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.
The body is a tight, actionable rulebook: a concrete type-identification procedure, hard STE constraints, and correctly factored references for card/banner generation. The only gaps are a slightly generic accessibility section and the absence of an explicit end-to-end editing sequence.
Suggestions
Trim the accessible images/videos section to the project-specific rules (alt="" convention, Card img, banner text fallback) and drop generic WCAG-style guidance Claude already knows.
Add a short ordered workflow for editing a page (identify type, apply type rules, apply STE if how-to/reference, then image/card checks) so the sections have an explicit application order.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Lean and imperative nearly everywhere ("Each page has one type. Do not mix types.", "One instruction per sentence. Max 20 words in procedures, 25 in descriptions."), but the accessibility section repeats generic alt-text and video guidance Claude already knows. Anchor 4 ('efficient; minor instances of over-explanation') fits better than 5. | 4 / 5 |
Actionability | Instruction-only skill with concrete, checkable rules: a two-question type algorithm ("action or cognition? study or work?"), hard limits (max 20/25 words), ordering rules with inline examples ("If X, do Y.", "Set the variable.", alt=""). Per the rubric's code-vs-instruction note, absent code is not penalized when guidance is this actionable; not below 5 because specific examples cover the common cases. | 5 / 5 |
Workflow Clarity | A clear entry point exists ("Before you write, identify the type") plus per-topic rules, but the body is a rulebook rather than a sequenced edit/review workflow — no explicit ordering across sections or review checklist. No destructive/batch operations, so the cap at 3 does not apply; anchor 4's 'minor gaps' is the best fit. | 4 / 5 |
Progressive Disclosure | A ~40-line overview keeps core operating rules inline and defers detail via two well-signaled, verified, one-level-deep references (card-images.md, card-design.md), each annotated with scope and a read-when hint. Design values correctly live in scripts/ and data in assets/, matching anchor 5's clear overview with easy navigation. | 5 / 5 |
Total | 18 / 20 Passed |