Skip to Content
Cockpit

Cockpit

Cockpit is your analytics hub — a single place to see how your agent fleet is performing, how the platform is getting smarter over time, and how code quality is trending across your repositories. Access via Operate → Cockpit in the sidebar.

Cockpit is organized into three tabs.

Fleet

The Fleet tab (/cockpit/fleet) is your operational command center — a real-time view of every agent in your workspace.

  • KPI cards — at-a-glance numbers: fleet size, agents active right now, runs in the last 24 hours, pass rate, overall fleet health, average cost per run, and cost today.
  • Run volume chart — a live bar chart of throughput over the last 30 minutes, so you can see activity spikes as they happen.
  • Outcome breakdown — a stacked bar showing how runs are distributed: completed, running, queued, and halted.
  • Fleet agents table — every agent with its status, run count, average duration, cost, and trust score. Sort and filter to find the agent you care about.

Compounding

The Compounding tab (/cockpit/compounding) shows the platform’s “flywheel” — the idea that every agent run makes the next one better, cheaper, and more reliable.

  • Platform IQ — a headline score with its 30-day trend, summarizing how much your workspace has learned.
  • IQ composition — a breakdown of what’s driving that score (knowledge, reliability, and more).
  • Compounding signals — six metrics tracking the flywheel: run velocity, Cortex knowledge growth, agent reliability, workflow patterns discovered, routing precision, and cost per run.
  • Compounding gain — month-over-month cost tracking that shows how much cheaper your runs have gotten as the platform learns.
  • The flywheel — a visual loop showing how agent runs feed knowledge back into the system: Run → Trajectories → Cortex atoms → Better routing → next Run.

Review

The Review tab (/cockpit/review) brings code-review intelligence into Cockpit — a roll-up of pull-request analysis across all your repositories. It’s the same engine as the dedicated Review surface, summarized here so you can spot quality trends at the fleet level: repo health, open vs. merged PRs, average risk scores, and finding counts.

See also: Review, Cortex, Agents, Cendra

Last updated on