Content
58%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The skill's core tool workflow is excellent — well-sequenced, command-driven, and backed by an explicit verification loop — but it sits inside a bloated body of textbook SQL and database-architecture material Claude already knows. Worse, every script, asset, and reference file the workflow depends on is missing from the bundle, so the workflow cannot actually execute as shipped.
Suggestions
Move the Query Generation Patterns, Performance Optimization, Multi-Database Decision Matrix, and Sharding & Replication sections into a reference file (e.g., references/database-design-reference.md), keeping only a one-line pointer per topic in SKILL.md — this would cut the body by more than half and fix the broken reference.
Ship the scripts and assets the workflow depends on (schema_analyzer.py, index_optimizer.py, migration_generator.py, assets/sample_schema.sql, assets/sample_schema.json, assets/sample_query_patterns.json), or rewrite the workflow around guidance Claude can execute without them.
Delete or drastically condense the generic Best Practices section ("Use meaningful names", "Encrypt sensitive data", "Validate inputs") — these are database-commonsense items Claude already applies without instruction.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Over half the body restates textbook knowledge Claude already has — JOIN/CTE/window-function tutorials, GROUPING SETS, generic best practices ("Use meaningful names", "Encrypt sensitive data"), and the PostgreSQL/MySQL/SQLite comparison and sharding/replication matrices. These are several large padded sections rather than a few trimmable sentences. | 2 / 5 |
Actionability | The tool workflow provides concrete, flag-level commands (e.g., `python3 schema_analyzer.py --input schema.sql --generate-erd --output-format json -o analysis.json`) and runnable SQL examples, but the commands reference scripts and sample assets that are not present in the bundle, so they are not fully copy-paste executable as shipped — concrete commands with a gap. | 4 / 5 |
Workflow Clarity | The tool workflow is a clear four-step sequence (analyze → optimize → migrate → verify), each step with a full command, and includes an explicit validation feedback loop: "Re-run step 1 on the *target* schema and assert the issues found in the first pass are gone; run `migration_generator.py --validate-only` before handing over the migration" — exactly the feedback-loop rigor the rubric demands for database operations. | 5 / 5 |
Progressive Disclosure | The body inlines roughly 200 lines of material that belongs in separate reference files (query generation patterns, sharding and replication, database comparison matrices), its only reference pointer — "See references/database-design-reference.md" — targets a file that does not exist, and every referenced script and asset (schema_analyzer.py, assets/sample_query_patterns.json, etc.) is missing from the bundle entirely. | 2 / 5 |
Total | 13 / 20 Passed |