Content
60%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 concise, cleanly sequenced outline for a simple read-only task, but it never descends to an executable level: no gh commands, no API endpoints, no output examples. The dangling LICENSE.txt reference, the unprefixed trailing '---', and the version-number footer are small structural defects that keep an otherwise well-organized simple skill from scoring higher.
Suggestions
Add concrete executable commands for each step, e.g. resolve with 'gh pr view --json number,url' and fetch with 'gh api repos/{owner}/{repo}/pulls/{n}/comments' and '.../pulls/{n}/reviews', so the guidance is copy-paste actionable.
Handle edge cases explicitly: what to do when no active PR exists for the branch, and whether resolved/outdated comment threads are included or filtered.
Fix the trailing structure: remove the stray '---' separator, and either ship the referenced LICENSE.txt in the bundle or drop the reference and version-number footer.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean with no concept over-explanation, but the trailing provenance footer ('From the Cursor Team Kit plugin (cursor/plugins, v1.2.0)... See LICENSE.txt') is trimmable padding that includes a version number — time-sensitive information not placed in a deprecated section. This is a minor instance of content that could be trimmed, matching anchor 4 rather than the every-token-earns-its-place level of 5. | 4 / 5 |
Actionability | All four workflow steps are high-level hints — 'Resolve the active PR', 'Fetch review comments and discussion comments', 'Group feedback by severity and actionability' — with no concrete commands such as 'gh pr view --json url' or 'gh api repos/{owner}/{repo}/pulls/{n}/comments', and no examples of the grouped output format. This matches anchor 2 (high-level hints missing the specific steps to execute); it avoids a 1 because the steps do specify concrete scope and output structure. | 2 / 5 |
Workflow Clarity | The four-step sequence (resolve → fetch → group → return) is clearly ordered and unambiguous, and the read-only task means the destructive/batch validation cap does not apply. Minor gaps remain — no handling for 'no active PR exists' or resolved-vs-unresolved comment threads — keeping it below the explicit-checkpoint level of 5. | 4 / 5 |
Progressive Disclosure | The skill is under 50 lines with well-organized sections (Trigger, Workflow, Output), which per the simple-skill exception could merit a 5. However, scoring against the actual bundle, the body references 'LICENSE.txt' which does not exist, and the stray trailing '---' separator makes the provenance footer parse as a second frontmatter block rather than a labeled section — minor organization gaps matching anchor 4. | 4 / 5 |
Total | 14 / 20 Passed |