Lean body that assumes competence — no boilerplate on what data validation or Pandas is — and every section adds GX-specific knowledge (object model, gxe namespace, batch_parameters key shapes, result_format values) Claude likely lacks.

Provides fully executable, copy-paste-ready code across expectation authoring, ValidationDefinition.run, Checkpoint with real Actions, a result-triage script, and a CI gate plus GitHub Actions YAML with specific class names and kwargs.

Clear sequence (Authoring -> Running Option A/B -> Parsing result -> CI integration) with an explicit non-zero-exit verification gate and the if: always() checkpoint that preserves Data Docs on failure.

No local bundle files exist; the four reference links resolve to real docs.greatexpectations.io URLs and the body is organized into clearly signaled sections, which earns the top score per the simple-skill note.

Lists four concrete actions ('Authors...ExpectationSuites, builds ValidationDefinitions and Checkpoints, runs validation against tabular batches, and parses the JSON result for CI gating'), matching the score-3 anchor for multiple specific concrete actions.

Explicitly answers what it does and when to use it via an explicit 'Use when the user works with Great Expectations on Pandas, SQL, or Spark data' clause.

Includes natural terms a user would say — 'Great Expectations', 'Pandas, SQL, or Spark data', 'ExpectationSuites', 'Checkpoints' — giving good coverage of how users name this need.

A clearly named niche (GX Core data validation) with library-specific triggers that are unlikely to fire for unrelated skills.