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.
An exemplary overview-style SKILL.md: lean, well-structured, with well-signaled one-level-deep references and a clear sequenced workflow ending in validation. Its weakness is actionability — the rules are directional but omit the concrete specifics (token names, class examples, import paths) that would make them executable without opening every reference file.
Suggestions
Include one short concrete example in the body, such as a semantic token name and its palette counterpart, or a one-line CSS import of the shared contract's entry point, so core rules are actionable before opening reference files.
Name the entry points of the 'complete CSS contract' (e.g., the actual import path or exported files) rather than only referencing 'components.css' abstractly.
Add a brief error-recovery step after workflow step 9, e.g., what to do when tests or screenshots reveal a token or layout violation.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~45-line body is entirely directive with no concept explanations or padding ('Keep one primary action per decision area', 'Reserve round geometry for avatars, status dots...'). Every token earns its place, matching the lean-and-efficient top anchor. | 5 / 5 |
Actionability | As an instruction-only skill, absence of code is not penalized per se, but the guidance lacks concrete specifics: no token names, no sample class composition, and no path or entry-point name for the 'complete CSS contract' (only 'components.css' is named). This matches 'some concrete guidance but incomplete; missing key details' rather than level 4, which expects mostly executable guidance with concrete code or commands. | 3 / 5 |
Workflow Clarity | Nine clearly sequenced steps with an explicit validation checkpoint ('Validate the affected routes with existing tests and real browser screenshots'). It is not 5 because there is no error-recovery feedback loop telling Claude what to do when validation fails; it is above 3 because the sequence and a real checkpoint are present. | 4 / 5 |
Progressive Disclosure | The body is a lean overview pointing to five one-level-deep reference files (tokens.md, consumer-adapters.md, application-surfaces.md, terminal-ui.md, embedded-surfaces.md), all verified to exist, each clearly signaled with a read-condition ('before choosing colors', 'when designing or auditing a terminal interface'). This matches the top anchor: clear overview, well-signaled references, easy navigation. | 5 / 5 |
Total | 17 / 20 Passed |