FEATSRV
/api/featsrv · live
build 1.0.0
7 · System health
OK
▲ 200 · p95 41ms
2–4 · Requests served (24h)
94,812
▲ 6.4% vs prior day
5 · Leakage flags open
3
▼ 2 resolved this week
1 · PIT training rows generated
2.41M
— last run 03 Aug 2026

Online feature request volume — before / after real‑time rollout

Monthly request count against the Redis‑backed online store. The dashed line marks the cut‑over from batch‑only serving (Jan 2024) to real‑time point‑in‑time serving. Same shape of chart the CLI's --health and online/* endpoints roll up into.

◂ Before — batch only Real‑time cut‑over, Jan 2024 ▸
Batch serving Real‑time serving

Recent CLI activity

Last 8 sessions written to Response‑FeatLock/
LIVE TAIL
Timestamp Service Entity Status Latency

Service call mix (24h)

Share of total requests by endpoint

Feature lineage — source to consumer

Every value served online or offline traces back through the point‑in‑time store. Flow width is proportional to feature‑value volume over the trailing 30 days. Hover a ribbon to isolate its path.

Note The point‑in‑time store is the single write path for both online and offline consumers — this is what the leakage audit (Tab 05) verifies: no feature value reaches a consumer with a feature_timestamp later than the event_timestamp it was requested against.

Leakage exposure by feature domain

Percent of served feature‑values at risk of temporal leakage, and the projected volume of training rows that would need regeneration if unresolved.

Leakage risk by portfolio

% of feature‑values at risk

Projected rows requiring regeneration

Next 24 months, by portfolio

Leakage trend

Open flags over the selected window
Feb 2026
Aug 2026

Entity risk‑score distribution

Feature‑confidence score, 300–850 band

Open leakage audit findings

/leakage-audit — filtered, sortable
Entity Feature Portfolio Δ hours Severity

Feature importance summary

Ranked by mean absolute contribution across the active underwriting and pricing models. Matches the CLI's feature-importance service output, including its rank‑bar rendering.

Top 14 features

Importance score, 0.0–1.0 scale

Point‑in‑time training set preview

First rows of the most recent PIT training pull, exactly as saved to pit_training_result.json.

— rows
claim_id policy_id event_timestamp feature_timestamp claim_severity_score prior_claims_24m label_charge_off

Batch offline features — last payload

Response body of the most recent /offline POST.

entity_id entity_type telematics_score credit_band exposure_months region

Online feature lookup

Simulates the CLI's real‑time Redis‑backed lookups. Enter an ID matching the sample formats below, or pick one of the recent entities.

Policy features

Format: POL‑0000

Claim features

Format: CLM‑00000