Phases
Planning / Specialist Execution / Review / Final ArtifactThese phases explain how work advances through the system.
Workflow
Runs
A run is the durable execution record for work moving through the runtime. It is where status, phases, provider usage, review pressure, messages, and artifacts come together.
Guide summary
Use this page to understand the surface before you act inside it.
Track execution phases, statuses, continue/retry paths, provider usage, and review posture.
Guide type
WorkflowThis guide reflects the current product workflow and surface ownership.
Sections
8Summary first, then steps, mistakes, and recovery notes.
Related guides
5Written against the current product structure and core execution workflow.
Main decision point
Continue vs Retry vs ReportReplay and step detail should inform that choice.
Related action
Open Mission ReplayReplay is often the fastest way to understand what actually happened.
Guide section
What it is
A run is the execution record for a mission or task moving through the runtime.
When to use it
Use runs whenever work should move through the execution engine and become a real deliverable.
Where to find it
Find them in Runs and through mission/task detail surfaces.
What happens next
You use this surface as part of the broader mission -> task -> run -> artifact workflow.
Common mistake
Reading only the status badge
Related action
Use Report this Run when the run detail surface is still unclear after Replay and troubleshooting.
Guide section
Why it matters
It preserves status, steps, artifacts, cost usage, and review posture in a durable place rather than losing that context in chat history.
Guide section
How it fits into the workflow
Runs are the center of execution. Missions define the objective, tasks define lanes, and artifacts capture the result.
Guide section
Step-by-step usage
- Launch the run from a mission or task.
- Follow phase and status transitions.
- Review specialist output and review pressure.
- Continue or retry if the runtime needs intervention.
- Open the artifact and replay once execution completes.
Guide section
Inputs and outputs
Inputs needed
Mission or task objective
Operator team
Provider posture
Relevant context
Outputs produced
Execution steps
Operator communication
Mission Replay events
Artifacts
Guide section
Common mistakes
- Reading only the status badge
- Ignoring review findings
- Retrying before the underlying cause is fixed
Guide section
Troubleshooting
Failed or stuck runs usually need Replay, step detail, provider posture, or support rather than blind retrying.
Guide section
Statuses and intervention paths
Queued
The run is waiting for runtime capacity or the next process slot.
Running
The run is currently moving through steps or background execution.
Waiting for approval
A human decision is needed before the system should continue.
Completed
The runtime reached a terminal success posture and usually produced an artifact.
Failed
A step or blocking condition stopped progress. Inspect before retrying.