Content
50%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 highly actionable for a review-style skill, with a concrete baseline prompt, explicit thresholds, example phrases, and a clear output/approval structure. Its weaknesses are significant redundancy — the same standards are restated in four to five different sections — and a monolithic single-file layout with no progressive disclosure despite being ~190 lines.
Suggestions
Consolidate the repeated standards: state each rule once (e.g. in "Non-Negotiable Additional Standards") and cut the restatements in "Primary Review Questions", "What to Flag Aggressively", "Preferred Remedies", and "Approval Bar" that restate the same rules in interrogative, flag, remedy, and blocker form.
Split the long example lists — "What to Flag Aggressively", "Preferred Remedies", and the "Review Tone" phrase bank — into a references/ file, keeping a short summary inline in SKILL.md.
Make the review workflow explicit and sequenced (obtain the diff/scope → run the core prompt → apply standards → prioritize per Output Expectations → decide against the approval bar), instead of leaving the process implicit across reference sections.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The same ~7 rules (1k-line limit, spaghetti branching, wrappers/casts, canonical helpers, layering, atomicity) are restated four to five times across "Non-Negotiable Additional Standards", "Primary Review Questions", "What to Flag Aggressively", "Preferred Remedies", and "Approval Bar" — e.g. "Do not let a PR push a file from under 1k lines to over 1k lines" reappears as "Did this change enlarge a file or component past a healthy size boundary?" and "A file crossing 1000 lines due to the PR" and "the PR pushes a file from below 1000 lines to above 1000 lines". This is noticeably verbose with several padded sections (anchor 2), beyond anchor 3's "some unnecessary explanation". | 2 / 5 |
Actionability | Concrete, usable guidance for an instruction-only skill: a copy-pasteable "Core Prompt" blockquote, an explicit threshold ("under 1k lines to over 1k lines"), ready-made review phrases ("this pushes the file past 1k lines. can we decompose this first?"), a numbered "Output Expectations" priority order, and an explicit approval bar. It falls short of anchor 5 only because it never specifies how to obtain the changes under review (e.g. diff commands or scope). | 4 / 5 |
Workflow Clarity | The material is organized as reference sections (standards, questions, flags, remedies, tone, output, approval) rather than a sequenced process; the actual workflow (start from the core prompt, apply standards, escalate, prioritize findings, decide approval) is implicit rather than stepwise, with no checkpoints between gathering the diff and issuing findings. This matches anchor 3's "sequence present but checkpoints missing or implicit" rather than anchor 4's explicit sequence. | 3 / 5 |
Progressive Disclosure | The body has good section headers but is a ~190-line monolith with no bundle files; content that could live in reference files (the long "Preferred Remedies" list, the "Review Tone" phrase bank, the "What to Flag Aggressively" list) is all inline. That fits anchor 3 ("some structure... content that should be separate is inline"); the >50-line simple-skill exception for scoring 5 does not apply. | 3 / 5 |
Total | 12 / 20 Passed |