Skip to content

Feature: feat/panel to dev - #479

Open
github-actions[bot] wants to merge 1 commit into
devfrom
feat/panel
Open

github-actions[bot] wants to merge 1 commit into
devfrom
feat/panel

Conversation

@github-actions

@github-actions github-actions Bot commented Oct 7, 2026 •

Copy link
Copy Markdown
Contributor

Automated PR tracking changes from feat/panel into dev.


Note

Low Risk
Additive documentation and a new tested library package; legacy judging in packages/api is untouched until a later cutover.

Overview
Introduces Panel, a standalone hackathon judging product under judging/, with a long-form PLAN.md, MIT license, README, and ADRs for a standalone boundary and a pure @panel/core.

Phase 0 deliverable: new workspace package @panel/core — zero runtime deps, ESLint no-restricted-imports blocking host/@query imports in library code — implementing portable judging logic: phase gating, rubric validation, visit timers, table distance (zones/coordinates), dispatch (coverage / uncertainty), ranking (z-score, Bayesian shrink, optional Bradley–Terry blend), uncertainty helpers, and shared event log / outbox / WebSocket channel types. Logic is ported from today’s club judge stack with tests asserting rank() matches the legacy pipeline at pairwiseWeight = 0.

Monorepo wiring adds judging/* to pnpm-workspace.yaml and includes judging/core in the root pnpm test script. No @panel/server, DB, web UI, or changes to existing packages/api judge routers yet.

Reviewed by Cursor Bugbot for commit aaad2da. Bugbot is set up for automated code reviews on this repo. Configure here.

Ranking and dispatch have to be extractable later, so they live under judging/ as pure functions with no host imports.

Co-authored-by: Cursor <cursoragent@cursor.com>

@cursor cursor Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

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

Fix All in Cursor

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

Want higher recall? High effort reviews run extra passes and find more bugs. A team admin can switch effort levels in the Cursor dashboard.

Reviewed by Cursor Bugbot for commit aaad2da. Configure here.

}
}
return out;
}

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pairwise ranking collapses on tied rubrics

Medium Severity

scaleStrengths maps Bradley-Terry strengths onto the rubric min/max. When those anchors are equal, every compared project is assigned that same value, so pairwise outcomes do not affect score or placement. Rubric ties are common after z-normalization of single-vote judges, which is the case pairwise blending is meant to resolve.

Fix in Cursor Fix in Web

Reviewed by Cursor Bugbot for commit aaad2da. Configure here.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant