Content
88%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.
A strong, lean runner skill: fully executable commands with fallbacks, a clearly sequenced workflow with explicit verification and a re-run feedback loop, and correct delegation of the review bar to an external skill. The only deductions are minor repetition of the do-not-score warnings and an inline report template that slightly bloats the single-file body.
Suggestions
Collapse the repeated do-not-score warnings (intro and Step 1) into one statement to trim redundancy.
Move the verbatim code-review-report.md template block into a references/ file (e.g. references/report-template.md) and keep only the required-sections summary inline.
State the fallback ordering once (CLI pull → curl from main) and drop the duplicated 'if the CLI pull fails' framing to shorten Step 1.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and imperative with zero concept explanation ("Checks and scoring live in flows-review-checks. Do not copy them here."), but the do-not-score warnings repeat across the intro and Step 1 ("Do not score from memory" vs "Do not score from whatever is already in the app's skills/ folder"), leaving minor trimmable redundancy. It fits the efficient-with-minor-trimming 4 anchor rather than every-token-earns-its-place 5. | 4 / 5 |
Actionability | Everything is executable: exact pull commands ("npx @cognite/cli@latest apps skills pull --skill flows-review-checks"), curl fallbacks with full URLs, a concrete Glob pattern, exact output filenames, and a copy-paste report template with required labels. This matches the fully-executable 5 anchor; it is not 4 because no key detail is missing for the runner's portion of the work. | 5 / 5 |
Workflow Clarity | The sequence is explicit (pre-checks → pull checks → write artifacts → runner verify → stop condition) with a 3-item verification checklist and a real feedback loop: "Re-run this skill until Must Fix open: 0". Validation checkpoints are present, so the missing-validation cap does not apply; this fits the 5 anchor with an explicit error-recovery loop. | 5 / 5 |
Progressive Disclosure | No bundle files exist (references/, scripts/, assets/ are absent), and the body appropriately delegates all check/scoring detail to the external flows-review-checks skill at one level with clear signaling. Sections are well organized, but the ~100-line body inlines the full report markdown template verbatim — good structure with a minor organization gap, fitting the 4 anchor rather than the cleanly-split 5 anchor. | 4 / 5 |
Total | 18 / 20 Passed |