Content
40%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 body is a well-organized catalogue stub that describes and points to the upstream skill, but it contains essentially no usable guidance: the install workflow's key step is missing and the only 'command' opens a webpage. As a SKILL.md it fails its primary job of telling the agent how to actually run the capability.
Suggestions
Replace the placeholder bash block with the actual install commands (e.g., the real clone/copy steps and the destination skills directory) so the workflow is executable end-to-end.
Inline the core actionable content — the design-spec template or UI/UX vocabulary list the skill is supposed to add to prompts — instead of deferring everything to the external GitHub repo.
Drop the duplicated description and Source sections that restate the frontmatter, keeping only the invocation and install guidance to remove redundant tokens.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is short and free of concept explanations Claude already knows, but roughly half of it duplicates the frontmatter verbatim (description repeated in 'What it does', upstream URL and category repeated in 'Source'), so it could be tightened. Not 4 because the duplicated description and Source section are unnecessary tokens; not 2 because there is no padding or over-explanation beyond that redundancy. | 3 / 5 |
Actionability | The only 'code' block is a comment plus `open https://github.com/google-labs-code/skills` with the instruction 'Inspect the upstream README for exact paths' — no install command, path, or concrete step is given. This is a high-level hint with the specific executable steps missing, matching anchor 2. Not 3 because there is no executable guidance at all; not 1 because a concrete URL and a concrete invocation instruction (skill name / trigger phrases) are provided. | 2 / 5 |
Workflow Clarity | A rough sequence exists (inspect upstream README → install the bundle → ask the agent to invoke by name or trigger), but the critical step (how to install the bundle, into which directory) is undefined and the README-reading step is deferred to an external resource. Not 3 because the key step is not just missing a checkpoint, it is unspecified; not 1 because a recognizable sequence is present. | 2 / 5 |
Progressive Disclosure | The skill is under 50 lines with clean, well-labeled sections (What it does / Source / How to use) and a single, clearly signaled external reference; no nested or buried references exist. Not 5 because the single reference is an external URL rather than an in-bundle file, so the skill defers all real content outside its own structure — a minor organization gap per the bundle-structure guideline. | 4 / 5 |
Total | 11 / 20 Passed |