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 tight, well-structured review procedure with concrete commands, explicit per-step completion checkpoints, and a verification feedback loop. The main blemish is a broken reference to references/workflows.md and prose-only (rather than literal code) guidance.
Suggestions
Fix or remove the broken reference: the intro cites references/workflows.md → "Code review a PR", but no workflows.md exists in the bundle.
For the report step, include a one-line literal output template (e.g. `**{what}** — {file:line} — {scenario} — {fix}`) to move actionability from prose direction to copy-paste ready.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes Claude's competence — no padding, no explaining what a diff or git is, every line earning its place — matching the score-5 anchor. | 5 / 5 |
Actionability | Concrete commands (git diff, gh pr diff, ado pr diff) and a precise one-line output format make the guidance executable, but it is prose direction rather than copy-paste code blocks, so it sits at 4 rather than 5. | 4 / 5 |
Workflow Clarity | Five sequenced steps (Scope, Trace, Hunt, Verify, Report) each with an explicit "Done when" checkpoint, and step 4 is a verify/feedback loop that confirms or downgrades blocking findings — matching the score-5 anchor. | 5 / 5 |
Progressive Disclosure | The overview points one level deep to clearly signaled references (lenses.md, findings.md) that both exist, but the intro also references references/workflows.md which is not present in the bundle, so organization is good with a minor gap keeping it at 4. | 4 / 5 |
Total | 18 / 20 Passed |