Runtime Operations

The dashboard has several runtime-facing operator surfaces even though execution happens elsewhere.

This is an important product boundary. The runtime is a separate system, but the platform still owns the operator controls that define how that execution system should behave.

Main Runtime-Facing Pages

Environments for execution-environment definitions
Agentic Settings for runtime defaults, safeguards, and execution posture
Platform Settings for control-plane timing, supervision, and fleet-side operations
runtime build history and customization surfaces where enabled

What Those Pages Actually Cover

The current dashboard and platform API expose concrete controls around:

execution image and environment posture
live visibility defaults
prompt warning thresholds
agent throughput and backlog posture
transport and timeout settings
platform attachment and reporting behavior
workspace setup and snapshot handling
result capture and publishing
secret-provider access and subagent limits

That is a meaningful operator surface, not just an afterthought settings page.

Why It Matters

These controls let operators shape runtime behavior without rebuilding the whole product or editing every workflow by hand. The platform owns the policy and defaults. The runtime later enforces them during execution.

How It Connects To The Rest Of The System

Runtime Operations is the umbrella view over several more specific runtime-facing surfaces:

Environments defines where specialist work runs
Agentic Settings defines how specialists generally behave while running
Platform Settings defines how the control plane supervises that execution
Live Containers and Live Logs provide evidence about how those settings behave in practice

That is why this page matters. It is the operator bridge between configuration and observed execution.