Three side-panel additions on the ranking dashboard's project detail
sheet:
- Project country and team name as outline badges in the header,
matching the row chips on the list view.
- Collapsible 'Description' box (closed by default) that reveals the
full project description without leaving the panel.
- Expanding a juror row now shows their per-criterion scores in
addition to the free-text feedback. Boolean criteria render with
the form's trueLabel/falseLabel (or 'Yes'/'No' fallback); numeric
criteria show the raw value next to the criterion label from the
active form's criteriaJson.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The dashboard now computes its own composite ranking score on the
client, blending (balanced-or-raw) average score with (balanced-or-raw)
advance pass rate via the existing scoreWeight / passRateWeight
sliders. Both inputs are toggled independently:
- 'Balance juror grading style (score)' — existing useBalancedRanking
- 'Balance juror approval rate (advance vote)' — new useBalancedPassRate
Both default to true and persist per-round. The pass rate is balanced
the same way scores are: each juror's personal yes-rate gives them a
Bernoulli stddev, each vote is z-normalized against that, and the
project's mean z is rescaled to the round's overall yes rate. A 'yes'
from a juror who rarely says yes counts more than a 'yes' from a
lenient juror.
List rows now show two chips — score (Bal/Raw X.XX) and pass rate
(Bal Yes% / Yes% N%) — so admins can see what's driving the order.
The threshold cutoff and live re-sort effect both use the same
composite formula.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The side panel only read Evaluation.binaryDecision, which is null when
the round's evaluation form stores the 'proceed to next round' answer
as a boolean criterion in criterionScoresJson (the current round's
shape). project.getFullDetail now resolves a unified `decision` field
per assignment using the same fallback pattern as the ranking router:
prefer the column, fall back to a type='advance' (or legacy 'move to
the next stage' boolean) criterion looked up by id in the active form.
Also: project country in the rankings list now renders whenever it's
present, not only when teamName is also set.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The ranking dashboard now refetches roundEvaluationScores every 30
seconds. When a juror submits a new evaluation, the next refetch
updates the raw and balanced averages, and the existing re-sort
effect (now also keyed on snapshot + evalScores, not just the toggle)
re-orders the list in place. Manual reorders persisted on the
snapshot still take precedence — admins who have dragged rows aren't
overruled by score updates.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Each ranking row now displays a small chip with the score that's
currently driving the rank — the juror-balanced average when the
round's useBalancedRanking toggle is on, the raw juror average when
it's off. The chip is labeled 'Bal' or 'Raw' so the source is
unambiguous. Per-juror score pills stay alongside; full Raw +
Balanced detail still lives in the side panel.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Project-level averages (Raw + Balanced in the side panel, observer
project detail score card, observer preview dialog Avg Score) now
show two decimals (e.g. 8.33 instead of 8.0/8.3) so admins can see
the actual computed value. Per-juror individual scores keep one
decimal — they're submissions, not aggregates. ScorePill gains an
optional precision prop so call sites can opt into 2-decimal display
where the value is an aggregate.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds the 'How scores are calculated' affordance to:
- the admin ranking dashboard side panel (next to the Avg Score card)
- the observer full project detail page (in the score card)
- the observer reports preview dialog (next to Evaluation Summary)
so all three audiences can open the same explainer dialog.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Reusable component used by admin and observer surfaces. Covers the
algorithm, a five-step plain-language walkthrough, a worked example
with three jurors of different grading styles, edge cases, and why
both Raw and Balanced are always shown.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Extends the ranking router's roundEvaluationScores response with
per-juror grading stats (mean, stddev, count) plus the round's overall
mean/stddev. The side-sheet juror rows render 'typical X.XX →
contributes Y.YY' next to each Score badge whenever balanced is on,
making the z-rescaling visible per individual rather than only as a
project-level number.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Removes the per-row '⇢ X.X' annotation from the ranking list — the
list view stays clean. The side panel's stats area gains a combined
Avg Score card that shows Raw and Balanced side-by-side, with the
active one (per the round's toggle) bolded and tagged 'used for
ranking'. Pass Rate and Evaluators move below into a 2-col grid.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The two existing sort sites (initial init + threshold cutoff) now read
from the local toggle. A second effect re-sorts the list when the
toggle flips, but only when no manual reorder is pinned to the
snapshot — persisted manual reorders always win, matching prior
behavior.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
A Switch at the top of the project side panel writes
useBalancedRanking onto Round.configJson via the existing round.update
mutation. The flip is shared across all viewers because the value
lives in the round's persisted config; hydration runs on every
roundData refetch so the UI converges quickly when another admin
flips it.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Defaults to true so existing rounds preserve current behavior; toggled
per-round from the ranking dashboard side panel.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The observer full project page used to call getProjectDetail without
a round, getting cross-round contaminated stats. It now resolves a
default — the currently OPEN round the project is in, falling back
to the most recently CLOSED one — and renders a selector chip in
the score card whenever the project participated in more than one
candidate round. Initial selection respects the ?round= query param.
A new observer procedure (getProjectRoundsForObserver) returns the
project's open or closed rounds for the picker.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Threads the active roundId through ProjectPreviewDialog and its two
callers (filtering tabs, expandable juror table). When a round is in
scope, the preview's stats card now matches the per-juror list and
the page-level round selector. The roundId prop is optional so the
component still works in any future caller that lacks round context.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The admin dashboard fetches its side-sheet detail from project.getFullDetail
(not analytics.getProjectDetail as the audit assumed), and that procedure
had the same cross-round contamination bug. Add an optional roundId to
its input, filter the SUBMITTED-evaluations query when provided, and pass
roundId from the dashboard's useQuery so the Avg Score / Pass Rate /
Evaluators card now matches the per-juror list rendered below it.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Previously the edition-level branch of analytics.getProjectRankings
(programId mode) pooled every juror's evaluations across every round
into a single z-normalization context. A juror's mean and stddev are
not stable across round types — quick intake screening produces a
very different grading profile than a deep evaluation round, and
mixing them yields a meaningless personal calibration.
The rollup now groups points by roundId, computes one balance context
per round, and aggregates per-project as the unweighted mean of the
per-round balanced averages. roundId mode is unchanged.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The procedure pulled every SUBMITTED evaluation for a project across
every round it ever participated in, then computed Avg Score / Pass
Rate / Evaluators from that pool. Meanwhile the per-juror list rendered
in the admin sheet filters to the current round, producing a card that
disagreed with the visible list. With roundId in the input, callers
opt into round-scoped stats; omitting it preserves the old aggregate
behavior for any caller that hasn't been updated yet.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
15 TDD-style tasks covering the round-scoping bug fixes for
getProjectDetail and getProjectRankings, the per-round toggle, the
side-panel deeper display, the shared score explainer dialog, and the
decimal display audit.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Captures the per-round toggle, side-panel deeper display, "How scores
are calculated" explainer dialog, and the cross-round contamination
fixes for getProjectDetail and getProjectRankings.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Initial ranking order, advancement cutoff line, and per-row "advancing"
highlight now all use the juror-balanced (z-score corrected) average,
falling back to the raw avgGlobalScore when no balanced score exists.
The init effect waits for evalScores so the sort has the data it needs.
Admin drag-reorders still take precedence — saved reorders override the
default sort exactly as before.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Top-level selector in the URL (?round=...) drives every single-round tab
(Overview, Analytics, Juror Consistency, Diversity) and narrows the Pipeline
tab to the selected program. Cross-Round keeps its own multi-select because
it compares multiple rounds by design.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds a shared juror-balancing utility (z-score normalization per juror,
rescaled back onto the raw 1-10 scale) and wires it into:
- Admin reports page: Top-10 project table now shows "Raw Avg" and
"Balanced" columns side by side, and the summary stats row shows a
balanced-average tile. Sort defaults to balanced so harsh and lenient
graders no longer skew the ranking.
- Ranking dashboard: each project row shows a green/amber balanced-score
chip next to the raw average when the two differ by ≥0.05, making it
obvious when juror calibration moved a project's effective ranking.
Also adds AI Juror Calibration Advisory — a mutation that takes
anonymized per-juror stats, calls OpenAI, and produces a plain-language
explanation of the cohort's grading patterns plus per-juror severity
(normal / notable / outlier) with a one-sentence narrative. The advisory
describes the statistical balance that already runs; it does not
introduce a new weighting layer. Rendered as a panel in the Juror
Consistency tab when a specific round is selected.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
getProjectRankings and getAllProjects in the analytics router were
loading every assignment for each project regardless of the selected
round, so per-round rankings in the admin reports were counting
evaluations from prior rounds and mixing their scores into averages.
A Round 2 view with 3 reviews per project was showing 5 evaluations
and a skewed average that included Round 1 data.
Filter the nested assignments include by the round (or program) the
request is scoped to, matching the pattern already used by the ranking
service and by getStatusBreakdown.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The applicant timeline sidebar was treating any previous round with
status ROUND_CLOSED as "advanced past", which caused the Grand Finale
to show "You are here" as soon as the semi-final round closed — before
admins declared which teams actually advanced.
Require explicit PASSED/COMPLETED project state on every prior round
before marking the next one as current. A closed-but-unresolved round
now correctly keeps the applicant anchored to it.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The ranking dashboard showed 0/X Yes for every project in rounds using
the 'advance' criterion type because the matcher only looked for
type === 'boolean' with a "move to the next stage" label.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Seven scenarios covering the new admin procedures:
- non-admin users are rejected on adminStart/adminSubmitOnBehalf
- admin can list a juror's assignments with COI flag surfaced
- admin can complete the full draft→autosave→submit cycle with the
voting window already closed, confirming bypass works
- COI-declared assignments still block admin submission
- feedback minimum length is enforced on admin submit
- adminAutosave refuses to overwrite a SUBMITTED evaluation
- audit log captures admin id, juror id, and bypassedWindow flag
Also swaps a lucide icon alias (FileEdit → Pencil) in the juror
assignments page to avoid the deprecated alias form.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When a juror cannot connect during an evaluation round, an admin
can now submit evaluations for them.
Router — new admin procedures:
- adminStart / adminAutosave: create and save drafts for any juror.
- adminSubmitOnBehalf: submit bypassing ROUND_ACTIVE and voting-window
checks. COI block and feedback/criterion validation still enforced.
Audit log records both admin and juror IDs plus bypassedWindow flag.
- getJurorAssignmentsForRound: list a juror's assignments + eval state.
UI — two new admin pages under /admin/rounds/[roundId]/jurors/[userId]/:
- evaluate: list of pending + completed assignments, COI flagged.
- evaluate/[projectId]: evaluation form reusing the juror's scoring UI,
with an "acting on behalf" banner and confirmation dialog before
submit. Back button returns to the assignments list.
Entry point: FilePen icon on each juror row in JuryProgressTable.
Refactor: extracted the scoring form JSX into shared
EvaluationFormFields component so the juror page and the admin proxy
page render identical inputs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Queue payload exposes the field as `category` (aliased from
`competitionCategory`), not `competitionCategory` itself, so the
previous lookup was always undefined and every project rendered as
"No category". Now shows "Startup" / "Business Concept" correctly.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Jury dashboard now shows "Submitted" badge (green) with "Edit
Rankings" button when juror has already voted, instead of always
showing "Vote Now" — prevents confusion about whether vote saved
- Award-master page now shows project logos next to project names
- Backend getMyAwardDetailEnhanced now returns logo URLs
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Replace custom download-only file list with full FileViewer component
that supports inline preview (PDF, video, images, Office docs),
open in new tab, and download
- Add project logos next to project names in award voting cards
- Backend now returns full file metadata (mimeType, size, pageCount,
language detection) and project logo URLs
- Award jurors can access files for eligible projects (access control
already added in prior commit)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Jury dashboard now shows active award voting banners with project
count, deadline countdown, and direct link to vote
- Award voting page shows full project details: description, team
members, tags, and downloadable files in expandable cards
- Award jurors can now download files for eligible projects (added
awardJuror access check to file.getDownloadUrl)
- Backend query enhanced to include files and team members
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The juror dropdown was always empty because the page requested
perPage: 200 but the user.list API caps at 100 (Zod validation).
Fixed to perPage: 100 with role filter for JURY_MEMBER/AWARD_MASTER.
Added "Import from Jury Group" section to the awards juror tab:
select a jury group, see members with checkboxes (already-assigned
shown as disabled), bulk-add selected members via bulkAddJurors.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Email lookups used findUnique (case-sensitive on PostgreSQL) but user
input was lowercased, causing login failures for users with mixed-case
emails stored in the DB (e.g. Laurent_Faure@dietsmann.com). Also
normalized 7 affected emails to lowercase on the production DB.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Files uploaded by admins with roundId but no requirementId were not
counted in the finalization page Docs column.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The admin upload flow accepted roundId but never wrote it to the
ProjectFile record, causing all admin-uploaded files to appear under
"General". Fixed the create call, the listByProject filter, and the
listByProjectForStage grouping to also use the direct roundId field.
Jury assignments on the project detail page are now grouped by round
with per-round completion counts instead of a flat list.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The category migration used DROP CONSTRAINT but the index was created
with CREATE UNIQUE INDEX, so it was never actually removed. This
prevented saving business concept criteria when startup criteria
already existed at the same version number.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Project names now link to their detail page on all finalization tabs.
Submission/intake rounds show a docs submitted/required column.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Adds an "Additional Roles" checkbox section below the primary Role
dropdown. Admins can now grant a user multiple dashboard views (e.g.,
Observer + Jury Member) without changing their primary role. The roles
array is saved alongside the primary role and used by the role switcher.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Adds bulkInviteMembers procedure to juryGroup router and integrates
BulkInviteForm into the jury group members tab. Also removes the
JURY_MEMBER-only filter from the user search — any user can now be
added to a jury group.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Previously showed "Select a jury group below" but the selector was
buried at the bottom (or not rendered at all when no jury was assigned).
Now shows the actual dropdown + "New Jury" button right where the
empty state message was.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
On startup, check for any failed migration in _prisma_migrations and
automatically mark it as rolled-back before running migrate deploy.
This prevents a partially-applied migration from permanently blocking
all future deployments. Also reduce max retries from 30 to 6.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The migration partially applied (column added, then failed on DROP
CONSTRAINT). Make every statement idempotent with IF EXISTS / IF NOT
EXISTS so it can safely re-run after resolve --rolled-back.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The migration failed on deploy because the constraint
EvaluationForm_roundId_version_key did not exist in the target DB.
Using DROP CONSTRAINT IF EXISTS makes this safe for databases where
the constraint was already removed.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
