Content
53%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 contains genuinely useful, terse prescriptive content (hard-ban tables, platform defaults, framework code patterns), but its reference architecture is broken — every 'REQUIRED' file is a dead path and the existing audit script is orphaned — and roughly a fifth of the tokens are slogan rhetoric, pseudo-quantification, or knowledge Claude already has. Process gates exist but are scattered rather than sequenced.
Suggestions
Ship the eight referenced files (mobile-design-thinking.md, touch-psychology.md, platform-ios.md, platform-android.md, etc.) or delete the "Mandatory Reference Reading" table — currently every "REQUIRED" reference is a dead path, while the one real bundle file (scripts/mobile_audit.py) is never mentioned in SKILL.md.
Present one ordered workflow (assess MFRI → ask clarifying questions → read platform refs → complete MOBILE CHECKPOINT → implement → release checklist) instead of scattering process gates across sections 1, 2, 3, 8, and 10.
Cut the MFRI pseudo-formula (its stated −10→+10 range is inconsistent with its 1–5 dimensional inputs), the Fitts' Law primer, and the circular "When to Use" filler to reclaim tokens.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The anti-pattern tables and code patterns are token-dense and terse, but the body pads with slogan rhetoric repeated three times ("Mobile-First · Touch-First · Platform-Respectful" appears in the title, philosophy block, and frontmatter), a Fitts' Law primer restating knowledge Claude already has ("Finger ≠ cursor", "Accuracy is low"), a broken MFRI pseudo-formula, and the circular filler "This skill is applicable to execute the workflow or actions described in the overview." Mostly efficient, but several sections could be cut. Not anchor 2 because the bulk of the document is compressed rule tables rather than padded explanation. | 3 / 5 |
Actionability | Concrete, mostly executable guidance throughout: a realistic React Native FlatList pattern (React.memo, useCallback, keyExtractor, getItemLayout), and ❌/✅ tables with specific replacements ("SecureStore / Keychain", "useCallback + memo", "Stable ID"). Not anchor 5 because the Flutter snippet is trivial ("const Text('Static')"), and the mandated execution path — "you are not allowed to design UI" without reading the reference files — points entirely at files that do not exist in the bundle. | 4 / 5 |
Workflow Clarity | A real process exists (MFRI assessment → ask-before-assuming → read references → MOBILE CHECKPOINT → release checklist) and genuine checkpoints are present (the fill-in checkpoint with "Cannot complete → go back and read", the MFRI required-action table, the ship checklist). But the sequence is never presented as ordered steps — it must be inferred from section numbering scattered across 11 thematic sections — so it matches anchor 3 ("sequence present but checkpoints missing or implicit") rather than anchor 4's "clear sequence". | 3 / 5 |
Progressive Disclosure | Scored against the actual bundle: the "Mandatory Reference Reading" table signals eight required .md files (mobile-design-thinking.md, touch-psychology.md, mobile-performance.md, mobile-backend.md, mobile-testing.md, mobile-debugging.md, platform-ios.md, platform-android.md), none of which exist in the bundle, while the one asset that does exist (scripts/mobile_audit.py, ~36KB) is never referenced from SKILL.md at all. The disclosure structure is broken at its core — dead required references plus an orphaned real file — matching anchor 2 rather than anchor 3, whose references at least exist. | 2 / 5 |
Total | 12 / 20 Passed |