// PLATFORM KPIS

NINETEEN NUMBERS.
ONE DASHBOARD.

The problem is not “Google Analytics for the VPS.” The problem is answering how the factory is doing without opening five apps. Sigma aggregates 19 live KPIs from DRA, BRI, RRV, AIF, NLG, TKN, Kronos, SRM, Herald, GHT, PRL, FMR. Each query is try/except — missing table returns 0, never kills the page. Sigma owns only sigma_alert_rules (auto-create). 14-day DRA+BRI+RRV bar chart. Panels: content pipeline, AI infra cost, platform health (job success, SRM open, Herald unacked). Rules: metric_key + gt/gte/lt/lte/eq + threshold. Kronos POST /api/cron/eval-rules fires Herald emit_alert plus a Telegram call that no-ops without tokens. CSV of all metrics. MCP or login on metric APIs. Refresh every 5 minutes. Nexus is a different pane (health grid, no rules engine).

Premier 19 KPIs Owns alert rules Herald on breach
Open Sigma How it works
19
Metrics
14d
Throughput chart
5m
Auto-refresh
1
Owned table
// HOW IT WORKS

Five steps from a SQL one-liner to a Herald ping

Adding a metric is a _METRIC_QUERIES entry. Alerting and CSV pick it up. Do not invent a second warehouse.

01
Read others’ tables
try/except per KPI. 0 if DRA never existed.
02
See the grid
Tiles + 14-day chart + three panels. Login.
03
Define a rule
sigma_alert_rules: metric, operator, threshold.
04
Kronos evaluates
POST /api/cron/eval-rules — no session, job auth.
05
Herald fires
Telegram extra is optional/dead. CSV anytime.
Open Sigma
// TILES

Factory pulse — CSS mock

Not Datadog. Zeros are honest missing tables, not a outage skin.

meltuc.tech/sigma/app — CSS mock, not a product screenshot
tkn_cost_7d—
herald_unacked—
kronos_errors_7d—
// RULES

Thresholds you typed

gt/gte/lt/lte/eq. Evaluated on a schedule, not on this landing.

meltuc.tech/sigma/app · rules — CSS mock, not a product screenshot
Tablesigma_alert_rules
Croneval-rules
BusHerald
// vs NEXUS

KPIs + rules vs ops wall

Nexus polls /health and prefers Sigma API. Sigma is the metric registry.

meltuc.tech/sigma/ vs /nexus/ — CSS mock, not a product screenshot
Sigma19 queries + rules
Nexusno tables of its own
Ownsigma_alert_rules
// COST COPY

tkn_* is the spend lens

README still says MiniMax in places; the keys are tkn_cost_*. Do not invent a second billing product.

// WHAT IT REFUSES

No write to other apps’ tables. No fake 100s.

Will not break if GHT is missing. Public landing is not the KPI JSON.

// WHO THIS IS FOR

The owner who wants one number wall

If you needed per-blueprint drill-in, open that blueprint.

// FEATURES

Read, chart, threshold, emit

No invented APM traces.

🔢
19 KPIs
_METRIC_QUERIES map.
📉
14-day chart
DRA+BRI+RRV, zero-fill.
🧱
Three panels
Content, infra, health.
🚨
Rules
sigma_alert_rules.
⏰
Kronos eval
/api/cron/eval-rules.
📣
Herald
emit_alert on breach.
📄
CSV
All metrics.
🔌
MCP or login
Dual auth APIs.
♻️
5 min refresh
Plus manual button.
// GET STARTED

Open the wall. Zero is a valid reading.

Login for /sigma/app. This page is not live KPI JSON.

Open Sigma

Requires a MelTuc account. Create one free.

// SAID SLOWLY

What stays true over time

Cross-platform analytics — pipeline throughput, AI costs, and ecosystem health. Health stays cheap. If this page disagrees with the signed-in app, trust the app — the product contract and tests enforce that rule.