Getting Started
Learn what IncidentFlow does and how the first investigation flow works.
IncidentFlow helps teams move from noisy operational signals to useful incident context.
It ingests and correlates signals from systems such as Kubernetes, Grafana, Slack, Argo CD, and status providers, then exposes that context through the web app and MCP tools.
Core Concepts
Incidents
An incident is a grouped operational problem with normalized context, related signals, status, and investigation artifacts.
Integrations
Integrations connect IncidentFlow to external systems. Each integration is authorized per workspace and access is enforced by the platform API.
MCP Tools
MCP tools let AI clients ask IncidentFlow for operational context. Tools are registered with read/write metadata so clients can distinguish read-only diagnostics from memory-writing actions.
Workspace Memory
Workspace memory stores private incident knowledge such as runbooks, RCAs, postmortems, and service notes. Every workspace memory operation is scoped to the authenticated workspace.
First Flow
- Connect a workspace.
- Add at least one integration, such as Kubernetes or Grafana.
- Install or authorize the IncidentFlow MCP server.
- Ask an AI client to inspect current incident context.
- Save useful runbooks or RCAs back into workspace memory when appropriate.