AI TRACE

AI TRACE visualizes the structured execution path of STRATUM's AI harness. It follows observable Mission activity from user intent through Skills, specialist Roles, planning and execution stages so an operator can see which parts of the configured AI Operator are participating in the current Mission.

The trace is built from structured tool and Mission events. It is designed for operational explainability and does not expose private model reasoning.

AI Studio defines the reusable Operator DAG: Roles, Skills, context, guardrails, policies and execution stages that shape STRATUM AI behavior. AI TRACE shows how that configured operator participates in a real Mission. A Skill can become active because it applies to the requested outcome, a specialist Role can contribute recommendations or constraints, and the Planner can consume those inputs while Mission Control reports the current planning or execution stage.

The connections in TRACE show the active path through the Operator for the Mission being planned or executed, connecting the reusable AI Studio DAG to actual operational work. The operator can therefore answer which Skills, Roles, policies or execution stages influenced the Mission without inspecting hidden prompt text.

The TRACE header reports structured tool events and can focus the graph on the active path. Mission Control provides the detailed timeline beside it, including the current phase, planning activity and concrete tool events such as guest-status discovery or infrastructure reads.

These views complement each other. Mission Control shows what the Mission is doing now, while AI TRACE shows how the configured STRATUM harness and Operator DAG are participating in that work.

The visualization is driven by structured tool and Mission events, so changes in the trace correspond to observable execution activity. That makes the graph useful for understanding tools used, systems observed, Skills selected, specialist Roles involved, policies evaluated, execution stages reached and evidence returned.

For production infrastructure, this creates explainability at the layer operators can act on. The trace provides a visual record of the harness path while the Mission engine, authorization, policy and verification systems remain the authorities for what can execute and what counts as success.