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.
A well-structured, actionable six-step workflow with concrete commands, decision branches, and test-based validation. It loses points for a couple of non-executable placeholder snippets, some minor redundancy and emphasis padding, and a shallow failure-recovery loop.
Suggestions
Replace the placeholder jq filter and the empty 'Find all imports' bash block with executable equivalents (e.g., a concrete `gh api` jq expression and an actual `rg 'from ["'']package'` example).
Remove the duplicate explanation of the '(dev)' marker and trim emphasis sentences ('This is the most important step', 'Do not skim') since the surrounding instructions already convey them.
Deepen the step-5 failure loop: instruct re-running the affected package tests after each fix and stopping to report if failures persist.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly lean commands and terse decision rules, but has minor removable padding: the '(dev)' marker is explained twice (lines 34 and 72) and emphasis lines like 'This is the most important step' and 'Do not skim' add tokens without adding information. | 4 / 5 |
Actionability | Provides concrete executable pnpm/gh/npm commands for most steps, but two spots are incomplete: the release-filter jq snippet uses non-executable placeholder syntax (test("v?{from}|v?{to}|...")) and step 4 opens with an empty bash block containing only a comment before deferring vaguely to 'Use Grep'. | 4 / 5 |
Workflow Clarity | The six steps are clearly sequenced with explicit semver/dependency-type decision branches, an early checkpoint ('If already up to date, say so and stop'), and test-based validation with a feedback loop ('If tests fail, investigate and fix'). The recovery loop is shallow — no fix-and-re-run detail or checklist — so it stops short of a 5. | 4 / 5 |
Progressive Disclosure | A single well-sectioned file with no bundle files and a cohesive one-workflow scope; nothing clearly demands extraction into references. The ~175-line body and the long inline example listing (lines 40-66) keep it at 'good structure' rather than a clear overview with well-signaled references. | 4 / 5 |
Total | 16 / 20 Passed |