Content
71%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 well-structured, actionable instruction skill: concrete artifact paths, a complete EARS pattern with example, a task template, and explicit phase-gate confirmations. Its main weaknesses are redundant trigger sections that repeat the use/skip decision four times and the absence of validation or example structure for the design phase.
Suggestions
Merge "Use this first when", "When to use this skill", and "Decision rule" into a single activation section to remove the repeated use/skip criteria and cut tokens.
Add a brief example structure or content checklist for design.md so Phase 2 guidance is as concrete as the EARS requirements example.
Include a validation checkpoint for each phase (e.g., verify every task links back to a requirement before asking for confirmation) to add a feedback loop to the phase gates.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body uses tight bullet lists and explains nothing Claude already knows, but the use/skip criteria are restated across four overlapping sections ("Use this first when", "When to use this skill", "Decision rule", plus "Common mistakes" and "Minimal checklist"). This repetition goes beyond the minor-trimming level of the efficient anchor. | 3 / 5 |
Actionability | Gives concrete executable guidance: exact artifact paths ("specs/<spec_name>/requirements.md"), a complete EARS pattern with a filled-in example, and a copy-ready task template. It falls short of fully-executable because Phase 2 and Phase 4 guidance stays high-level with no example structure for design.md content. | 4 / 5 |
Workflow Clarity | Four phases are clearly sequenced with explicit confirmation checkpoints ("Only start implementation after the user confirms the task plan", "Require confirmation between requirements, design, and task breakdown"). It misses the top anchor because there is no quality-gate validation of the artifacts themselves or feedback loop for revising a rejected spec. | 4 / 5 |
Progressive Disclosure | A single self-contained file with well-organized sections and no bundle files to split out; all content appropriately lives at this level. The sibling-skill references ("../ui-design/SKILL.md", "../postgresql-development-cloudbase/SKILL.md") are clearly signaled one level deep under "Then also read". | 5 / 5 |
Total | 16 / 20 Passed |