1 REGRESSION in rentveh (2 new failing tests) — 23rd consecutive plateau night otherwise unchanged — 1760 total, 856 passed, 762 failed, 142 skipped, 0 flaky.

## Run identity
- Marker: `test_reports/nightly_marker.json` — `date: "20260801"` (UTC-run-start-stamped, as usual for this pipeline)
- Source JSON: `test_reports/nightly_20260801.json`
- Started 2026-08-01T18:45:04.636Z UTC, finished 2026-08-01T23:28:12Z UTC = **2026-08-02 04:58:12 IST** (belongs to today, 2026-08-02)

## Totals
1760 total, 856 passed, 762 failed, 142 skipped, 0 flaky.

## REGRESSIONS (new vs baseline memory `project_pwa_test_coverage.md`)

`rentveh.spec.js` grew **6 → 10 failures** (+4, both android-chrome and desktop-chrome). Two new unique tests now fail:

- `submit → row lands in DB with payable amount` — vehicle picker never resolves: `page.click('.pick-row:has-text("TEST-RV-SPEC") .pi')` times out at 30000ms, then the follow-up `.exp-card` visibility check fails (element not found, 5000ms).
- `TM approval flow: team mode → approve → status approved` — cascades from the above: `.exp-card:has(.b-submitted)` never appears (30000ms timeout), and the DB-side assertion `rows[0].status` throws `TypeError: Cannot read properties of undefined` (no submitted row exists to approve).

This correlates with the still-uncommitted local edit to `rentveh/index.html` / `rentveh/sw.js` (both show modified in `git status`, flagged pending investigation on prior nights).

The 2 previously-known rentveh failures are unchanged: `SW cache version matches CACHE_VER in app` (stale `SW_VER` constant, still `v3` in spec vs shipped `v5`) and `regular user (not TM/custodian) sees no Record Payment button` / `TM/custodian: Record Payment → custodian transfer prefilled` (tied to the same edit).

## Known plateau (unchanged, 26 other spec files byte-identical in failure count to 2026-07-31→08-01)

Same "server/DB/proxy partially unreachable" family, 23rd consecutive night: `fundcustodian` 154, `sales_billing` 78, `vendors` 74, `upi_pay` 68, `customers` 50, `custodian` 40, `recce_client` 32, `recce` 32, `sales` 32, `safe_bottom` 28, `stores` 26, `counters` 18, `yagya-portfolio` 18, `contacts` 14, `dispatch` 14 (`HUB_JWT_SECRET not set`, 6th night unfixed), `finance`/`installation`/`tour_planner` 10 each, `hub`/`oc_admin`/`vrs` 8 each, `admin`/`smoke` 6 each, `client` 4, `debug_tour`/`hr` 2 each.

`tour_pg.spec.js` — 0 failures again tonight (5th flip between 0/2 in ~2 weeks; still not durable either way, watch on).

Green (0 failures): `activity`, `btl`, `drive_consolidator`, `expense`, `learn`, `printing`, `production`, `tour_pg`, `tour_pg_sharing`, `vehicle`.

## Method
Per-file counts computed directly from the JSON (`expected`/`unexpected`/`skipped` per spec × test × project), cross-checked against `stats.expected=856, stats.unexpected=762, stats.skipped=142`.

## Baseline status
**Not re-baselined** — the 2026-06-20 baseline (0 failed) in `project_pwa_test_coverage.md` remains the trusted baseline. Action items carried forward: fix `HUB_JWT_SECRET` sourcing for `dispatch.spec.js`; bump rentveh's `SW_VER` to v5; investigate/commit the pending `rentveh/index.html` edit (now confirmed to have broken 2 more tests); ops uptime fix for the dev stack still outstanding (23rd night).