feat(ui): confidence-calibration curve card on the analytics dashboard (#2192)

[andriypolanski]

Summary

Adds a read-only Confidence calibration card to /app/analytics visualizing predicted merge-confidence bands vs realized kept-rate from computeCalibration, plus the recommended confidence floor. Extends Calibration with per-bin aggregates and surfaces the payload on the operator dashboard.

Closes #2192, Part of #1967.

Changes

Area	Change
src/review/ops.ts	CalibrationBin + buildCalibrationBins(); bins on Calibration
src/services/operator-dashboard.ts	Surface calibration via computeCalibration
apps/gittensory-ui/src/components/site/app-panels/calibration-card-model.ts	UI types + trend/status helpers
apps/gittensory-ui/src/components/site/app-panels/calibration-card.tsx	TrendChart curve, per-bin rows, recommended-floor Stat + BoundaryBadge
apps/gittensory-ui/src/routes/app.analytics.tsx	Wire card after cycle time
Tests	calibration-card.test.tsx, ops.test.ts, operator-dashboard.test.ts

UI Evidence

Required before opening the PR — upload screenshots via GitHub drag-and-drop (do not commit images):

Page / Feature	Before	After
/app/analytics

Test plan

• CalibrationCard — empty bins
• CalibrationCard — single populated bin
• CalibrationCard — full multi-bin curve + recommended floor
• buildCalibrationBins — empty / single / full-curve cases
• computeCalibration — bins folded from merged + reverted samples
• Operator dashboard payload includes empty calibration.bins
• npm run ui:test — calibration-card.test.tsx (8/8)
• npm run ui:typecheck
• npm run test:unit -- test/unit/ops.test.ts test/unit/operator-dashboard.test.ts (44/44)
• Manual: seed + refresh /app/analytics
• Before/after screenshots uploaded to PR (required — issue is visual)

Notes

• Issue is labeled visual (owner-led). Confirm eligibility before opening a contributor PR.
• Recommended floor Stat includes BoundaryBadge boundary="private-api" — operator-only tuning signal.
• Bins use fixed 10-point bands from 50% through 100% confidence.

[superagent-security]

Superagent didn't find any vulnerabilities or security issues in this PR.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.23%. Comparing base (2dbfc46) to head (da33d6e).
⚠️ Report is 13 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #4818   +/-   ##
=======================================
  Coverage   94.23%   94.23%           
=======================================
  Files         441      441           
  Lines       38793    38809   +16     
  Branches    14129    14131    +2     
=======================================
+ Hits        36556    36572   +16     
  Misses       1577     1577           
  Partials      660      660           

Flag	Coverage Δ
shard-1	55.33% <70.58%> (?)
shard-2	41.28% <100.00%> (?)
shard-3	38.38% <5.88%> (?)
shard-4	48.55% <70.58%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
src/review/ops.ts	99.24% <100.00%> (+0.10%)	⬆️
src/services/operator-dashboard.ts	100.00% <ø> (ø)

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[loopover-orb]

Warning

🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨🟨

⏸️ Gittensory review result - manual review recommended

_{Review updated: 2026-07-11 02:26:11 UTC}

8 files · 1 AI reviewer · no blockers · readiness 100/100 · CI green · clean

⏸️ Suggested Action - Manual Review

• Touches a guarded path — held for manual review: This PR changes guardrail-protected path(s): src/services/operator-dashboard.ts (matched src/services/**).

Review summary
This PR adds a read-only confidence-calibration card to the analytics dashboard, extending `Calibration`/`computeCalibration` with `buildCalibrationBins()` (5 fixed confidence-band buckets folded from the same kept/reverted samples already computed for the existing avg/recommendedFloor logic) and threading `calibration` through `operator-dashboard.ts` and the new `CalibrationCard`/`calibration-card-model.ts` into `app.analytics.tsx`. The binning logic is correct (half-open bins except the last, which is inclusive of 1.0) and is derived from the same `isKept` boolean used for the existing kept/rev arrays, so there's no risk of the two calculations diverging. Tests cover empty/single/full-curve cases for both the pure bin-builder and the component, and wiring is complete end-to-end (ops.ts → operator-dashboard.ts → route → card) with no schema/migration change needed since this is a computed, non-persisted field.

Nits — 6 non-blocking

• src/review/ops.ts:96-119 `buildCalibrationBins` silently drops any sample with `confidence < 0.5` (it never lands in a bin), so `bins` sample-size totals don't reconcile with `mergedCount`/`keptAvgConfidence` — worth a one-line comment noting the curve intentionally only covers the >=50% band, or the card copy should say so.
• apps/gittensory-ui/.../calibration-card-model.ts:35-37 `calibrationTrendValues` maps a `null` keptRate (empty bin) to `0`, which is visually indistinguishable on the `TrendChart` from a real 0% kept-rate bin with samples — consider a distinct sentinel or gap in the series.
• test/unit/operator-dashboard.test.ts has a stray blank line with trailing whitespace right after the new `calibration` assertion block — minor cleanup.
• apps/gittensory-ui/.../calibration-card.tsx:74 the `96` chart height and src/review/ops.ts:114 the `.toFixed(3)` precision are unexplained magic numbers; naming them (or reusing an existing constant if one exists for other cards) would help future readers.
• `CalibrationCard` is a ~100-line single component (per-bin rows, curve, stats all inline) — consider extracting a `CalibrationBinRow` subcomponent for consistency with how other analytics cards in this dir are typically decomposed.
• Touches a guarded path — held for manual review — A maintainer must review and merge this change.

Signal	Result	Evidence
Code review	✅ No blockers	1 reviewer
Linked issue	✅ Linked	#2192
Related work	✅ No active overlap found	No same-issue or scoped active PR overlap found.
Change scope	✅ 20/20	Low review scope from cached public metadata (1 linked issue).
Validation posture	✅ 25/25	PR body includes validation/test evidence.
Contributor workload	✅ 10/10	Author activity: 143 registered-repo PR(s), 96 merged, 25 issue(s).
Contributor context	✅ Confirmed Gittensor contributor	andriypolanski; Gittensor profile; 143 PR(s), 25 issue(s).
Gate result	⚠️ Not blocking	Advisory; not blocking this PR.

Review context

• Author: andriypolanski
• Role context: outside_contributor
• Public audience mode: oss maintainer
• Lane context: Repository is configured for direct PR review.
• Public profile languages: Python, TypeScript, JavaScript, Rust, Cuda, Kotlin, MDX, Scala
• Official Gittensor activity: 143 PR(s), 25 issue(s).
• PR-specific overlap: none found.

Contributor next steps

• Keep the PR focused and include validation evidence before maintainer review.

Signal definitions

• Related work = same linked issue, overlapping active PRs, or title/path similarity.
• Change scope = cached public metadata such as size labels, draft state, and review-burden hints.
• Validation posture = whether the PR provides enough public validation/test evidence for maintainer review.
• Contributor workload = public contributor activity and cleanup pressure, not a repo-wide quality failure.
• Contributor context = public GitHub/Gittensor identity context; non-Gittensor status is not a blocker.

Visual preview

Route	Viewport	Before (production)	After (this PR's preview)	Diff
/app/analytics	desktop			—
/app/analytics	mobile			—

Click any thumbnail to open the full-size screenshot. Before = production · After = this PR's preview deploy.

_{🟩 Safe / merged · 🟦 Advisory · 🟨 Held for review · 🟥 Blocked / closed}

---

💰 Earn for open-source contributions like this. Gittensor lets GitHub contributors earn for the work they already do — register to start earning →.

Checked by Gittensory, a quiet PR intelligence layer for OSS maintainers.

• Re-run Gittensory review