Content
68%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.
A well-structured, code-heavy skill body with concrete executable examples and sensible use of a one-level-deep reference file. Its main gaps are the absence of explicit validation/verification steps in the design workflow (capping workflow clarity) and code examples that depend on undefined project scaffolding.
Suggestions
Add explicit validation checkpoints to the Schema Design Process, e.g., 'Step 4: Verify — apply the migration in a transaction, test RLS policies with a non-superuser role, and roll back on failure.'
Make the seed script self-contained or note its prerequisites (schema files, client setup) so the example is actionable without guesswork.
Move the RLS policy details and seed/ERD examples into reference files (e.g., references/rls-policies.md, references/seed-and-erd.md), keeping SKILL.md as a lean overview.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is efficient: terse headers, bullet lists, and code with almost no re-teaching of concepts Claude already knows. Minor trimmable parts exist — the 60-line seed script and the 'When to Use' section that duplicates the frontmatter description — keeping it just short of the lean-anchor 5. | 4 / 5 |
Actionability | Mostly executable guidance: complete RLS SQL policies, a runnable TypeScript seed script, a Mermaid ERD block, and concrete commands ('npx prisma-erd-generator'). Not a 5 because the seed example imports undefined project files ('./client', '../src/lib/auth', './schema') so it is not standalone copy-paste ready. | 4 / 5 |
Workflow Clarity | The 'Schema Design Process' presents a clear three-step sequence (requirements to entities, relationships, cross-cutting concerns), but the workflow has no validation checkpoints — migration and RLS verification appear only as pitfall bullets ('RLS not tested — always test RLS with a non-superuser role') rather than as explicit steps. Per the rubric's database-operation rule, missing feedback loops in this context caps the score at 3. | 3 / 5 |
Progressive Disclosure | Good structure: the bulk full-schema example is pushed to references/full-schema-examples.md (a real file, one level deep, no nested references, clearly signaled with an arrow pointer), and sections are well organized. Not a 5 because sizable RLS, seed, and ERD examples remain inline in SKILL.md and could be split into reference files for a leaner overview. | 4 / 5 |
Total | 15 / 20 Passed |