Example Queries

Once the skill is loaded, try asking Claude:

"Help me write a Flink SQL query with tumbling windows to count orders per hour"

"Create a UDF that masks email addresses for GDPR compliance"

"What's the difference between interval joins and temporal joins?"

"Show me how to deploy a UDF to Confluent Cloud"

"Debug this error: Cannot generate watermark for rowtime column"

🔧 Key Patterns Covered

SQL Patterns

• Window aggregations (TUMBLE, HOP, SESSION, CUMULATE)
• All join types (interval, temporal, lookup, window)
• Deduplication and Top-N queries
• MATCH_RECOGNIZE for pattern detection
• CDC handling with Debezium

UDF Development

• Java ScalarFunction and TableFunction templates
• Python @udf decorator patterns
• Maven POM with shade plugin
• Confluent Cloud deployment workflow

Process Table Functions (PTFs)

• 9 real-world examples from production use cases
• State management (ValueState, MapView, ListView)
• Timer patterns for session detection
• Changelog handling (consuming and producing CDC)
• Multiple input streams (co-partition patterns)

Kafka Integration

• Kafka source/sink table definitions
• Avro + Schema Registry configuration
• Upsert Kafka for stateful outputs
• Flink vs Kafka Streams comparison

🙏 Credits & Sources

This skill incorporates patterns from:

• MartijnVisser/flink-ptf-examples — PTF examples
• confluentinc/flink-table-api-java-examples — Official Confluent examples
• Apache Flink Documentation
• Confluent Cloud Flink Documentation

📖 Related Resources

• Kafka in Action (co-authored by Viktor Gamov)
• FLIP-440: ProcessTableFunction
• Querying Streams YouTube Series

🤝 Contributing

Contributions welcome! Please:

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Run validation: python scripts/validate.py
5. Submit a pull request

📄 License

Apache License 2.0 — See LICENSE for details.

---

Maintainer: Viktor Gamov — Principal Developer Advocate at Confluent

"Tables as movies, not photographs" 🎬