Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A highly actionable, well-sequenced multi-phase workflow with strong validation feedback loops. Its main weakness is conciseness: significant cross-phase repetition and a monolithic body with sections that could be split into references.
Suggestions
Deduplicate guidance repeated across phases (no-screenshots rule, commit cadence, design-reference reads) by stating each once in Core Principles and referencing it from phases.
Move the full Live Preview Status Protocol into a references file and keep a one-paragraph summary plus the JSON shape in SKILL.md to reduce body length.
Replace the Example Workflow section with a brief pointer or drop it, since it restates the eight phases already detailed above.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Content is operational rather than generic, but there is notable duplication — the "no screenshots" rule, commit guidance, and design-reference reads each repeat across phases, and the "Important Notes" and "Example Workflow" sections restate the phase content — so it could be tightened rather than earning a 3. | 2 / 3 |
Actionability | Provides copy-paste-ready bash (npm install, git init, npm run dev), exact script invocations with full paths, named MCP tools, AskUserQuestion option tables, JSON file shapes, and per-framework marker syntax — fully executable guidance. | 3 / 3 |
Workflow Clarity | Eight clearly sequenced phases with explicit GATE checkpoints and a validate→fix→re-run feedback loop (axe-core audit re-run until exit code 0, Playwright snapshot after each change), matching the score-3 anchor. | 3 / 3 |
Progressive Disclosure | Real, well-signaled one-level-deep references exist (references/design-aesthetics.md, scripts/axe-audit.js, asset templates), but the 815-line body is monolithic with the full Live Preview Status Protocol and Example Workflow inline that could be externalized, so it does not fully meet the score-3 "content appropriately split" bar. | 2 / 3 |
Total | 10 / 12 Passed |