Skip to main content
Version: Next

Incident flow

Production problems are tracked as incidents attached to the task they relate to — not as task statuses. Each task carries an incident log (incidents.json side file), and incidents have their own status machine.

Incident lifecycle

reported → investigating → production-fix → fixed → verified → closed
StatusMeaning
reportedIncident created and logged.
investigatingRoot cause being diagnosed.
production-fixA fix is being applied to production.
fixedFix applied.
verifiedFix confirmed to resolve the incident.
closedIncident fully resolved and closed.

Agent

The /task-incident agent handles incidents: investigate, apply a production fix, and document the root cause. Production problems are redirected here rather than being filed as new tasks.

CLI commands

insight-flow incident-create --id Nxx --title "..." --severity critical|high|medium|low
insight-flow incident-status --id Nxx --incident <iid> --status investigating|production-fix|verified|closed
insight-flow incident-resolve --id Nxx --incident <iid> --rootCause "..." --fix "..."
insight-flow incident-list --id Nxx

See the CLI: Incidents group for full flags. A task's open incident count surfaces on its dashboard card via the openIncidentCount summary field.