Sigma is the analytics layer for the entire MelTuc ecosystem. It aggregates 19 live KPIs from across every active blueprint — DRA research output, BRI idea velocity, AI cost burn, Kronos job success rates, SRM ticket volume, Herald alert count, GHT repo tracking, and more — into a single dashboard that auto-refreshes every 5 minutes. Set threshold alert rules on any metric and Sigma notifies you via Telegram and Herald the moment a number crosses your line.
The centrepiece is a 14-day stacked bar chart showing DRA, BRI, and RRV pipeline throughput per day — giving you an instant visual read on whether content generation is accelerating, stalling, or idle. Supporting stat panels break out AI cost tracking (7d and 30d), platform health indicators, and content pipeline depth at a glance.
Sigma queries a dozen existing blueprint tables every 5 minutes and surfaces 19 live KPIs — with zero data duplication. Its only table of its own holds your alert rules.
/sigma/api/platform-metrics.
The top section of the Sigma dashboard is a 4-column KPI grid of headline metrics.
Each tile shows a label and a live number — refreshed on page load and every 5 minutes
automatically. Teal values are content/throughput metrics. Amber values are AI cost numbers.
Red values are error/failure counts. All values come from a single
GET /sigma/api/platform-metrics
call that runs 19 separate SQL queries in sequence, each individually wrapped in try/except
so a missing table never breaks the rest of the dashboard.
Below the KPI grid, a Chart.js bar chart shows per-day DRA, BRI, and RRV output over the
trailing 14 days. The chart uses a
generate_series query to fill in
zero values for days with no output — so a flat day is clearly visible rather than absent.
Teal bars are DRA research reports, amber bars are BRI ideas, and blue bars are RRV blog posts.
The legend, axis ticks, and gridlines all use the platform color system. The chart redraws
cleanly on every refresh without page reload.
Below the chart, three key-value panels drill deeper into the three dimensions of platform health. The Content Pipeline panel shows 30-day DRA output, RRV drafts waiting to publish, and PRL prompt library size. The AI Infrastructure panel shows FMR active model count alongside 7d and 30d Claude API cost. The Platform Health panel computes the Kronos job success rate from the 7-day run and error counts, and shows open SRM tickets and Herald unacked alerts.
Sigma reads from a dozen existing blueprint tables and owns only one table of its own (alert rules). Every query is individually wrapped in try/except, so a missing table degrades gracefully without breaking the dashboard.
tkn_turns.cost_usd over 7-day and 30-day windows — always current, never stale.kronos_runs in the last 7 days. Job success rate computed automatically.consumed_at IS NULL — your newsletter backlog depth at a glance.metric_key + operator (>, >=, <, <=, ==) + threshold. Kronos evaluates them on a schedule and fires a Telegram message plus a Herald alert the instant a threshold is breached.?format=csv, or query the same data programmatically — metric and alert-rule endpoints accept a SWIP session or an MCP key.19 live KPIs. 14-day throughput chart. AI cost burn. Threshold alerts. All in one dashboard.
Requires a MelTuc account. Create one free.