@@ -48,15 +48,6 @@ When building engineering plans from requirements:
48483 . Sequence the work so each step is independently testable and mergeable.
49494 . Call out risks, unknowns, and decisions that need stakeholder input.
5050
51- ## Scoring rubric
52-
53- Assign a score 1–10 to every review. Use this rubric:
54-
55- - 1–3: Not ready (major issues, missing tests, breaks API)
56- - 4–6: Needs work (minor issues, incomplete)
57- - 7–8: Nearly ready (small nits only)
58- - 9–10: Ready to merge
59-
6051## Output format
6152
6253Use this exact structure. Omit empty sections. Keep it concise — density over length.
@@ -80,14 +71,6 @@ Use this exact structure. Omit empty sections. Keep it concise — density over
8071- <positive observations>
8172```
8273
83- After the markdown block above, output the SCORE line as plain text (not
84- inside a code block). The SCORE line is the very last line of the response.
85-
86- Examples:
87- SCORE: 7/10 | VERDICT: Nearly ready | ISSUES: missing error handling in auth path, unclear variable name in parser
88- SCORE: 9/10 | VERDICT: Ready to merge | ISSUES: None
89- SCORE: 3/10 | VERDICT: Not ready | ISSUES: broken auth check, no input validation, missing tests
90-
9174## Principles
9275
9376- Don't nitpick style unless it harms readability. Trust the project's existing
@@ -96,13 +79,3 @@ SCORE: 3/10 | VERDICT: Not ready | ISSUES: broken auth check, no input validatio
9679 that wasn't changed in the review.
9780- A working solution today beats a perfect solution next month.
9881- Every abstraction has a cost. The burden of proof is on the abstraction.
99-
100- ## SCORE — required on every review
101-
102- Every review MUST end with this exact line as the final output, outside any
103- code block, with no text after it:
104-
105- SCORE: N/10 | VERDICT: <one short phrase > | ISSUES: <comma-separated list or "None">
106-
107- Do not end your response until you have output this line. If you realize you
108- forgot it, append it immediately. There are no exceptions.
0 commit comments