Guides2 min read

Runbooks

Author reusable workflows, bind them to services, respond to incidents, and manage execution infrastructure.

Beta — Runbooks is Beta in v2.1. Use a published, compatible server/Agent pair and qualify your deployment and target permissions before enabling automatic writes. Passing configuration checks or a preview does not prove recovery.

Read Runbooks and safe remediation for the architecture, risk model, and safety guarantees before enabling writes.

Guides

# Guide Audience What you'll learn
1 Author and publish Authors, Admins Create drafts, configure all 9 step types, define typed inputs, manage prechecks and verification, publish immutable versions
2 Bind and respond Admins, Responders Attach workflows to services, configure triggers, preview and approve exact plans, handle cancellation and uncertain outcomes
3 Agents, pools, and secrets Admins, Operators Enroll Agents, choose pool semantics, grant and rotate scoped secrets, monitor fleet health
4 Permissions and security Admins All 9 capabilities, API scopes, service authorization, artifact access
5 Action examples Authors, Operators Kubernetes recovery, Docker inspection, systemd checks, HTTP verification, static Bash commands
6 Health and monitoring Operators Health dashboard, circuit breaker, remediation budgets, Prometheus alerts, worker evidence
7 Service bindings and triggers Admins Automatic activation, trigger conditions, concurrent edits, rule lifecycle

Reference and operations

  • Runbook reference — Step types, input types, risk classes, permissions, limits, state machines, and maintenance.
  • Agent operations — Installation, deployment topologies, and upgrade procedures.
  • Enterprise qualification — Correctness, physical target, fault, and load receipts.
  • Troubleshooting — Waiting executions, unavailable Agents, uncertain outcomes, and common errors.

Choose your starting point

If you are a… Start with…
Responder new to Runbooks Read the concepts, then Bind and respond for incident execution, exact-plan approval, detached suggestions, and uncertain outcomes
Author creating your first workflow Author and publish — start with a read-only diagnostics workflow
Administrator setting up infrastructure Agents, pools, and secrets → Permissions → Service bindings
Operator deploying Agents Agent operations → Health → Enterprise qualification
API integrator v2 API reference — bound executions use saved inputs; request parameters cannot override them

Begin with diagnostics and a manual binding before introducing approved writes or automatic triggers.

Last updated for v2.1.0

Edit this page on GitHub