Live

Live is the ambient monitoring surface for Colony Cloud. Route: /dashboard/pipeline-live

It gives operators a continuous, hands-off window into everything the autonomous pipeline is doing right now — without requiring any interaction. Open it on a secondary screen or a wall display to stay passively aware of fleet health.

What Live shows

Live composes six sections rendered top-to-bottom on a single scrollable page:

Section	What it shows
Live indicator	Connection status dot and a retry control. Green when the SSE stream is active; amber when falling back to HTTP polling; grey when disconnected.
LLM health banner	A dismissible banner that appears when any upstream model provider reports degraded availability. Disappears automatically when the provider recovers.
Feature hints	Contextual tips surfaced when the pipeline detects a configuration gap — for example, no Workers registered, or no repositories enabled. Disappears once the condition resolves.
Digest	A rolling summary of recent pipeline activity: issues completed, state changes, and any DLQ entries in the current session window.
Budget overview	Current org LLM spend for the rolling 24-hour window, against any configured daily cap. Turns amber as spend approaches the limit.
Pipeline live view	The main content area: a KPI row (Active, In flight, Blocked, Completed today), one workflow-diagram card per active workflow showing per-state issue counts, an Agent activity list of currently active issues, and a Recent transitions strip.

How Live gets its data

Live subscribes to the shared pipeline-state stream via PipelineStateContext. The connection is a server-sent event (SSE) stream; if the SSE connection drops, the context falls back to an HTTP poll on a 3-second cadence. The live indicator in the page header reflects the current connection mode. No manual refresh is needed — the view updates automatically whenever the pipeline state changes.

The data scope follows the org-wide repo selector in the sidebar. If you have scoped the selector to a specific repository, Live shows only that repository’s issues.

How Live differs from Pipeline

Both surfaces show pipeline activity, but they answer different questions:

	Pipeline	Live
Route	/dashboard/pipeline	/dashboard/pipeline-live
Mental model	Swim-lane board — what is each repo working on right now?	Fleet dashboard — KPIs, per-workflow diagrams, and a live activity feed
Layout	One swim lane per repository, issues as cards at their current station	KPI row + workflow diagram cards + Agent activity list + Recent transitions strip
Primary question	”What is actively flowing through my repos?"	"How many issues are active, in flight, or blocked right now?”
Best for	Repo-level daily monitoring	Fleet-level health monitoring; wall display ambient view

Tip

Pipeline is the default landing surface after sign-in. Live is a complement — use it when you want a fleet-wide health snapshot or need to monitor agent activity across all active workflows.

For the full tour of Pipeline, Insights, and Inbox, see Dashboards.