Content
28%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 an auto-generated skeleton: placeholder script invocations, boilerplate feature bullets, and generic best-practice lists that teach Claude nothing new. The reference and script files exist and are appropriately linked one level deep, but the references themselves are placeholder templates, and the body's bulk is padding.
Suggestions
Cut the 'Tech Stack', 'Best Practices Summary', and 'Common Commands' sections — they restate common knowledge and generic npm/docker commands unrelated to the skill's actual purpose.
Document each script's real CLI (actual flags, required arguments, example invocation and expected output) instead of '[options]' / '[arguments]' placeholders.
List each reference file exactly once and state what problem each solves, rather than repeating the same three paths across three sections; replace the placeholder content inside the reference files with real material.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~215-line body is heavily padded with knowledge Claude already has ('Write comprehensive tests', 'Validate all inputs', 'Measure before optimizing', 'Keep it simple') and vacuous feature bullets ('Automated scaffolding', 'Deep analysis', 'Expert-level automation'). This matches 'Severely verbose; heavily padded' — the Tech Stack and Best Practices Summary sections contribute nothing the model doesn't know. | 1 / 5 |
Actionability | Real script paths are cited (scripts/project_architect.py etc. all exist), which is concrete, but invocations use undefined placeholders — 'python scripts/architecture_diagram_generator.py [options]', 'python scripts/dependency_analyzer.py [arguments] [options]' — with no documented flags, expected inputs, or example output. This lands on 'Some concrete guidance but incomplete; missing key details'. | 3 / 5 |
Workflow Clarity | The 'Development Workflow' is a rough three-step sequence where step 3 ('Implement Best Practices') just points at reference files, and 'Run Quality Checks' says only 'Review recommendations / Apply fixes'. It matches 'Rough sequence present but many gaps; validation absent' — no checkpoints on what a passing check looks like. | 2 / 5 |
Progressive Disclosure | Bundle structure is real: three references/ files and three scripts/ files, all one level deep and all cited paths exist. However, the same three reference files are listed redundantly in three sections ('Reference Documentation', 'Development Workflow' step 3, and 'Resources'), and generic list content that belongs in (or nowhere near) the references is inlined in SKILL.md — 'Some structure but could be better organized'. | 3 / 5 |
Total | 9 / 20 Passed |