fix(v1): prefer final judge choices over draft boxes

[xeophon]

Overview

Prefer explicit final judge verdicts over boxed answers, so a draft boxed answer cannot mask a later final decision.

Alternative implementation to #2256.

Details

• Read direct choices from Final Judgment, Final Answer, and Final Verdict lines before applying any boxed-answer fallback.
• When more than one explicit final verdict appears, use the last one.
• If no explicit final verdict is present, keep the existing boxed-answer and general marker behavior.
• Keep the change contained to parse_judge_choice without adding another parsing helper.

Note

Prefer final judge markers over boxed answers in parse_judge_choice

• Adds an early scan in parse_judge_choice in score.py for lines starting with FINAL plus JUDGMENT|ANSWER|VERDICT followed immediately by a valid choice token, returning the last such match's canonical choice.
• Falls back to existing boxed-answer extraction and legacy verdict_re heuristics only when no explicit final-line choice is found; text_upper is recomputed after boxed extraction for the legacy path.
• Updates tests in test_scoring.py to cover precedence, multiple-marker, and non-final-label scenarios.
• Behavioral Change: judge outputs with both a boxed answer and a later explicit Final Judgment: line now resolve to the explicit line instead of the box; outputs relying solely on boxed answers are unaffected.

^{Macroscope summarized 59239af.}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-08-29T08:51:33.407332Z	59239af	New commits

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 6def05051e

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — This small parser fix changes production judge scoring and verdict classification. Unresolved review concerns identify concrete ambiguity around prose tokens and ordering between boxed and unboxed final verdicts, so the resulting behavior needs human verification.

^{You can add or adjust custom eligibility rules. Learn more.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 491c4789bd

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 011b6b7183

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 011b6b7. Configure here.}