Content
78%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, actionable skill body that delegates detail to four real reference files and leads with concrete commands and code. Its only weaknesses are minor: a few placeholder code gaps and an error-recovery loop that is implicit rather than an explicit validate→fix→retry checklist.
Suggestions
Replace the placeholder `database: /* DB adapter ... */` in the canonical server snippet with a concrete one-line example (e.g. a Drizzle or Kysely call) so the core pattern is copy-paste ready.
Make the 'When Typecheck or Runtime Fails' section an explicit validate→fix→retry loop (grep common-errors.md → apply fix → re-run typecheck → only proceed when green) to lift workflow clarity to the top anchor.
Tighten the 'Critical: Do Not Trust Internal Knowledge' section to a short imperative list; the prose around 'everything known from training data is suspect' restates guidance already implied by the numbered steps.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and assumes Claude's competence — no preamble about what auth libraries are — with purposeful bash/code blocks, though the 'Critical: Do Not Trust Internal Knowledge' section and a few explanatory notes could be trimmed slightly. It sits noticeably above the midpoint but not at the leanest anchor. | 4 / 5 |
Actionability | It provides concrete, executable guidance throughout (`ask src better-auth`, `rg "^export "`, full `betterAuth({...})` and `createAuthClient` snippets, framework/plugin/DB tables), but the canonical server snippet uses placeholders like `/* DB adapter ... */` and `// see references/plugins.md`, so it is not fully copy-paste ready. | 4 / 5 |
Workflow Clarity | A clear sequence runs Prerequisites → version resolution → enumeration → server/client setup → adapter/plugin/DB selection → error triage, with a numbered 'do not trust memory' checklist and a typecheck validation step plus an error-recovery feedback loop. It falls just short of the top anchor because the validate→fix→retry loop is implicit rather than a spelled-out checklist. | 4 / 5 |
Progressive Disclosure | SKILL.md is an overview that signals four real one-level-deep references (adapters.md, plugins.md, databases.md, common-errors.md — all present in the bundle), keeps quick-reference tables inline and pushes full matrices to those files, and closes with a clear References section; bundle files cross-link to each other but stay one level from the skill. | 5 / 5 |
Total | 17 / 20 Passed |