Content
63%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 SKILL.md body is a well-written, efficient router: it defines four abstract layers, gives sharp minimum guards, and routes by user purpose with an explicit fallback loop. Its fatal flaw in this bundle is that every one of the five documents it routes to is missing, which cripples actionability and progressive disclosure despite the excellent in-file organization. No destructive or batch operations are involved, so no validation cap applies.
Suggestions
Ship the five routed documents (brownfield.md, greenfield.md, writing-docs.md, enforcing-rules.md, best-practices.md) alongside SKILL.md — none exist in the bundle, so every routing path dead-ends and the skill's core guidance is unreachable (progressive_disclosure).
Add a fallback step to the Purpose Router for when a routed document is missing or fails to answer the purpose, e.g. 'If the routed document is unavailable, apply the Common Foundation and Minimum Guards above and say so' — this both fixes navigation and gives an executable path when files are absent (actionability, workflow_clarity).
Inline a minimal placement/import-direction decision procedure (the essence of the Common Foundation guards) in SKILL.md so the most common decisions can be made without depending entirely on the external documents (actionability).
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with non-obvious guidance ('Shared is determined by code-level independence, not by generic names', 'Judge the application context the code knows, not the package it imports') and explains nothing Claude already knows, but a few passages could be trimmed — e.g. 'Real projects may split them further, but this means at least four concepts are needed. That is why they are abstract layers.' and the 60-plus-word Domain-to-Data guard sentence. | 4 / 5 |
Actionability | Concrete decision rules, a dependency diagram, and worked classification examples ('ProductCard can be Shared if it only renders injected props, while a generic-looking buildSearchParams is not Shared') are present, but the primary instruction — 'read the first matching document' — cannot be executed: none of the five routed files (brownfield.md, greenfield.md, writing-docs.md, enforcing-rules.md, best-practices.md) exist in the bundle, so the key details are missing. This matches 'some concrete guidance but incomplete; missing key details' rather than anchor 4, whose gaps are minor rather than structural. | 3 / 5 |
Workflow Clarity | The router gives a clear sequenced workflow ('Classify the user's purpose in one sentence, then read the first matching document') with an explicit checkpoint and error-recovery loop ('After reading the routed document, check whether it directly answers the user's purpose. If not, say what is missing and route to the next required source'). Not anchor 5 because there is no handling for the case where a routed document is unavailable, which this bundle demonstrates is a real failure mode. | 4 / 5 |
Progressive Disclosure | The structure is the right pattern — a lean overview with a purpose-keyed table of one-level-deep, clearly signaled links — which would score 5 if the files existed. However, all five referenced documents are absent from the bundle, so the references are dangling and the navigation chain is broken; that is worse than the 'minor organization gaps' of anchor 4 but better than anchor 2, since the in-file structure and signaling are genuinely good. | 3 / 5 |
Total | 14 / 20 Passed |