README: first -llm-judge results

[davidbzyk]

The six -llm-judge tasks, one trial each on 2026-10-13 (team and judge both Codex on gpt-6-luna, medium): per-scale 1-5 scores and the weighted reward, mean 0.41 at $0.24 per task. Per-trial CSV in research/results/llm-judge-2026-10-13.csv. Notes that these are new team runs, not the reviews the recall grader scored.

[coderabbitai]

Warning

Review limit reached

Your organization has reached its usage spending cap. Adjust your spending cap in the billing tab.

Next included review available in 25 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used the included review currently available. Your 88 included PR review attempts over the past 7 days set your current allowance at 1 review per hour.

Learn how review limits work.

Review configuration:

⚙️ Run configuration

• Configuration used: Organization UI
• Review profile: CHILL
• Plan: Essentials
• Run ID: a3f948a6-5b8d-4dd0-b966-e93273f4fd5e

📥 Commits

Reviewing files that changed from the base of the PR and between f5693db and d830541.

⛔ Files ignored due to path filters (1)

• research/results/llm-judge-2026-10-13.csv is excluded by !**/*.csv

📒 Files selected for processing (1)

• README.md

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

_{Comment @coderabbitai help to get the list of available commands.}

[Copilot]

🟢 Approval recommended

The documented figures are consistent with the CSV data and configured scoring weights.

0 open findings

What changed in this PR

Publishes the first six -llm-judge evaluation results and summarizes their methodology and outcomes.

Changes:

• Adds per-task scores, token usage, and costs.
• Documents aggregate results and key observations.

File	Description
README.md	Summarizes results and updates project status.
research/​results/​llm-judge-2026-10-13.csv	Records per-trial evaluation data.

🧠 Review effort: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[github-actions]

Custom Code Review

No issues found on d830541.

_{codex-ci · gpt-6-luna · effort high · Codex 0.161.0}

[github-actions]

Custom Code Review

No issues found on d830541.

What the change does

• The README now reports results from six `-llm-judge` tasks run on 2026-10-13, including a reward summary, review costs, and task-level scores across five criteria.
• It adds notes describing how the judge assessed findings, how test requests affected validity scores, and how these runs relate to the recall grader results.
• The per-trial results link points to `research/results/llm-judge-2026-10-13.csv`.
• The roadmap text now says the cross-model and `-llm-judge` results are above, with more models planned.

_{codex-ci · gpt-6-luna · effort high · Codex 0.161.0}