test-json-sql-filter_skill

Overview

This skill validates a structured filter primitive by running an end-to-end test that creates sample data and applies a WHERE-style condition. It focuses on filtering primitive fields (numeric comparisons) and confirming the resulting collection matches the expected subset. The test is self-contained and deterministic, making it ideal for CI checks of filtering logic.

How this skill works

The skill generates an internal collection named $papers with citation counts [100, 250, 50, 180] and associated titles. It applies a filter operation equivalent to WHERE citations > 100 and produces a $high_cited collection. The test then asserts that $high_cited contains exactly the two expected items (Transformers: 250, Scaling Laws: 180).

When to use it

• Validate numeric comparison behavior of filter primitives
• Automated tests in CI to catch regressions in WHERE logic
• Verify end-to-end behavior of a query layer without external dependencies
• Demonstrate expected filter semantics for reviewers or maintainers
• Quick sanity checks after refactoring filtering code

Best practices

• Keep test data minimal and self-contained to avoid external flakiness
• Use clear, deterministic expectations (exact item counts and values)
• Test boundary conditions (equal, greater-than, less-than) in companion tests
• Name collections and assertions to clearly indicate intent (e.g., $high_cited)
• Run the test as part of a pipeline to catch regressions early

Example use cases

• CI job that runs unit-level verification of filtering primitives
• Documentation example showing how WHERE > numeric comparisons behave
• Regression test after optimizing query execution or changing comparison operators
• Teaching example for engineers learning how structured filters are evaluated
• Pre-merge check to ensure no change breaks simple numeric filters