Conciseness

The body is tables, tight code examples, and guideline bullets — no explanations of concepts Claude already knows. It falls just short of the 5-anchor because eleven full inline code examples partially duplicate content in the reference files (e.g., the complete error-handling stored procedure is detail that sql-scripting.md covers), giving minor trim potential consistent with the 4-anchor.

Actionability

Every example is complete, executable SQL with correctly qualified catalog.schema names, real signatures (h3_longlatash3(longitude, latitude, 7), returnType => ..., schemaHints => ..., CREATE CONNECTION with bearer_token secret), and inline comments. Copy-paste-ready examples cover the common case for each of the eleven feature areas, matching the 5-anchor.

Workflow Clarity

The navigation flow (Quick Reference table → example pattern → 'When to Read' column per reference file) is clear, and the final guidelines include a verification step ('Test SQL via CLI ... before deploying', with a DATABRICKS_WAREHOUSE_ID fallback). As a feature-catalog skill it has no multi-step sequence with explicit checkpoints, which fits the 4-anchor's 'clear sequence with most checkpoints; minor validation gaps' better than the 5-anchor.

Progressive Disclosure

The Quick Reference table maps every feature to its reference file, the Reference Files table adds a 'When to Read' column, and all five referenced files exist and are substantive (432–1361 lines). References are exactly one level deep — they link only to internal anchors, with no nested reference chains — matching the 5-anchor's clear overview with well-signaled one-level-deep references.

Specificity

The capability statement 'Databricks SQL (DBSQL) advanced features and SQL warehouse capabilities' names the domain, and the trigger list enumerates many concrete features, but the what is stated as generic nouns with no concrete action verbs, unlike the 5-anchor's 'Extract text and tables... fill forms'. It exceeds the 2-anchor ('names the domain but actions are minimal') thanks to the feature enumeration, but does not list specific actions.

Completeness

Both what and when are explicitly present: 'MUST be invoked when the user mentions: ...' and 'SHOULD also invoke when...' make the when maximally explicit with concrete trigger phrases. Held at 4 rather than 5 because the what ('advanced features and SQL warehouse capabilities') is generic, where the 5-anchor pairs concrete actions with its trigger clause.

Trigger Term Quality

Comprehensive natural-term coverage with synonyms and both syntax and colloquial variants: 'DBSQL', 'Databricks SQL', 'SQL warehouse', 'stored procedure', 'CALL procedure', 'materialized view', 'CREATE MATERIALIZED VIEW', 'pipe syntax', '|>', 'geospatial', 'H3', 'ST_', 'spatial SQL', 'COLLATE', 'ai_query', 'http_request', 'read_files', 'Lakehouse Federation', 'recursive CTE', 'WITH RECURSIVE', 'temp table', 'temporary view', 'pipe operator'. This matches the 5-anchor's coverage of natural terms including synonyms.

Distinctiveness Conflict Risk

Databricks-specific triggers (DBSQL, H3, ai_query, Lakehouse Federation) give it a clear niche with minimal conflict risk, but broadly generic terms like 'temp table', 'temporary view', 'recursive CTE', 'geospatial', and 'ST_' could overlap with general SQL or PostGIS questions and a sibling databricks-core skill — minor overlap risk consistent with the 4-anchor.