Content
75%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 PR-review workflow with concrete commands, an explicit stage gate, checklists, and report templates. Its main weaknesses are unresolved external references and a missing final verification step before posting review comments.
Suggestions
Add a verify-before-publish checkpoint before the `gh api .../reviews` POST (e.g. confirm severity counts and comment paths) to close the workflow-clarity gap.
Resolve or stub the referenced bundle files (backend-quality/*, frontend-quality/*, style/*) so the checklist links lead somewhere instead of 404.
Normalize the deeply nested relative references (`../../../design/api/index.md`, `../../../code/syntax.md`) or inline the few rules that matter into the body.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly lean — commands, checklists, and templates — with project-specific rules (e.g. the i18n static-key rule) that earn their tokens; only minor framing prose could be trimmed. | 4 / 5 |
Actionability | It provides copy-paste-ready gh/git/pnpm commands and a concrete GitHub API review-posting JSON template, with minor gaps where guidance stays high-level (e.g. 'propose 3–5 test cases') and checklist targets are unresolved placeholders. | 4 / 5 |
Workflow Clarity | A clear staged sequence (Step 0 → Phase 1 → parallel Phases 2–6 → report) with an explicit stage-gate checkpoint and feedback loop; the only gap is a verify-before-publish checkpoint right before the irreversible API post. | 4 / 5 |
Progressive Disclosure | The overview points to one-level-deep, clearly signaled checklist references (security.md, react-performance.md, style/*.md), but none of the referenced files exist in the bundle and several use deeply nested relative paths, keeping it below a 5. | 4 / 5 |
Total | 16 / 20 Passed |