OneNym.idDiscuss deployment
English

THE HARDENED STACK / Appendix C

Incident Runbook Templates

A runbook lets you make the hard decisions once, calmly, instead of for the first time in the middle of the night. This appendix provides four starting templates that IT teams adapt to their own tools and escalation paths.

This page summarises the chapter. The full text is in the book.

Principles that apply to every runbook

Preserve evidence before changing anything you do not have to, contain the spread before trying to remove the attacker, communicate to the people your plan names, and keep a timeline as you work. The runbooks are purely defensive procedures for systems you own. Store them somewhere reachable when your main systems are down, such as on paper or a separate system.

Each runbook includes a point where you decide whether the event is minor or should trigger the wider incident-communications plan. The book advises escalating when in doubt and writing a simple low/medium/high/critical scale, with a named owner per level, in advance.

Phishing and compromised accounts

For a reported phishing email, thanking the reporter is treated as a control, because people who are dismissed stop reporting. Keep the original with headers, find other recipients, remove copies in bulk, block sender and URLs, and check anyone who clicked for new forwarding rules, app consents and unusual sign-ins. If credentials were entered, run the compromised-account runbook in parallel.

For a compromised account, capture sign-in and audit logs, disable sign-in rather than delete, revoke sessions and refresh tokens, then reset the password and the MFA registrations. The book stresses that attackers often enrol their own second factor, which survives a password reset. Then look for persistence and judge the blast radius honestly, widening to the whole tenant if the account was an administrator.

Ransomware and lost devices

Suspected ransomware is a leadership-level event from the start. Isolate affected machines from the network instead of powering them off, protect and verify the immutable or offline backup copy, preserve evidence, bring in outside help early, and leave any payment decision to leadership, legal and the insurer. Recover by clean rebuild from a known-good restore point, not by restoring the latest backup, which may contain the attacker.

For a lost or stolen device, establish whether it was managed and encrypted, revoke its sessions, issue a remote lock or wipe, and reset credentials if needed. If encryption cannot be confirmed, treat the data as exposed and follow the breach-assessment process. Every incident, however small, ends with a short blameless review.

KEY POINTS

Key points

  • Store runbooks where you can reach them when the main network is down, such as on paper.
  • Write a low/medium/high/critical severity scale with a named owner for each level.
  • Make resetting MFA registrations a mandatory step in every account-compromise response.
  • Confirm today that you have an immutable or offline backup copy and have restored from it.
  • Keep an incident timeline during the work, not reconstructed afterwards.
  • Hold a short blameless review after every real incident and make one improvement.

QUESTIONS

Common questions

What should IT do when a user reports a phishing email?
Thank the reporter, keep the original message with full headers, and find out whether they clicked, opened an attachment or entered credentials. Search for other recipients, remove all copies, block the sender, domain and URLs, and check anyone who interacted for new forwarding rules, app consents or unusual sign-ins. If credentials were entered, treat it as a compromised account at the same time.
Is resetting the password enough after an account compromise?
No. Attackers frequently register their own MFA method early, and that second factor survives a password reset. Revoke all sessions and refresh tokens, reset both the password and the MFA registrations, and look for persistence such as new inbox or forwarding rules, OAuth consents, delegate permissions and changed recovery details.
Should you shut down computers hit by ransomware?
The Hardened Stack advises isolating affected machines from the network rather than powering them off, unless a specialist advises otherwise. Disconnecting stops the spread, while shutting down can destroy memory and disk state that help the investigation and, with some ransomware, keys that might aid recovery.
Why not restore immediately from the latest backup after ransomware?
If the latest backup was taken after the attacker was already inside, restoring it brings the attacker back along with the data. Recovery should follow a clean rebuild and a known-good restore point, with credentials reset broadly and systems brought back in a planned order so cleaned systems are not reinfected.
What should happen when a company laptop is lost or stolen?
Gather the facts, check whether the device was managed and encrypted, and revoke its sessions and tokens. Issue a remote lock and, if recovery is unlikely and the device holds sensitive data, a remote wipe. If full-disk encryption cannot be confirmed, treat the data as exposed and follow the data-breach assessment process, which may involve legal reporting duties.