Content
86%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 lean, highly actionable skill body: executable commands throughout, Wrong/Correct mistake pairs with sources, and well-organized one-level-deep references. The only notable improvements are inlining the command-construction order and trimming the redundant Setup/Tension sections.
Suggestions
Move the command-construction order (currently only in references/create-flag-compatibility-matrix.md) into the body as a short numbered sequence so the workflow is explicit in SKILL.md.
Drop or merge the Setup example into the first Core Patterns example, since they are nearly identical commands.
Compress the two "Tension" subsections into one-line cross-references (e.g., "For integration selection, see choose-ecosystem-integrations/SKILL.md § Common Mistakes") to save tokens.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dominated by copy-paste commands with short, CLI-specific explanations (e.g., "Router-only compatibility mode ignores template, deployment, and add-on intent"), and it never explains concepts Claude already knows. Not 5 because the Setup section largely duplicates the first Core Patterns example, and the two "Tension" subsections are commentary that could be trimmed to one-line cross-references; not 3 because there is no real padding or over-explanation. | 4 / 5 |
Actionability | Every section is a fully executable npx command, and each Common Mistake gives a Wrong/Correct pair (e.g., "npx @tanstack/cli create my-app --add-ons -y" vs. "--add-ons clerk,drizzle -y") covering the common cases: build tool choice, non-interactive runs, blank presets, router-only, and templates. This matches the copy-paste-ready anchor exactly. | 5 / 5 |
Workflow Clarity | For a single-purpose skill the action is unambiguous ("build a deterministic tanstack create command before running generation") and the Wrong/Correct pairs act as inline checkpoints, but the actual command-construction sequence (framework -> mode -> add-ons -> toolchain -> -y) lives only in references/create-flag-compatibility-matrix.md rather than in the body. Not 5 because the ordering is not stated in SKILL.md itself; not 3 because the patterns plus mistake catalog make the sequence and failure modes largely inferable and there is no validation gap for a destructive operation. | 4 / 5 |
Progressive Disclosure | The body is an overview of patterns and pitfalls, while flag-by-flag detail is appropriately split into five real, one-level-deep reference files, all verified to exist and clearly listed in the References section with descriptive titles ([Create flag compatibility matrix], [Framework adapter options], etc.). This matches the well-signaled, clearly-split anchor. | 5 / 5 |
Total | 18 / 20 Passed |