Skip to Content
ModulesRunbooks

Runbooks

Runbooks are structured remediation playbooks that define step-by-step actions to resolve specific types of incidents. AtlasAI can generate runbooks from AI analysis, and optionally execute them automatically based on your trust level.

Key Features

  • AI-generated runbooks — The reasoning engine creates runbooks based on RCA results and historical resolutions
  • Manual authoring — Build runbooks with a visual step editor, including conditional branches and approval gates
  • Step types — Shell commands, API calls, Kubernetes actions, cloud provider operations, and human approval steps
  • Risk classification — Each step is tagged Low, Medium, or High risk with rollback instructions
  • Version control — Full version history with diff view and rollback capability
  • Execution engine — Run runbooks on Edge Agents with real-time output streaming
  • Approval workflows — Require manual approval for high-risk steps while auto-executing safe ones
  • Runbook library — Searchable catalog of proven runbooks organized by category and service

How to Access

Navigate to Runbooks in the left sidebar. The library view shows all saved runbooks. To generate a runbook, click Generate Runbook from an RCA result inside an incident.

Basic Usage

  1. Navigate to Runbooks → Library
  2. Click New Runbook to create manually, or generate one from an incident’s RCA
  3. Add steps to the runbook — each step requires a type, target, command, and risk level
  4. Add approval gates before high-risk steps (e.g., “Requires SRE approval before database failover”)
  5. Test the runbook against a staging environment using Dry Run
  6. Save and publish to the library
  7. When an incident matches, the runbook can be suggested automatically or triggered manually

See also

RCA LabAutomation