Skip to content
Home › Studio › Overview

Overview

Studio has two top-level entry surfaces — the Landing (/) and the Overview dashboard (/overview). The landing keeps the workspace calm and oriented; the dashboard surfaces the numbers a maintainer wants at a glance.

Landing (/)

The landing page is what you see immediately after /studio opens the browser. It is intentionally low-density:

  • Status pills — active LLM profile, configured DB profile count, and any pending review queue. Each pill links to the page that resolves it, and renders a warning chip when the underlying state is missing (no LLM configured, zero DB profiles, etc.).
  • Action grid — four cards: Browse, New run, Ask, Audit. Each card greys out with a reason text when its preconditions aren't met (e.g. "No DB profile configured" on Browse and New run).
  • Recent activity — a five-row excerpt of the most recent runs across every command (run, run-apply, generate, ask, schedule), each row navigable to its Run detail page.
  • Footer link to the Overview dashboard for token + cost detail.

The landing replaces a previous behaviour where / redirected to the dashboard immediately; the dashboard is still one click away, but a fresh session starts in a quieter place.

Overview dashboard (/overview)

The dashboard is denser — it answers "how much have I used AMX, and is it working?"

Six stat cards (3-column grid on md: and up, stacks on mobile):

Card What it shows
LLM model Active model name. Click → Settings → LLM
Total runs All-time count across every command kind. Click → Runs list
Success rate Percentage of runs that finished cleanly. Click → Runs list
Input tokens All-time input tokens summed across every recorded run
Output tokens All-time output tokens summed across every recorded run
Total cost (USD) All-time USD spend (frozen at run time, never re-priced)

Cost-card click-throughs land on the System → Token usage section where the same totals can be windowed (Today / 24 h / 7 d / 30 d / All time) and broken out by (provider, model).

Below the stat cards, a Recent runs card lists the eight most recent runs with an activity icon, run ID, command, scope summary, DB profile, model, duration, relative timestamp, and a status pill. Every row is clickable to the corresponding Run detail page.

Empty states

A fresh install lands on a landing with three placeholder cards:

The dashboard mirrors this — every stat card shows — until at least one run has landed, and the Recent runs card renders an icon + "No runs yet" line.