Content
52%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 has a clear workflow and good structure, anchored by a concrete review framework and argument routing. Its main weakness is heavy persona padding: overlapping identity/role sections and generic question checklists that restate knowledge Claude already has, inflating token cost without adding guidance.
Suggestions
Collapse 'Your Identity', 'Your Role', and 'Core Responsibilities' into one short section; the three currently repeat the same persona claims (depth, breadth, mentorship mindset) at length.
Cut the generic review checklists (scalability/reliability/maintainability question lists) — a senior engineer persona already knows to ask about bottlenecks, failure modes, and threat models; keep only the domain-specific framing that differs from defaults.
Move the ~50-line System Design Review markdown template to a reference file (e.g. references/review-template.md) and point to it from the Task Execution section, trimming SKILL.md to the routing logic and communication style.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The 'Your Identity', 'Your Role', and 'Core Responsibilities' sections heavily overlap, and generic checklists (e.g. 'What are the bottlenecks?', "What's the threat model?") restate senior-engineer knowledge Claude already has. This is noticeably verbose with several padded sections rather than just incidental over-explanation. | 2 / 5 |
Actionability | The Review Framework template, Socratic dialogue examples, and $ARGUMENTS routing are concrete and usable, but the majority of the body (identity, role, communication style) is abstract persona description with no executable detail. | 3 / 5 |
Workflow Clarity | Task Execution gives a clear branch sequence per argument (--review, --career, general) with 'Ask Clarifying Questions' as an early checkpoint and a 6-step engagement flow. As a conversational skill with no destructive or batch operations, no validation cap applies. | 4 / 5 |
Progressive Disclosure | The body is well organized with clear section headers and no buried references, and no bundle files exist to misroute. The ~50-line review framework template is inline content that could live in a reference file, which is a minor placement gap. | 4 / 5 |
Total | 13 / 20 Passed |