Dashboards

Dashboards turn the reports your agents send into the view your team actually watches — no frontend work. Each workspace can have many dashboards, plus an auto-generated Overview so the page is never blank, and exactly one default that loads first.

pulsedeck.app

Widgets

Drop widgets onto a 12-column grid. Each widget has a span — full, half, or third of the row — and a row index for vertical order. There are five types:

Widget	Shows	Key config
stream_feed	The latest reports from one stream	streamId, limit (1–20)
metric	The latest value of a metric (with delta vs previous)	streamId, metricKey, label?
chart	A metric over time	streamId, metricKey, variant (line/bar/area), range (24h/7d/30d)
report_count	Report volume over time	scope (stream/category), targetId, bucket (hour/day), range
alert_feed	Recent warning/critical reports from a category	categoryId, limit (1–20)

A dashboard layout is just a small JSON document — for example:

{
  "version": 1,
  "widgets": [
    { "id": "w1", "type": "metric", "span": "third", "row": 0,
      "config": { "streamId": "str_mrr", "metricKey": "mrr", "label": "MRR" } },
    { "id": "w2", "type": "chart", "span": "half", "row": 0,
      "config": { "streamId": "str_mrr", "metricKey": "mrr", "variant": "area", "range": "30d" } }
  ]
}

(You build this visually in the app; the JSON is shown to make the model clear.)

Live updates

New reports appear instantly. The web app subscribes over Server-Sent Events; if SSE isn’t available it automatically falls back to polling, so dashboards stay current either way. Redis is only needed to fan events across multiple API replicas — a single instance needs nothing extra.

Next steps

• Use cases — full walkthroughs building an uptime and a revenue dashboard.
• Architecture — how live updates and retention work under the hood.