gamussa/flink-sql
Apache Flink SQL, Table API, and UDF development for both OSS Flink and Confluent Cloud
92
89%
Does it follow best practices?
Impact
98%
1.22xAverage score across 5 eval scenarios
Advisory
Suggest reviewing before use
Quality
Discovery
100%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is an excellent skill description that hits all the marks. It provides specific concrete actions, uses natural trigger terms that Flink developers would actually search for, includes an explicit and detailed 'Use when' clause with six enumerated scenarios, and occupies a clearly distinct niche. The description is comprehensive without being verbose.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions: writing Flink SQL queries (with subtypes like windows, joins, aggregations, MATCH_RECOGNIZE), building Table API pipelines, creating UDFs, deploying to Confluent Cloud, converting between DataStream and Table API, and troubleshooting errors. | 3 / 3 |
Completeness | Clearly answers both 'what' (Flink SQL, Table API, and UDF development for OSS Flink and Confluent Cloud) and 'when' with an explicit numbered 'Use when:' clause listing six specific trigger scenarios. | 3 / 3 |
Trigger Term Quality | Excellent coverage of natural terms users would say: 'Flink SQL', 'Table API', 'UDF', 'Confluent Cloud', 'watermarks', 'windowing', 'event-time processing', 'MATCH_RECOGNIZE', 'DataStream', 'state management'. These are the exact terms a developer working with Flink would use. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive niche targeting Apache Flink SQL and Confluent Cloud specifically. The combination of Flink SQL, Table API, UDFs, and Confluent-specific patterns makes it very unlikely to conflict with other skills like general SQL or generic streaming skills. | 3 / 3 |
Total | 12 / 12 Passed |
Implementation
72%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a strong, well-structured skill that covers a complex domain comprehensively. Its greatest strengths are the highly actionable code examples across SQL, Java, and Python, and the excellent progressive disclosure with clear references to detailed guides. The main areas for improvement are trimming some explanatory content that Claude already knows (changelog concepts, what PDFs are equivalent explanations) and adding validation checkpoints to multi-step workflows like UDF deployment.
Suggestions
Add validation checkpoints to the UDF deployment workflow (e.g., verify artifact upload with `confluent flink artifact list` before running CREATE FUNCTION)
Move the environment setup section (Maven dependencies, pip install, env vars) to the confluent-cloud.md reference file to reduce the main file's length and keep it focused on patterns
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is mostly efficient with good use of tables and code blocks, but it's quite long for a top-level SKILL.md. Some sections like the mental model explanation and changelog modes could be trimmed since Claude understands these concepts. The environment setup section with boilerplate dependency XML and env vars is verbose for the overview file and could be referenced out. | 2 / 3 |
Actionability | Excellent actionability throughout — every pattern includes fully executable SQL or Java/Python code that is copy-paste ready. The UDF deployment section includes concrete CLI commands, and the troubleshooting table maps specific errors to specific fixes. | 3 / 3 |
Workflow Clarity | The UDF deployment section has a clear sequence (build → upload → register), but lacks validation checkpoints (e.g., verifying the artifact upload succeeded before registering). The critical rules section lists important constraints but doesn't integrate them into workflows. For a skill involving deployment and state management, explicit validation steps would strengthen this. | 2 / 3 |
Progressive Disclosure | Excellent progressive disclosure with a clear overview in the main file and well-signaled one-level-deep references to six separate files (udf-guide.md, sql-patterns.md, confluent-cloud.md, troubleshooting.md, ptf-guide.md, kafka-patterns.md). The quick reference table at the top also links to the PTF guide. References are clearly labeled with descriptions. | 3 / 3 |
Total | 10 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
Reviewed
Table of Contents