Merge S8 (a1): ops — the product watches itself

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> # Conflicts: # BUILD-STATE.md # CLAUDE.md # src/snapshotScheduler.js
2026-07-11 14:38:06 -04:00
parent 841e1c9475 044941d406
commit aaafc3e0f2
11 changed files with 1081 additions and 3 deletions
@@ -84,6 +84,34 @@ work-order 1.6 closed (canonical player keys + slate join invariant),
 Phase 2 slate UX, Phase 3 mobile P0. DEPLOY GATE: Coolify by Sat 10 AM ET;
 first snapshot locks against freshly posted lines (manual internal trigger).

+## Session 8 (A1 board, 2026-07-11) — SHIPPED ✅ OPS: THE PRODUCT WATCHES ITSELF
+
+Branch off day1/a1-board tip. 2398 → 2437 tests (208 suites). Backend-only.
+- **Settlement alarm** (`snapshotScheduler`): outcome/ledger settle pass
+  THROW → ntfy high. Morning zero-settle alarm: ledger settle found
+  pre-today Postgres rows but settled none → page once per ET date
+  (signal = settleLedger's own return values — survives the Redis-TTL
+  failure mode that silently killed morning settles in S60; documented in
+  `opsWatch.zeroSettleAlarm`). Genuinely empty yesterday never alerts.
+- **Persistent-failure pager** (`opsWatch.createFailureTracker`, pure):
+  3 consecutive error / skipped-'no props' slots for a sport pages once
+  (high); any good slot resets and re-arms. Single-slot misses stay quiet.
+- **Quota** (`opsWatch.checkQuotaDaily`): odds-api ≥80% after a snapshot
+  run → one alert/day, Redis-deduped (`ops:quota_day:{provider}:{date}`).
+- **Box health** (`services/systemHealth`, pure + injectable): fs.statfs
+  + os mem → `{disk_pct, mem_pct}`; pages high at disk>85 / mem>90.
+- **Daily pulse** 13:00 UTC (9 AM EDT; `PULSE_HOUR_UTC` to move), ONE
+  notification: ledger rows yesterday (`ledgerService.countRowsForDate`,
+  'n/a' when Supabase off — never a fabricated 0), settles last 24h,
+  quota pct, disk/mem, 'desk pack: see /desk'. Dedupe: in-process date +
+  `ops:pulse:{date}`.
+- **docs/OPS-RUNBOOK.md**: Uptime Kuma monitors (vyndr.app 200, api
+  /api/health keyword 'healthy', /api/snapshot/summary), Coolify
+  deploy-failure webhook → ntfy, ntfy phone-subscription steps.
+- Live acceptance alert POSTed to ntfy.sh/vyndr-pipeline-kev2026.
+- All alert copy VOICE v1.1 — deadpan, numbers, zero exclamation points
+  (tests lint for `!`). Spec: `specs/session-8-ops-watch.md`.
+
 ## Session 60 (night2/full-board, 2026-07-11) — BUILT ✅ THE WHOLE BOARD (awaiting "merge the train")

 Branch `night2/full-board` off d10bb4c — ZERO pushes to main. 2352 →
@@ -854,6 +854,31 @@ phased plan in the Session-57 conversation / BUILD-STATE Next section).
  surface is `/welcome`. Landing capture sits below FAQ. Subscribe route is
  10/min IP-limited → keep `tests/integration/newsletterRoute.test.js` under 10
  subscribe requests (same lesson as analyze.test.js).
+## Ops Self-Watch (Session 8, A1 board — non-obvious)
+- **Pure alarm logic lives in `src/services/opsWatch.js`** (failure tracker,
+  zero-settle signal, quota daily check, pulse assembly) + `src/services/
+  systemHealth.js` (statfs/os, injectable). `snapshotScheduler` only WIRES
+  them — keep new alarm logic in opsWatch so it stays unit-testable.
+- **Zero-settle signal = the ledger settle's own return values** (Postgres
+  `game_date < today AND outcome IS NULL`), NOT `snapshot:{sport}:previous` —
+  the snapshot key expires in exactly the failure mode the alarm exists to
+  catch (the S60 SNAP_TTL bug). Scoped to `SETTLEABLE_SPORTS` (mlb);
+  NBA/WNBA rows legitimately stay pending and must not page daily. Morning
+  slot only (`morningHourUtc` = first configured hour ≥ 06 UTC).
+- **Failure pager pages ONCE per losing streak** (exactly at count 3), not
+  once per failing slot; `status:'skipped', reason:'no grades'` is NOT a
+  failure (props arrived, grader refused) and resets the counter.
+- **Daily pulse** fires at `PULSE_HOUR_UTC` (default 13 = 9 AM EDT), one
+  notification, dual dedupe (in-process date + Redis `ops:pulse:{date}`).
+  Missing data renders 'n/a' — a pulse never fabricates a zero (Data
+  Semantics Rule applies to ops copy). All alert copy: no exclamation
+  points, ever (VOICE v1.1) — tests lint for it.
+- **Single-suite jest runs of scheduler tests can hang at exit** when redis
+  is down (ioredis reconnect timer keeps the process alive) — PRE-EXISTING
+  at baseline, full-suite runs are unaffected. Don't chase it as a leak in
+  new code; inject cacheGet/cacheSet in tests to avoid creating a client.
+- Box-side ops (Uptime Kuma, Coolify deploy-failure webhook, ntfy phone
+  setup) = `docs/OPS-RUNBOOK.md`.

 ## Active Skills
 - vyndr-voice (all user-facing output)
@@ -0,0 +1,106 @@
+# VYNDR OPS RUNBOOK — the product watches itself (Session 8, A1 board)
+
+The in-app half (settle alarms, failure pager, quota check, box health, daily
+pulse) lives in `src/snapshotScheduler.js` + `src/services/opsWatch.js` +
+`src/services/systemHealth.js` and pushes to ntfy via `src/utils/opsNotify.js`.
+This runbook is the BOX-SIDE half: the monitors and hooks that catch the cases
+the app cannot report on itself — a dead container, a failed deploy, a dead box.
+
+Channel: `https://ntfy.sh/vyndr-pipeline-kev2026`
+(override with `NTFY_URL` / `NTFY_TOPIC`; both default in `opsNotify.js`).
+Zero out-of-pocket: ntfy.sh is free, Uptime Kuma is already on the box.
+
+---
+
+## 1. Phone subscription (do this first — alerts nobody reads are logs)
+
+Android / iOS:
+1. Install **ntfy** (Play Store, App Store, or F-Droid).
+2. Open the app → **+ Subscribe to topic**.
+3. Server: `https://ntfy.sh` (default). Topic: `vyndr-pipeline-kev2026`.
+4. Android: in the subscription's settings enable **Instant delivery**
+   (foreground service) and exempt ntfy from battery optimization
+   (Settings → Apps → ntfy → Battery → Unrestricted). Without this,
+   Android can delay high-priority pages by hours.
+5. Send yourself a probe and confirm the phone buzzes:
+   `curl -s -d "probe" https://ntfy.sh/vyndr-pipeline-kev2026`
+
+Desktop (optional): open `https://ntfy.sh/vyndr-pipeline-kev2026` in a browser
+tab and allow notifications.
+
+Note: the topic name is the only secret. Anyone who knows it can read and
+write the channel. If it ever leaks, rotate via `NTFY_TOPIC` on the container
+and re-subscribe the phone.
+
+## 2. Uptime Kuma monitors (external heartbeat — catches a dead container)
+
+Uptime Kuma already runs on the box. Add THREE monitors
+(**Add New Monitor** for each):
+
+| # | Name | Type | URL | Pass condition | Interval |
+|---|------|------|-----|----------------|----------|
+| 1 | `vyndr web` | HTTP(s) | `https://vyndr.app` | status 200 | 60s, retries 3 |
+| 2 | `vyndr api health` | HTTP(s) — **Keyword** | `https://api.vyndr.app/api/health` | keyword `healthy` present | 60s, retries 3 |
+| 3 | `vyndr snapshot summary` | HTTP(s) | `https://api.vyndr.app/api/snapshot/summary` | status 200 | 300s, retries 3 |
+
+Why keyword on #2: `/api/health` answers 200 `"status":"healthy"` only when
+Redis AND Supabase check out; degraded mode returns 503 `"degraded"`. The
+keyword check fails on BOTH a dead container and a degraded one.
+Monitor #3 is the public cache-only snapshot read — it proves the pipeline's
+output surface is serving, and it can never drain provider quota.
+
+Wire Kuma to the pager:
+1. **Settings → Notifications → Setup Notification → ntfy**.
+2. Server URL `https://ntfy.sh`, topic `vyndr-pipeline-kev2026`,
+   priority: default 5 (Kuma sends its own priority on down events).
+3. Check **Default enabled** so future monitors inherit it, then attach the
+   notification to all three monitors above.
+
+## 3. Coolify deploy-failure -> ntfy
+
+Coolify owns deploys, so the app can't report its own failed deploy. Two
+options, in order of preference:
+
+**A. Native notification channel (Coolify v4):**
+1. Coolify → **Notifications** (team level).
+2. If your Coolify version lists **ntfy**: server `https://ntfy.sh`, topic
+   `vyndr-pipeline-kev2026`, enable only the **Deployment Failed** (and
+   optionally **Deployment Success**) events.
+3. If it doesn't list ntfy, use the generic **Webhook** channel pointed at
+   `https://ntfy.sh/vyndr-pipeline-kev2026`. ntfy accepts any POST body as
+   the message text — a JSON payload arrives readable, just ugly.
+
+**B. Per-app webhook (any Coolify version):**
+1. App → **Webhooks** → add `https://ntfy.sh/vyndr-pipeline-kev2026` for the
+   deployment-failed event.
+2. Verify by triggering a deploy of a branch that fails its build; the phone
+   should receive the payload within seconds.
+
+Either way, send one manual probe to prove the path before trusting it:
+`curl -s -d "coolify webhook path test" https://ntfy.sh/vyndr-pipeline-kev2026`
+
+## 4. What the app already pages on its own (for reference — do not duplicate)
+
+| Alert | Source | Priority |
+|-------|--------|----------|
+| Snapshot success / 0-props / hard failure | `snapshotService` (S56) | default / low / high |
+| Missed cron slot (per-minute watchdog) | `snapshotScheduler` (S56) | high |
+| Settlement pass THREW (outcomes or ledger) | `snapshotScheduler` (S8) | high |
+| Morning settle closed with 0 settles while Postgres held pre-today pending rows | `snapshotScheduler` + `opsWatch.zeroSettleAlarm` (S8) | high |
+| Sport produced nothing 3 consecutive slots | `opsWatch.createFailureTracker` (S8) | high |
+| odds-api >= 80% quota (once per day) | `opsWatch.checkQuotaDaily` (S8) | high |
+| Disk > 85% or memory > 90% | `systemHealth` via daily pulse (S8) | high |
+| Daily pulse, 9 AM ET (13:00 UTC, `PULSE_HOUR_UTC` to move) | `snapshotScheduler` (S8) | default |
+
+Kill switches: `PIPELINE_ALERTS=0` silences everything app-side;
+`NODE_ENV=test` is silent by design.
+
+## 5. Weekly 5-minute sanity pass
+
+1. Phone got yesterday's 9 AM pulse. If not: check `ops:pulse:*` in Redis and
+   the container log for `[opsWatch] armed`.
+2. `GET /api/internal/snapshot/status` (header `x-internal-key`) shows
+   `cron_armed: true`, fresh `last_snapshot`, `overdue: false`.
+3. Uptime Kuma dashboard: three greens.
+4. Disk line in the pulse trending up week over week means log rotation or
+   Redis growth needs a look before the 85% page ever fires.
@@ -0,0 +1,83 @@
+# Session 8 (A1 board) — Ops: the product watches itself
+
+## Problem
+The pipeline alerts on snapshot success/failure and missed cron slots (S56),
+but the record can still die silently: a settlement pass that throws or
+quietly settles nothing, a sport erroring slot after slot, quota creeping to
+the block threshold, a full disk. Kev must be paged by the product, not
+discover it days later.
+
+## Scope (extends existing scheduler/services — nothing rebuilt)
+1. **Settlement alarm** (`snapshotScheduler.js`)
+   - Settle pass (outcomes or ledger) THROWS -> ntfy priority `high`.
+   - Zero-settle alarm: at the MORNING slot only (first configured hour
+     >= 06 UTC; 14 UTC default = 10 AM ET, the first slot after overnight
+     completions), alert when the ledger settle pass found rows from before
+     today but settled none. **Chosen signal (documented): the ledger settle
+     pass's own return values** (`settleLedger` fetches
+     `game_date < today AND outcome IS NULL` from Postgres). It is the
+     cheapest (zero extra reads) and the most reliable: Postgres survives the
+     Redis TTL expiry that silently killed morning settles in Session 60 —
+     a `snapshot:{sport}:previous` signal would vanish in exactly that
+     failure mode. Scoped to SETTLEABLE sports (mlb) because NBA/WNBA/soccer
+     rows legitimately stay pending until their settle feed exists.
+     A genuinely empty yesterday fetches 0 rows -> no alarm. Deduped once
+     per ET date via Redis `ops:settle_zero:{date}`.
+2. **Persistent snapshot failure** (`opsWatch.createFailureTracker`, pure)
+   - Per-sport consecutive count of bad outcomes (`status==='error'` or
+     `skipped` with reason `'no props'`) across cron slots (in-process Map).
+   - Pages ONCE (priority high) when a sport reaches 3 consecutive bad
+     slots; stays quiet at 4+; any good outcome resets and re-arms.
+3. **Quota alert** (`opsWatch.checkQuotaDaily`)
+   - After each snapshot run: odds-api usage >= 80% -> one alert per day
+     (Redis dedupe `ops:quota_day:{provider}:{YYYY-MM-DD}`). Complements the
+     once-per-MONTH quotaTracker warn.
+4. **Box health** (`src/services/systemHealth.js`, pure + injectable)
+   - `getSystemHealth({fsImpl, osImpl})` -> `{disk_pct, mem_pct}` via
+     `fs.promises.statfs('/')` + `os.freemem/totalmem`.
+   - `healthIssues()` flags disk > 85% / mem > 90%; checked in the daily
+     pulse; issues alert at priority high.
+5. **Daily pulse** (scheduler, 13:00 UTC = 9 AM EDT, dedupe by date
+   in-process + Redis `ops:pulse:{date}`), ONE notification:
+   ledger rows written yesterday (Supabase count via
+   `ledgerService.countRowsForDate`, `n/a` when unconfigured), settles in
+   the last 24h (from `outcomes:{sport}:log`), odds-api quota pct,
+   disk/mem, `desk pack: see /desk` placeholder.
+6. **Runbook** — `docs/OPS-RUNBOOK.md`: Uptime Kuma monitors, Coolify
+   deploy-failure webhook -> ntfy, phone subscription steps.
+
+## Data shapes
+- `createFailureTracker(threshold=3).record(sport, result)` ->
+  `{ sport, count, shouldPage }`
+- `zeroSettleAlarm(ledgerResults, settleable=['mlb'])` ->
+  `{ alarm, settled, pending }`
+- `checkQuotaDaily(deps)` -> `{ alerted, deduped?, pct }`
+- `getSystemHealth()` -> `{ disk_pct: 0-100|null, mem_pct: 0-100|null }`
+- `buildPulseMessage(pieces)` -> single multi-line string, VOICE v1.1
+  (deadpan, numbers, zero exclamation points).
+
+## Acceptance criteria
+- Settle-pass throw -> one high-priority ntfy; zero-settle morning alarm
+  fires only when Postgres held pre-today unsettled rows; empty yesterday
+  never alerts.
+- 3 consecutive bad slots page exactly once per losing streak; success
+  resets.
+- Quota >= 80% alerts at most once per day.
+- Pulse is ONE notification, fields present, `n/a` when Supabase off,
+  no `!` anywhere in any alert copy.
+- One real test alert POSTed to the live ntfy channel (HTTP result in the
+  session report). Full jest green.
+
+## Test plan
+- `tests/unit/opsWatch.test.js` — tracker (increment/page-once/reset,
+  per-sport isolation), zero-settle signal, morning-hour resolution, quota
+  dedupe (per-day, injected now), pulse assembly (fields + no-`!` lint),
+  recent-settle counting.
+- `tests/unit/systemHealth.test.js` — injected fs/os math, statfs failure
+  degrades to null, threshold flags at 85/90 boundaries.
+- `tests/unit/opsScheduler.test.js` — scheduler wiring: settle-throw alert,
+  3-slot page via ticks, pulse fires once at 13:00 with Redis dedupe.
+
+## Zero out-of-pocket
+ntfy.sh (existing channel) + Uptime Kuma (already on the box) + node
+built-ins (`fs.statfs`, `os`). No new dependency, no paid service.
@@ -361,6 +361,27 @@ async function defaultGetPlayerStats(name, sport) {
  return { found: false }; // no free settled-result feed yet → pending
 }

+/**
+ * Session 8 (A1 board, ops) — count of ledger rows for one game_date (the
+ * daily pulse's "rows written yesterday"). Returns null (NOT 0) when Supabase
+ * isn't configured or the count fails — the pulse renders "n/a", never a
+ * fabricated zero.
+ */
+async function countRowsForDate(gameDate, opts = {}) {
+  if (!gameDate) return null;
+  if (!opts.sb && !isConfigured()) return null;
+  try {
+    const sb = opts.sb || defaultClient();
+    const { count, error } = await sb.from('ledger_entries')
+      .select('id', { count: 'exact', head: true })
+      .eq('game_date', gameDate);
+    if (error) return null;
+    return count || 0;
+  } catch {
+    return null;
+  }
+}
+
 /**
 * 30-day aggregate over the PUBLIC model record (user_id NULL): hit rate,
 * beat-the-close rate, pending count. Percentages are null below
@@ -453,6 +474,7 @@ module.exports = {
  settleLedger,
  settleAllLedgers,
  applyRevision,
+  countRowsForDate,
  getModelAggregate,
  MIN_AGG_SAMPLE,
  __internals: {
@@ -0,0 +1,181 @@
+'use strict';
+
+/**
+ * opsWatch — pure ops-alarm logic for the scheduler (Session 8, A1 board).
+ *
+ * The record must never die silently. This module holds the TESTABLE pieces
+ * the scheduler wires to ntfy (via opsNotify):
+ *
+ *   1. createFailureTracker — per-sport CONSECUTIVE snapshot-failure counter.
+ *      Single-slot errors are normal before books post lines; 3 in a row is a
+ *      broken pipeline. Pages exactly once per losing streak (at the
+ *      threshold crossing), re-arms on any good outcome.
+ *   2. zeroSettleAlarm — "settle pass finished but the record did not
+ *      advance." SIGNAL CHOICE (documented): the ledger settle pass's own
+ *      return values. settleLedger fetches `game_date < today AND outcome IS
+ *      NULL` from POSTGRES, so the signal survives the Redis TTL expiry that
+ *      silently killed morning settles in Session 60 — a
+ *      snapshot:{sport}:previous read would vanish in exactly that failure
+ *      mode, and it costs zero extra reads. Scoped to SETTLEABLE sports (mlb)
+ *      because NBA/WNBA/soccer rows legitimately stay pending until they have
+ *      a settled-result feed. A genuinely empty yesterday fetches 0 rows ->
+ *      settled 0 / pending 0 -> no alarm, ever.
+ *   3. checkQuotaDaily — odds-api >= 80% -> one alert per day (Redis dedupe
+ *      by date key). Complements quotaTracker's once-per-MONTH warn.
+ *   4. buildPulseMessage / countRecentSettles — the 9 AM ET daily pulse, one
+ *      notification, VOICE v1.1: deadpan, numbers, no exclamation points.
+ *
+ * Everything here is pure or fully injectable — no requires of redis/ntfy.
+ */
+
+/** Sports with a real settled-result feed (mlb game logs). Keep in sync with
+ *  outcomeService/ledgerService's MLB-only settlement until Phase 4.5. */
+const SETTLEABLE_SPORTS = ['mlb'];
+
+const PAGE_THRESHOLD = 3;
+
+/** A snapshot result that means "the pipeline produced nothing": a hard error
+ *  or an empty odds feed. 'no grades' (props arrived, grader refused) and 'ok'
+ *  are NOT failures for this counter. */
+function isBadSnapshotResult(result) {
+  if (!result) return false;
+  if (result.status === 'error') return true;
+  return result.status === 'skipped' && String(result.reason || '') === 'no props';
+}
+
+/**
+ * Per-sport consecutive-failure counter (in-process; a restart re-arms, which
+ * is acceptable — the missed-cron watchdog covers a crashed scheduler).
+ * record() returns { sport, count, shouldPage }: shouldPage is true ONLY at
+ * the exact threshold crossing, so a sport that keeps failing pages once per
+ * losing streak, not once per slot.
+ */
+function createFailureTracker(threshold = PAGE_THRESHOLD) {
+  const counts = new Map();
+  return {
+    record(sport, result) {
+      const sp = String(sport || '').toLowerCase();
+      const bad = isBadSnapshotResult(result);
+      const count = bad ? (counts.get(sp) || 0) + 1 : 0;
+      counts.set(sp, count);
+      return { sport: sp, count, shouldPage: bad && count === threshold };
+    },
+    count(sport) { return counts.get(String(sport || '').toLowerCase()) || 0; },
+    threshold,
+  };
+}
+
+/** The daily "close the book" slot: the first configured UTC hour >= 06
+ *  (hours 0-5 UTC are late-night ET slots of the PREVIOUS ET day). Falls back
+ *  to the smallest configured hour when nothing is >= 06. */
+function morningHourUtc(hoursUtc) {
+  const hours = (hoursUtc || []).filter((h) => Number.isInteger(h));
+  if (hours.length === 0) return 14;
+  const daytime = hours.filter((h) => h >= 6);
+  return (daytime.length > 0 ? daytime : hours).sort((a, b) => a - b)[0];
+}
+
+/**
+ * Evaluate the zero-settle signal from settleAllLedgers results.
+ * alarm === true only when settleable-sport rows EXISTED for settlement
+ * (fetched from Postgres: settled + pending > 0) and none settled.
+ */
+function zeroSettleAlarm(ledgerResults, settleable = SETTLEABLE_SPORTS) {
+  const rows = (Array.isArray(ledgerResults) ? ledgerResults : [])
+    .filter((r) => r && settleable.includes(String(r.sport || '').toLowerCase()));
+  const settled = rows.reduce((n, r) => n + (r.settled || 0), 0);
+  const pending = rows.reduce((n, r) => n + (r.pending || 0), 0);
+  return { alarm: settled === 0 && pending > 0, settled, pending };
+}
+
+/** ET calendar date (YYYY-MM-DD) of a Date; UTC fallback if Intl is absent. */
+function dateET(d = new Date()) {
+  try {
+    return new Intl.DateTimeFormat('en-CA', {
+      timeZone: 'America/New_York', year: 'numeric', month: '2-digit', day: '2-digit',
+    }).format(d);
+  } catch {
+    return d.toISOString().slice(0, 10);
+  }
+}
+
+/**
+ * Quota alert: odds-api usage >= thresholdPct -> alert once per UTC day
+ * (Redis dedupe key ops:quota_day:{provider}:{date}). All deps injectable:
+ * { getStatus, cacheGet, cacheSet, notify, now, providerId, thresholdPct }.
+ * Never throws — an ops check must not break the snapshot run.
+ */
+async function checkQuotaDaily(deps = {}) {
+  const providerId = deps.providerId || 'odds-api';
+  const thresholdPct = deps.thresholdPct != null ? deps.thresholdPct : 0.8;
+  try {
+    const status = await deps.getStatus(providerId);
+    const pct = status && Number.isFinite(status.pct) ? status.pct : null;
+    if (pct == null || pct < thresholdPct) return { alerted: false, pct };
+    const day = (deps.now ? deps.now() : new Date()).toISOString().slice(0, 10);
+    const key = `ops:quota_day:${providerId}:${day}`;
+    if (await deps.cacheGet(key)) return { alerted: false, deduped: true, pct };
+    await deps.cacheSet(key, '1', 2 * 24 * 3600);
+    await deps.notify(
+      `${providerId} at ${Math.round(pct * 100)}% of ${status.quotaType || 'period'} quota `
+      + `(${status.used}/${status.limit}). Blocks at 95. PropLine stays primary; `
+      + `the backup path is thinning.`,
+      { title: 'VYNDR quota', priority: 'high', tags: ['warning', 'chart_decreasing'] },
+    );
+    return { alerted: true, pct };
+  } catch (err) {
+    return { alerted: false, error: err.message };
+  }
+}
+
+/** Count settled outcomes across logs whose settledAt falls inside the last
+ *  windowHours (default 24). Pure — feed it the outcomes:{sport}:log arrays. */
+function countRecentSettles(logs, nowMs, windowHours = 24) {
+  const cutoff = nowMs - windowHours * 3600 * 1000;
+  let n = 0;
+  for (const log of logs || []) {
+    for (const o of Array.isArray(log) ? log : []) {
+      const t = o && o.settledAt ? new Date(o.settledAt).getTime() : NaN;
+      if (Number.isFinite(t) && t >= cutoff) n += 1;
+    }
+  }
+  return n;
+}
+
+/**
+ * Assemble the ONE daily-pulse notification body. VOICE v1.1: deadpan,
+ * numbers, no exclamation points. Missing data renders as "n/a" — a pulse
+ * never fabricates a zero.
+ * pieces: { dateEt, ledgerRows (number|null), settles24h (number|null),
+ *           quota ({pct, used, limit}|null), health ({disk_pct, mem_pct}|null) }
+ */
+function buildPulseMessage(pieces = {}) {
+  const na = (v) => (v == null ? 'n/a' : String(v));
+  const quota = pieces.quota && Number.isFinite(pieces.quota.pct)
+    ? `${Math.round(pieces.quota.pct * 100)}% (${pieces.quota.used}/${pieces.quota.limit})`
+    : 'n/a';
+  const h = pieces.health || {};
+  const disk = Number.isFinite(h.disk_pct) ? `${h.disk_pct}%` : 'n/a';
+  const mem = Number.isFinite(h.mem_pct) ? `${h.mem_pct}%` : 'n/a';
+  return [
+    `VYNDR pulse — ${pieces.dateEt || dateET()}`,
+    `ledger rows yesterday: ${na(pieces.ledgerRows)}`,
+    `settles last 24h: ${na(pieces.settles24h)}`,
+    `odds-api quota: ${quota}`,
+    `disk ${disk} · mem ${mem}`,
+    'desk pack: see /desk',
+  ].join('\n');
+}
+
+module.exports = {
+  createFailureTracker,
+  isBadSnapshotResult,
+  zeroSettleAlarm,
+  morningHourUtc,
+  checkQuotaDaily,
+  countRecentSettles,
+  buildPulseMessage,
+  dateET,
+  SETTLEABLE_SPORTS,
+  PAGE_THRESHOLD,
+};
@@ -0,0 +1,60 @@
+'use strict';
+
+/**
+ * systemHealth — box vitals (Session 8, A1 board — ops).
+ *
+ * Pure + injectable: disk usage via fs.promises.statfs('/'), memory via
+ * os.freemem/os.totalmem. Returns percentages (0-100, rounded) or null when a
+ * probe fails — the caller renders "n/a", never a fabricated number (Data
+ * Semantics Rule applies to ops copy too). Zero dependencies, zero cost.
+ */
+
+const fsp = require('fs').promises;
+const os = require('os');
+
+/** Alerting thresholds — checked by the daily pulse. */
+const THRESHOLDS = { disk_pct: 85, mem_pct: 90 };
+
+const roundPct = (frac) => Math.min(100, Math.max(0, Math.round(frac * 100)));
+
+/**
+ * { disk_pct, mem_pct } — each null when its probe fails.
+ * opts: { fsImpl (statfs), osImpl (freemem/totalmem), path }.
+ */
+async function getSystemHealth(opts = {}) {
+  const fsImpl = opts.fsImpl || fsp;
+  const osImpl = opts.osImpl || os;
+  const out = { disk_pct: null, mem_pct: null };
+  try {
+    const s = await fsImpl.statfs(opts.path || '/');
+    if (s && Number.isFinite(s.blocks) && s.blocks > 0 && Number.isFinite(s.bavail)) {
+      out.disk_pct = roundPct(1 - s.bavail / s.blocks);
+    }
+  } catch { /* disk_pct stays null */ }
+  try {
+    const total = osImpl.totalmem();
+    const free = osImpl.freemem();
+    if (Number.isFinite(total) && total > 0 && Number.isFinite(free)) {
+      out.mem_pct = roundPct(1 - free / total);
+    }
+  } catch { /* mem_pct stays null */ }
+  return out;
+}
+
+/**
+ * Deadpan issue lines for anything over threshold (strictly >). Empty array =
+ * healthy. Copy carries the numbers; no punctuation theatrics.
+ */
+function healthIssues(health, thresholds = THRESHOLDS) {
+  const issues = [];
+  if (!health) return issues;
+  if (Number.isFinite(health.disk_pct) && health.disk_pct > thresholds.disk_pct) {
+    issues.push(`disk at ${health.disk_pct}% (threshold ${thresholds.disk_pct}%)`);
+  }
+  if (Number.isFinite(health.mem_pct) && health.mem_pct > thresholds.mem_pct) {
+    issues.push(`memory at ${health.mem_pct}% (threshold ${thresholds.mem_pct}%)`);
+  }
+  return issues;
+}
+
+module.exports = { getSystemHealth, healthIssues, THRESHOLDS };
@@ -60,9 +60,19 @@ function startSnapshotScheduler(opts = {}) {
  const settleLedgers = opts.settleAllLedgers || require('./services/ledgerService').settleAllLedgers;
  const notify = opts.notify || require('./utils/opsNotify').notify;
  const cacheGet = opts.cacheGet || require('./utils/redis').cacheGet;
+  const cacheSet = opts.cacheSet || require('./utils/redis').cacheSet;
  const now = opts.now || (() => new Date());
+  // Session 8 (A1 board) — ops: the product watches itself. Pure logic lives
+  // in services/opsWatch + services/systemHealth; this file only wires it.
+  const opsWatch = require('./services/opsWatch');
+  const getSystemHealth = opts.getSystemHealth || require('./services/systemHealth').getSystemHealth;
+  const healthIssues = opts.healthIssues || require('./services/systemHealth').healthIssues;
+  const getQuotaStatus = opts.getQuotaStatus || require('./services/quotaTracker').getQuotaStatus;
+  const countLedgerRows = opts.countLedgerRows || require('./services/ledgerService').countRowsForDate;
+  const failureTracker = opts.failureTracker || opsWatch.createFailureTracker();
  let lastFiredSlot = null;
  let lastOverdueSlot = null;
+  let lastPulseDate = null;

  // Session 56 — missed-cron watchdog. Runs every minute (independent of the
  // fire schedule): if a scheduled slot came and went without a snapshot, alert
@@ -83,8 +93,55 @@ function startSnapshotScheduler(opts = {}) {
    } catch { /* watchdog must never throw */ }
  };

+  // Session 8 — daily 9 AM ET pulse (13:00 UTC), ONE notification. Dedupe is
+  // belt-and-braces: in-process date + a Redis key that survives a restart
+  // inside the same day. Runs on the per-minute cadence, independent of slots.
+  const PULSE_HOUR_UTC = Number.parseInt(process.env.PULSE_HOUR_UTC || '13', 10);
+  const pulseTick = async () => {
+    try {
+      const d = now();
+      if (d.getUTCHours() !== PULSE_HOUR_UTC || d.getUTCMinutes() !== 0) return;
+      const dayKey = d.toISOString().slice(0, 10);
+      if (dayKey === lastPulseDate) return;
+      lastPulseDate = dayKey;
+      if (await cacheGet(`ops:pulse:${dayKey}`)) return;   // already sent today (pre-restart)
+      await cacheSet(`ops:pulse:${dayKey}`, '1', 2 * 24 * 3600);
+
+      const yesterdayEt = opsWatch.dateET(new Date(d.getTime() - 24 * 3600 * 1000));
+      let ledgerRows = null;
+      try { ledgerRows = await countLedgerRows(yesterdayEt); } catch { /* n/a */ }
+      let settles24h = null;
+      try {
+        const logs = [];
+        for (const sp of ['mlb', 'nba', 'wnba', 'soccer']) {
+          const raw = await cacheGet(`outcomes:${sp}:log`);
+          logs.push(Array.isArray(raw) ? raw : (raw && raw.log) || []);
+        }
+        settles24h = opsWatch.countRecentSettles(logs, d.getTime());
+      } catch { /* n/a */ }
+      let quota = null;
+      try { quota = await getQuotaStatus('odds-api'); } catch { /* n/a */ }
+      let health = null;
+      try { health = await getSystemHealth(); } catch { /* n/a */ }
+
+      await notify(
+        opsWatch.buildPulseMessage({ dateEt: opsWatch.dateET(d), ledgerRows, settles24h, quota, health }),
+        { title: 'VYNDR daily pulse', tags: ['newspaper'] },
+      );
+      // Box health pages separately at high priority — the pulse is a read,
+      // the threshold breach is an action item.
+      const issues = healthIssues(health);
+      if (issues.length > 0) {
+        await notify(`Box health: ${issues.join('; ')}. The record does not fit on a full disk.`, {
+          title: 'VYNDR box', priority: 'high', tags: ['warning'],
+        });
+      }
+    } catch { /* the pulse must never break the scheduler */ }
+  };
+
  const tick = async () => {
    await checkOverdue();
+    await pulseTick();
    const d = now();
    if (d.getUTCMinutes() !== 0) return;
    const h = d.getUTCHours();
@@ -92,20 +149,47 @@ function startSnapshotScheduler(opts = {}) {
    const slot = `${d.getUTCFullYear()}-${d.getUTCMonth()}-${d.getUTCDate()}-${h}`;
    if (slot === lastFiredSlot) return; // fire once per slot
    lastFiredSlot = slot;
+    // Session 8 — a settle pass that THROWS is a paged event, not a log line.
+    // Settlement failing quietly is how the record dies.
    try {
      const settled = await settleAll();
      const totalSettled = settled.reduce((n, r) => n + (r.settled || 0), 0);
      console.log(`[outcomes] cron fired ${h}:00 UTC — ${totalSettled} props settled vs real results`);
    } catch (e) {
      console.warn('[outcomes] settle run failed:', e.message);
+      await notify(`Outcome settlement THREW at ${h}:00 UTC — ${e.message}. Yesterday's grades are unsettled until the next slot.`, {
+        title: 'VYNDR settlement', priority: 'high', tags: ['rotating_light'],
+      });
    }
+    let ledgerResults = null;
    try {
-      const ledger = await settleLedgers();
-      const n = ledger.reduce((t, r) => t + (r.settled || 0), 0);
+      ledgerResults = await settleLedgers();
+      const n = ledgerResults.reduce((t, r) => t + (r.settled || 0), 0);
      console.log(`[ledger] settle pass — ${n} entries settled (outcome + CLV)`);
    } catch (e) {
      console.warn('[ledger] settle run failed:', e.message);
+      await notify(`Ledger settlement THREW at ${h}:00 UTC — ${e.message}. The public record did not advance.`, {
+        title: 'VYNDR settlement', priority: 'high', tags: ['rotating_light'],
+      });
    }
+    // Session 8 — zero-settle alarm, MORNING slot only (the book-closing pass).
+    // Signal = the ledger settle's own Postgres-backed return values (see
+    // opsWatch.zeroSettleAlarm for why not snapshot:{sport}:previous). Deduped
+    // once per ET date; a genuinely empty yesterday (0 rows fetched) never fires.
+    try {
+      if (h === opsWatch.morningHourUtc(HOURS_UTC) && ledgerResults) {
+        const z = opsWatch.zeroSettleAlarm(ledgerResults);
+        if (z.alarm) {
+          const dk = `ops:settle_zero:${opsWatch.dateET(d)}`;
+          if (!(await cacheGet(dk))) {
+            await cacheSet(dk, '1', 2 * 24 * 3600);
+            await notify(`Morning settle pass closed with 0 settles — ${z.pending} ledger rows from before today are still pending. Yesterday had graded rows; the record did not advance.`, {
+              title: 'VYNDR settlement', priority: 'high', tags: ['rotating_light'],
+            });
+          }
+        }
+      }
+    } catch { /* alarm evaluation must never break the tick */ }
    try {
      const results = await runAll();
      const ok = results.filter((r) => r.status === 'ok');
@@ -116,9 +200,22 @@ function startSnapshotScheduler(opts = {}) {
        const total = results.reduce((n, r) => n + (r.gradeCount || 0), 0);
        await notify(`Desk pack ready — ${total} props graded across ${ok.length} sports. vyndr.app/desk`, { title: 'VYNDR desk', tags: ['newspaper'] });
      }
+      // Session 8 — persistent-failure pager: 3+ CONSECUTIVE erroring slots for
+      // a sport pages once (single-slot errors are normal before lines post).
+      for (const r of results) {
+        const t = failureTracker.record(r.sport, r);
+        if (t.shouldPage) {
+          await notify(`${String(r.sport).toUpperCase()} snapshot has produced nothing for ${t.count} consecutive slots (latest: ${r.status}${r.reason ? ` — ${r.reason}` : ''}). One empty slot is a quiet book; ${t.count} is a broken pipe.`, {
+            title: 'VYNDR pipeline', priority: 'high', tags: ['rotating_light'],
+          });
+        }
+      }
    } catch (e) {
      console.warn('[snapshot] cron run failed:', e.message);
    }
+    // Session 8 — quota check after each snapshot run: odds-api >= 80% alerts
+    // once per day (Redis-deduped). Never throws (guarded inside checkQuotaDaily).
+    await opsWatch.checkQuotaDaily({ getStatus: getQuotaStatus, cacheGet, cacheSet, notify, now: () => now() });
  };

  // Session 60 (night2/D) — Phase 2.5 intraday refresh. Every
@@ -158,7 +255,9 @@ function startSnapshotScheduler(opts = {}) {
  // was invisible at boot, which made "is settlement scheduled?" unanswerable
  // from logs. This line makes it verifiable forever.
  console.log(`[settlement] armed — outcomes + ledger settle pass runs FIRST at each snapshot slot (${HOURS_UTC.join(',')} UTC), idempotent re-runs`);
-  return { interval, tick, refreshTick };
+  // Session 8 — same verifiability rule: every watchdog states itself at boot.
+  console.log(`[opsWatch] armed — settle alarms (throw + morning zero-settle), failure pager (${failureTracker.threshold} consecutive), quota daily check, pulse ${PULSE_HOUR_UTC}:00 UTC`);
+  return { interval, tick, refreshTick, pulseTick };
 }

 module.exports = { startSnapshotScheduler, HOURS_UTC, mostRecentExpectedSlot, isSnapshotOverdue };
@@ -0,0 +1,207 @@
+'use strict';
+
+// Session 8 (A1 board) — scheduler ops wiring: settle alarms, failure pager,
+// quota check, daily pulse. All deps injected; zero network, zero redis.
+
+const { startSnapshotScheduler } = require('../../src/snapshotScheduler');
+
+function memCache(seed = {}) {
+  const store = { ...seed };
+  return {
+    store,
+    cacheGet: async (k) => (k in store ? store[k] : null),
+    cacheSet: async (k, v) => { store[k] = v; return true; },
+  };
+}
+
+function collector() {
+  const msgs = [];
+  return { msgs, notify: async (m, o = {}) => { msgs.push({ m, o }); return { sent: true }; } };
+}
+
+// Base deps: quiet pipeline, everything healthy, injectable clock.
+function baseDeps(cache, notes, dRef) {
+  return {
+    ...cache,
+    notify: notes.notify,
+    now: () => dRef.d,
+    runAllSnapshots: async () => [{ sport: 'mlb', status: 'ok', gradeCount: 5 }],
+    settleAllOutcomes: async () => [{ sport: 'mlb', settled: 3, pending: 0 }],
+    settleAllLedgers: async () => [{ sport: 'mlb', settled: 3, pending: 1 }],
+    getQuotaStatus: async () => ({ pct: 0.1, used: 50, limit: 500 }),
+    getSystemHealth: async () => ({ disk_pct: 40, mem_pct: 50 }),
+    countLedgerRows: async () => 42,
+  };
+}
+
+function makeSched(overrides = {}, startIso = '2026-07-11T14:00:00Z') {
+  process.env.SNAPSHOT_CRON = '1';
+  const cache = memCache();
+  const notes = collector();
+  const dRef = { d: new Date(startIso) };
+  const logSpy = jest.spyOn(console, 'log').mockImplementation(() => {});
+  const warnSpy = jest.spyOn(console, 'warn').mockImplementation(() => {});
+  const sched = startSnapshotScheduler({ ...baseDeps(cache, notes, dRef), ...overrides });
+  clearInterval(sched.interval);
+  logSpy.mockRestore();
+  warnSpy.mockRestore();
+  return { sched, cache, notes, dRef };
+}
+
+afterEach(() => { delete process.env.SNAPSHOT_CRON; });
+
+describe('settlement alarm', () => {
+  test('outcome settle THROW pages high priority', async () => {
+    const { sched, notes } = makeSched({
+      settleAllOutcomes: async () => { throw new Error('mlb stats 500'); },
+    });
+    await sched.tick();
+    const alert = notes.msgs.find((x) => x.m.includes('Outcome settlement THREW'));
+    expect(alert).toBeTruthy();
+    expect(alert.o.priority).toBe('high');
+    expect(alert.m).toContain('mlb stats 500');
+    expect(alert.m).not.toContain('!');
+  });
+
+  test('ledger settle THROW pages high priority', async () => {
+    const { sched, notes } = makeSched({
+      settleAllLedgers: async () => { throw new Error('supabase timeout'); },
+    });
+    await sched.tick();
+    const alert = notes.msgs.find((x) => x.m.includes('Ledger settlement THREW'));
+    expect(alert).toBeTruthy();
+    expect(alert.o.priority).toBe('high');
+  });
+
+  test('morning zero-settle alarm fires once per day, only when rows were pending', async () => {
+    const { sched, notes, cache } = makeSched({
+      settleAllLedgers: async () => [{ sport: 'mlb', settled: 0, pending: 12 }],
+    });
+    await sched.tick(); // 14:00 UTC — the morning slot
+    const alarms = notes.msgs.filter((x) => x.m.includes('0 settles'));
+    expect(alarms).toHaveLength(1);
+    expect(alarms[0].m).toContain('12 ledger rows');
+    expect(alarms[0].o.priority).toBe('high');
+    expect(cache.store['ops:settle_zero:2026-07-11']).toBe('1');
+    expect(alarms[0].m).not.toContain('!');
+  });
+
+  test('redis dedupe suppresses a second zero-settle alarm the same ET day', async () => {
+    const { sched, notes } = makeSched({
+      settleAllLedgers: async () => [{ sport: 'mlb', settled: 0, pending: 12 }],
+      cacheGet: async (k) => (k === 'ops:settle_zero:2026-07-11' ? '1' : null),
+    });
+    await sched.tick();
+    expect(notes.msgs.filter((x) => x.m.includes('0 settles'))).toHaveLength(0);
+  });
+
+  test('no alarm at a non-morning slot even with zero settles', async () => {
+    const { sched, notes } = makeSched({
+      settleAllLedgers: async () => [{ sport: 'mlb', settled: 0, pending: 12 }],
+    }, '2026-07-11T19:00:00Z');
+    await sched.tick();
+    expect(notes.msgs.filter((x) => x.m.includes('0 settles'))).toHaveLength(0);
+  });
+
+  test('genuinely empty yesterday (0 fetched rows) never alarms', async () => {
+    const { sched, notes } = makeSched({
+      settleAllLedgers: async () => [{ sport: 'mlb', settled: 0, pending: 0 }],
+    });
+    await sched.tick();
+    expect(notes.msgs.filter((x) => x.m.includes('0 settles'))).toHaveLength(0);
+  });
+});
+
+describe('persistent snapshot-failure pager', () => {
+  test('three consecutive failing slots page exactly once; success re-arms', async () => {
+    let result = { sport: 'mlb', status: 'skipped', reason: 'no props', gradeCount: 0 };
+    const { sched, notes, dRef } = makeSched({
+      runAllSnapshots: async () => [result],
+    });
+    const slots = ['2026-07-11T14:00:00Z', '2026-07-11T19:00:00Z', '2026-07-11T22:00:00Z', '2026-07-12T01:00:00Z'];
+    for (const iso of slots.slice(0, 2)) { dRef.d = new Date(iso); await sched.tick(); }
+    expect(notes.msgs.filter((x) => x.m.includes('consecutive'))).toHaveLength(0);
+    dRef.d = new Date(slots[2]); await sched.tick();       // third consecutive
+    let pages = notes.msgs.filter((x) => x.m.includes('consecutive'));
+    expect(pages).toHaveLength(1);
+    expect(pages[0].o.priority).toBe('high');
+    expect(pages[0].m).toContain('MLB');
+    expect(pages[0].m).toContain('3 consecutive');
+    expect(pages[0].m).not.toContain('!');
+    dRef.d = new Date(slots[3]); await sched.tick();       // fourth — no re-page
+    expect(notes.msgs.filter((x) => x.m.includes('consecutive'))).toHaveLength(1);
+    // success resets, then three more failures page again
+    result = { sport: 'mlb', status: 'ok', gradeCount: 4 };
+    dRef.d = new Date('2026-07-12T03:00:00Z'); await sched.tick();
+    result = { sport: 'mlb', status: 'error', reason: 'down', gradeCount: 0 };
+    for (const iso of ['2026-07-12T14:00:00Z', '2026-07-12T19:00:00Z', '2026-07-12T22:00:00Z']) {
+      dRef.d = new Date(iso); await sched.tick();
+    }
+    expect(notes.msgs.filter((x) => x.m.includes('consecutive'))).toHaveLength(2);
+  });
+});
+
+describe('quota daily check', () => {
+  test('>= 80% after a snapshot run alerts once per day via redis dedupe', async () => {
+    const { sched, notes, dRef, cache } = makeSched({
+      getQuotaStatus: async () => ({ pct: 0.84, used: 420, limit: 500, quotaType: 'monthly' }),
+    });
+    await sched.tick();
+    dRef.d = new Date('2026-07-11T19:00:00Z');
+    await sched.tick(); // second slot, same day — deduped
+    const alerts = notes.msgs.filter((x) => x.o.title === 'VYNDR quota');
+    expect(alerts).toHaveLength(1);
+    expect(alerts[0].m).toContain('84%');
+    expect(cache.store['ops:quota_day:odds-api:2026-07-11']).toBe('1');
+  });
+});
+
+describe('daily pulse', () => {
+  test('fires ONE notification at 13:00 UTC with the assembled fields', async () => {
+    const { sched, notes } = makeSched({}, '2026-07-11T13:00:00Z');
+    await sched.tick();
+    await sched.tick(); // same minute — in-process dedupe
+    const pulses = notes.msgs.filter((x) => x.o.title === 'VYNDR daily pulse');
+    expect(pulses).toHaveLength(1);
+    expect(pulses[0].m).toContain('ledger rows yesterday: 42');
+    expect(pulses[0].m).toContain('odds-api quota: 10% (50/500)');
+    expect(pulses[0].m).toContain('disk 40% · mem 50%');
+    expect(pulses[0].m).toContain('desk pack: see /desk');
+    expect(pulses[0].m).not.toContain('!');
+  });
+
+  test('redis date key suppresses a duplicate pulse after a restart', async () => {
+    const { sched, notes } = makeSched({
+      cacheGet: async (k) => (k === 'ops:pulse:2026-07-11' ? '1' : null),
+    }, '2026-07-11T13:00:00Z');
+    await sched.tick();
+    expect(notes.msgs.filter((x) => x.o.title === 'VYNDR daily pulse')).toHaveLength(0);
+  });
+
+  test('does not fire off the pulse hour', async () => {
+    const { sched, notes } = makeSched({}, '2026-07-11T12:00:00Z');
+    await sched.tick();
+    expect(notes.msgs.filter((x) => x.o.title === 'VYNDR daily pulse')).toHaveLength(0);
+  });
+
+  test('box health over threshold pages separately at high priority', async () => {
+    const { sched, notes } = makeSched({
+      getSystemHealth: async () => ({ disk_pct: 92, mem_pct: 50 }),
+    }, '2026-07-11T13:00:00Z');
+    await sched.tick();
+    const box = notes.msgs.filter((x) => x.o.title === 'VYNDR box');
+    expect(box).toHaveLength(1);
+    expect(box[0].o.priority).toBe('high');
+    expect(box[0].m).toContain('disk at 92%');
+    expect(box[0].m).not.toContain('!');
+  });
+
+  test('supabase unconfigured renders n/a, never a fabricated zero', async () => {
+    const { sched, notes } = makeSched({
+      countLedgerRows: async () => null,
+    }, '2026-07-11T13:00:00Z');
+    await sched.tick();
+    const pulse = notes.msgs.find((x) => x.o.title === 'VYNDR daily pulse');
+    expect(pulse.m).toContain('ledger rows yesterday: n/a');
+  });
+});
@@ -0,0 +1,209 @@
+'use strict';
+
+// Session 8 (A1 board) — opsWatch: the pure ops-alarm logic.
+
+const {
+  createFailureTracker,
+  isBadSnapshotResult,
+  zeroSettleAlarm,
+  morningHourUtc,
+  checkQuotaDaily,
+  countRecentSettles,
+  buildPulseMessage,
+  dateET,
+} = require('../../src/services/opsWatch');
+
+function memCache(seed = {}) {
+  const store = { ...seed };
+  return {
+    store,
+    cacheGet: async (k) => (k in store ? store[k] : null),
+    cacheSet: async (k, v) => { store[k] = v; return true; },
+  };
+}
+
+describe('isBadSnapshotResult', () => {
+  test('error and skipped/no-props are bad; ok and skipped/no-grades are not', () => {
+    expect(isBadSnapshotResult({ status: 'error', reason: 'down' })).toBe(true);
+    expect(isBadSnapshotResult({ status: 'skipped', reason: 'no props' })).toBe(true);
+    expect(isBadSnapshotResult({ status: 'ok' })).toBe(false);
+    expect(isBadSnapshotResult({ status: 'skipped', reason: 'no grades' })).toBe(false);
+    expect(isBadSnapshotResult(null)).toBe(false);
+  });
+});
+
+describe('createFailureTracker', () => {
+  test('pages exactly once at the third consecutive failure, stays quiet after', () => {
+    const t = createFailureTracker(3);
+    const bad = { status: 'error', reason: 'x' };
+    expect(t.record('mlb', bad).shouldPage).toBe(false);   // 1
+    expect(t.record('mlb', bad).shouldPage).toBe(false);   // 2
+    const third = t.record('mlb', bad);
+    expect(third.shouldPage).toBe(true);                   // 3 -> page once
+    expect(third.count).toBe(3);
+    expect(t.record('mlb', bad).shouldPage).toBe(false);   // 4 -> no re-page
+    expect(t.count('mlb')).toBe(4);
+  });
+
+  test('any good outcome resets the counter and re-arms the pager', () => {
+    const t = createFailureTracker(3);
+    const bad = { status: 'skipped', reason: 'no props' };
+    t.record('mlb', bad); t.record('mlb', bad);
+    t.record('mlb', { status: 'ok' });                     // reset
+    expect(t.count('mlb')).toBe(0);
+    t.record('mlb', bad); t.record('mlb', bad);
+    expect(t.record('mlb', bad).shouldPage).toBe(true);    // re-armed -> pages again
+  });
+
+  test('counts are independent per sport', () => {
+    const t = createFailureTracker(3);
+    const bad = { status: 'error' };
+    t.record('mlb', bad); t.record('mlb', bad);
+    expect(t.record('wnba', bad).count).toBe(1);
+    expect(t.count('mlb')).toBe(2);
+  });
+});
+
+describe('zeroSettleAlarm', () => {
+  test('alarms when settleable rows existed and none settled', () => {
+    const z = zeroSettleAlarm([{ sport: 'mlb', settled: 0, pending: 12 }]);
+    expect(z).toEqual({ alarm: true, settled: 0, pending: 12 });
+  });
+
+  test('never alarms on a genuinely empty yesterday (0 rows fetched)', () => {
+    expect(zeroSettleAlarm([{ sport: 'mlb', settled: 0, pending: 0 }]).alarm).toBe(false);
+    expect(zeroSettleAlarm([]).alarm).toBe(false);
+    expect(zeroSettleAlarm(null).alarm).toBe(false);
+  });
+
+  test('no alarm when anything settled', () => {
+    expect(zeroSettleAlarm([{ sport: 'mlb', settled: 3, pending: 9 }]).alarm).toBe(false);
+  });
+
+  test('ignores sports without a settled-result feed (WNBA pendings are honest)', () => {
+    const z = zeroSettleAlarm([
+      { sport: 'mlb', settled: 0, pending: 0 },
+      { sport: 'wnba', settled: 0, pending: 40 },
+    ]);
+    expect(z.alarm).toBe(false);
+  });
+});
+
+describe('morningHourUtc', () => {
+  test('first configured hour >= 06 UTC (1,3 are late-night ET of the prior day)', () => {
+    expect(morningHourUtc([14, 19, 22, 1, 3])).toBe(14);
+    expect(morningHourUtc([19, 14])).toBe(14);
+  });
+  test('falls back to the smallest hour when nothing is daytime', () => {
+    expect(morningHourUtc([1, 3])).toBe(1);
+    expect(morningHourUtc([])).toBe(14);
+  });
+});
+
+describe('checkQuotaDaily', () => {
+  const notifyCollector = () => {
+    const msgs = [];
+    return { msgs, notify: async (m, o) => { msgs.push({ m, o }); return { sent: true }; } };
+  };
+
+  test('below 80% -> no alert', async () => {
+    const { msgs, notify } = notifyCollector();
+    const cache = memCache();
+    const r = await checkQuotaDaily({
+      getStatus: async () => ({ pct: 0.5, used: 250, limit: 500 }),
+      ...cache, notify, now: () => new Date('2026-07-11T15:00:00Z'),
+    });
+    expect(r.alerted).toBe(false);
+    expect(msgs).toHaveLength(0);
+  });
+
+  test('>= 80% alerts once, then dedupes for the rest of the day', async () => {
+    const { msgs, notify } = notifyCollector();
+    const cache = memCache();
+    const deps = {
+      getStatus: async () => ({ pct: 0.82, used: 410, limit: 500, quotaType: 'monthly' }),
+      ...cache, notify, now: () => new Date('2026-07-11T15:00:00Z'),
+    };
+    expect((await checkQuotaDaily(deps)).alerted).toBe(true);
+    const second = await checkQuotaDaily(deps);
+    expect(second.alerted).toBe(false);
+    expect(second.deduped).toBe(true);
+    expect(msgs).toHaveLength(1);
+    expect(msgs[0].m).toContain('82%');
+    expect(msgs[0].m).toContain('410/500');
+    expect(msgs[0].o.priority).toBe('high');
+    expect(msgs[0].m).not.toContain('!');
+    expect(cache.store['ops:quota_day:odds-api:2026-07-11']).toBe('1');
+  });
+
+  test('fires again the next day (date-keyed dedupe)', async () => {
+    const { msgs, notify } = notifyCollector();
+    const cache = memCache();
+    const base = {
+      getStatus: async () => ({ pct: 0.9, used: 450, limit: 500 }),
+      ...cache, notify,
+    };
+    await checkQuotaDaily({ ...base, now: () => new Date('2026-07-11T15:00:00Z') });
+    await checkQuotaDaily({ ...base, now: () => new Date('2026-07-12T15:00:00Z') });
+    expect(msgs).toHaveLength(2);
+  });
+
+  test('never throws — a broken status probe returns { alerted: false }', async () => {
+    const r = await checkQuotaDaily({ getStatus: async () => { throw new Error('redis gone'); } });
+    expect(r.alerted).toBe(false);
+    expect(r.error).toBe('redis gone');
+  });
+});
+
+describe('countRecentSettles', () => {
+  test('counts settledAt inside the window across multiple sport logs', () => {
+    const nowMs = Date.parse('2026-07-11T13:00:00Z');
+    const logs = [
+      [ { settledAt: '2026-07-11T14:05:00Z' }, { settledAt: '2026-07-09T14:00:00Z' } ], // 1 in, 1 out
+      [ { settledAt: '2026-07-10T14:30:00Z' }, { noSettledAt: true } ],                 // 1 in
+    ];
+    expect(countRecentSettles(logs, nowMs)).toBe(2);
+    expect(countRecentSettles([], nowMs)).toBe(0);
+    expect(countRecentSettles(null, nowMs)).toBe(0);
+  });
+});
+
+describe('buildPulseMessage', () => {
+  test('one message with every field, real numbers rendered', () => {
+    const msg = buildPulseMessage({
+      dateEt: '2026-07-11',
+      ledgerRows: 42,
+      settles24h: 38,
+      quota: { pct: 0.12, used: 61, limit: 500 },
+      health: { disk_pct: 41, mem_pct: 63 },
+    });
+    expect(msg).toContain('VYNDR pulse — 2026-07-11');
+    expect(msg).toContain('ledger rows yesterday: 42');
+    expect(msg).toContain('settles last 24h: 38');
+    expect(msg).toContain('odds-api quota: 12% (61/500)');
+    expect(msg).toContain('disk 41% · mem 63%');
+    expect(msg).toContain('desk pack: see /desk');
+  });
+
+  test('missing data renders n/a — never a fabricated zero', () => {
+    const msg = buildPulseMessage({ dateEt: '2026-07-11', ledgerRows: null, settles24h: null, quota: null, health: null });
+    expect(msg).toContain('ledger rows yesterday: n/a');
+    expect(msg).toContain('settles last 24h: n/a');
+    expect(msg).toContain('odds-api quota: n/a');
+    expect(msg).toContain('disk n/a · mem n/a');
+  });
+
+  test('VOICE v1.1 — no exclamation points, ever', () => {
+    const loud = buildPulseMessage({ ledgerRows: 999, settles24h: 999, quota: { pct: 0.99, used: 495, limit: 500 }, health: { disk_pct: 99, mem_pct: 99 } });
+    expect(loud).not.toContain('!');
+    expect(buildPulseMessage({})).not.toContain('!');
+  });
+});
+
+describe('dateET', () => {
+  test('renders the America/New_York calendar date (late UTC rolls back)', () => {
+    // 03:00 UTC Jul 11 = 11:00 PM ET Jul 10.
+    expect(dateET(new Date('2026-07-11T03:00:00Z'))).toBe('2026-07-10');
+    expect(dateET(new Date('2026-07-11T15:00:00Z'))).toBe('2026-07-11');
+  });
+});
@@ -0,0 +1,58 @@
+'use strict';
+
+// Session 8 (A1 board) — systemHealth: box vitals, pure + injectable.
+
+const { getSystemHealth, healthIssues, THRESHOLDS } = require('../../src/services/systemHealth');
+
+describe('getSystemHealth', () => {
+  test('computes disk_pct from statfs and mem_pct from os', async () => {
+    const fsImpl = { statfs: async () => ({ blocks: 1000, bavail: 400 }) };   // 60% used
+    const osImpl = { totalmem: () => 100, freemem: () => 25 };                // 75% used
+    const h = await getSystemHealth({ fsImpl, osImpl });
+    expect(h).toEqual({ disk_pct: 60, mem_pct: 75 });
+  });
+
+  test('a failed statfs degrades disk_pct to null, mem still reports', async () => {
+    const fsImpl = { statfs: async () => { throw new Error('EACCES'); } };
+    const osImpl = { totalmem: () => 10, freemem: () => 1 };
+    const h = await getSystemHealth({ fsImpl, osImpl });
+    expect(h.disk_pct).toBeNull();
+    expect(h.mem_pct).toBe(90);
+  });
+
+  test('zero/invalid totals degrade to null rather than dividing by zero', async () => {
+    const h = await getSystemHealth({
+      fsImpl: { statfs: async () => ({ blocks: 0, bavail: 0 }) },
+      osImpl: { totalmem: () => 0, freemem: () => 0 },
+    });
+    expect(h).toEqual({ disk_pct: null, mem_pct: null });
+  });
+
+  test('real (uninjected) call returns numbers on this box', async () => {
+    const h = await getSystemHealth();
+    expect(h.disk_pct === null || (h.disk_pct >= 0 && h.disk_pct <= 100)).toBe(true);
+    expect(h.mem_pct).toBeGreaterThanOrEqual(0);
+    expect(h.mem_pct).toBeLessThanOrEqual(100);
+  });
+});
+
+describe('healthIssues', () => {
+  test('thresholds: disk > 85, mem > 90 — strictly greater', () => {
+    expect(healthIssues({ disk_pct: 85, mem_pct: 90 })).toEqual([]);
+    expect(healthIssues({ disk_pct: 86, mem_pct: 90 })).toEqual(['disk at 86% (threshold 85%)']);
+    expect(healthIssues({ disk_pct: 40, mem_pct: 91 })).toEqual(['memory at 91% (threshold 90%)']);
+    expect(healthIssues({ disk_pct: 99, mem_pct: 99 })).toHaveLength(2);
+  });
+
+  test('null probes never flag; copy carries no exclamation points', () => {
+    expect(healthIssues({ disk_pct: null, mem_pct: null })).toEqual([]);
+    expect(healthIssues(null)).toEqual([]);
+    for (const line of healthIssues({ disk_pct: 99, mem_pct: 99 })) {
+      expect(line).not.toContain('!');
+    }
+  });
+
+  test('exported thresholds match the spec', () => {
+    expect(THRESHOLDS).toEqual({ disk_pct: 85, mem_pct: 90 });
+  });
+});