A critical vulnerability lands in the KEV catalogue on a Friday evening, the vendor ships an out-of-band fix, and the routine monthly patch cycle is suddenly useless. Most organisations have a patch policy, but few have a rehearsed procedure for the 24 to 72 hours after an emergency release, when someone has to decide whether to patch now, what testing can be cut, who approves the change and how to prove the fix took effect. This playbook covers that operation. It is written for vulnerability management leads, infrastructure and endpoint teams, change managers and the incident lead on call. Investigating whether the flaw was already exploited is covered in the Zero-Day Exploitation Playbook; this page is about getting the patch or mitigation in safely and quickly.
How to use this template
- Adapt before you need it: replace every [Organisation] placeholder, name the roles, and agree the trigger thresholds with the change advisory board (CAB) and the risk owner in advance. A playbook first read during an emergency is a document, not a capability.
- Link it to your maintenance plans: NIST SP 800-40 Rev. 4 recommends a maintenance plan per maintenance group for each risk response scenario, including emergency patching and emergency mitigation. This template is the emergency plan; your Patch Management Policy Template holds the routine one.
- Keep sections numbered: the numbers let the incident log refer to "section 4, mitigation applied" without ambiguity.
- Rehearse twice a year: run a tabletop using a recent real emergency release and time each step.
1. Scope and triggers
1.1 Scope. This playbook applies to all [Organisation] assets, including internet-facing appliances (VPN, firewall, load balancer, email gateway), servers, endpoints, cloud services under our control, operational technology where named in annex A, and the management tools that patch them.
1.2 Triggers. The playbook is invoked when any of the following is true and the affected product is present in our estate:
| Trigger | Example source | Default response |
|---|---|---|
| Vulnerability added to the CISA Known Exploited Vulnerabilities (KEV) catalogue affecting an internet-facing asset | CISA KEV feed | Emergency patch or mitigation, target 48 hours |
| Vendor out-of-band security release with stated in-the-wild exploitation | Vendor advisory or PSIRT bulletin | Emergency patch, target 48 hours for exposed assets, 7 days internal |
| Government emergency directive or national CERT alert naming the product | CISA emergency directive, NCSC alert | Follow the directive's actions and deadlines where stricter |
| Public proof-of-concept for a critical, pre-authentication flaw in an exposed product | Threat intelligence, vendor update | Emergency mitigation within 24 hours, patch within 72 hours |
| Evidence the flaw has been exploited in our own estate | SOC detection | Invoke the incident response plan; this playbook runs in parallel for the patch |
1.3 Not in scope. Routine monthly releases (handled by the Patch Tuesday process) and critical vulnerabilities without exploitation evidence that can wait for the next accelerated cycle. For reference, CISA BOD 22-01 sets two weeks for federal agencies to remediate KEV entries with CVE IDs from 2021 onwards; [Organisation] targets are shorter for exposed assets because mass exploitation typically starts within days of a public proof of concept.
2. Roles and contacts
| Role | Responsibility | Named person and deputy |
|---|---|---|
| Emergency patch lead | Owns the operation end to end; chairs the decision call; keeps the decision log | [Name], [Deputy] |
| Risk owner (senior) | Accepts residual risk where patching is deferred; approves taking a business service offline | [Name], [Deputy] |
| Emergency CAB approver | Approves the emergency change, or approves retrospectively within the agreed window | [Name], [Deputy] |
| Platform owners | Apply patches or mitigations for their maintenance group (network edge, Windows servers, Linux, endpoints, databases, SaaS admin) | [Names] |
| SOC lead | Raises detection for exploitation attempts; hunts on affected assets before and after patching | [Name] |
| Service owners | Confirm business impact and run smoke tests after patching | [Names per service] |
| Communications lead | Internal notices, customer and supplier communication | [Name] |
| Vendor contacts | Support case escalation, hotfix access | [Vendor, contract number, support line] |
Keep this table in the out-of-band location used by the Incident Response Plan Template, because the collaboration platform may itself be the affected product.
3. Detection and triage
The aim of triage is a decision within two hours of the trigger: patch now, mitigate now and patch later, or accept until the next window.
- Confirm the facts: read the vendor advisory itself, not a news summary. Record CVE IDs, affected versions and builds, fixed versions, whether exploitation is confirmed, whether a mitigation exists, and whether the fix needs a reboot or service restart.
- Find exposure: query the asset inventory and software bill of materials for affected products and versions; check the external attack surface for internet-facing instances; ask cloud and SaaS owners about managed services. Unknown versions count as vulnerable.
- Assign maintenance groups: sort affected assets by exposure (internet-facing first), privilege (identity providers, management planes, hypervisors) and business criticality.
- Check for exploitation: ask the SOC to search for published indicators and the vendor's detection guidance before patching, because patching can overwrite evidence. If any hit is found, escalate to the incident response plan and preserve images before changing the system.
- Decide: use the decision table below and record the result in the decision log with time and approver.
| Condition | Decision | Approval |
|---|---|---|
| Exploited, exposed asset, patch available, low regression risk | Patch now with abbreviated testing (section 4.2) | Emergency CAB approver |
| Exploited, exposed asset, patch high risk or not yet available | Apply vendor mitigation or remove exposure now; patch when ready | Emergency CAB approver |
| Exploited, no mitigation, cannot patch safely | Disconnect or restrict access to the service | Risk owner |
| Not exploited, internal only, compensating controls in place | Accelerated normal change within 7 days | Standard CAB |
| Asset end of life with no fix | Isolate and plan removal | Risk owner, recorded in risk register |
For prioritising across several simultaneous releases, combine KEV status with an exploitation probability score and a decision model. See the EPSS Guide and the SSVC Guide.
4. Containment
In this playbook containment means reducing exposure before and while the patch is applied.
4.1 Emergency mitigation. Apply the vendor's documented workaround where one exists: disabling a vulnerable feature, importing a mitigation configuration, a web application firewall rule, or restricting the management interface to an administrative network. NIST SP 800-40 Rev. 4 treats emergency mitigation as a separate scenario and says temporary mitigations should be replaced by permanent fixes on enforced schedules. Record each mitigation with an expiry date. During the 2024 Ivanti Connect Secure response, CISA ED 24-01 required agencies to import a vendor mitigation file and run the vendor's external integrity checker before patches existed, and warned that the mitigation degraded some management features: test mitigations as carefully as patches.
4.2 Abbreviated testing. SP 800-40 Rev. 4 describes emergency patching as the routine approach on an accelerated schedule with a brief canary check first. Agree in advance which tests can be cut:
| Routine step | Emergency version | Never skip |
|---|---|---|
| Full regression test in staging | Smoke test on one representative asset per maintenance group | Service starts and core transaction works |
| Pilot ring of several days | Canary of 1 to 5 percent for 1 to 4 hours | Canary health checked by a named person |
| Change window outside business hours | Next available window, or immediately for exposed assets | Service owner informed |
| Full CAB review | Emergency CAB approver, documented in the ticket | Retrospective CAB review within [5] working days |
| Backup before change | Snapshot or configuration export | A tested rollback path |
4.3 Rollback plan. Before each group is patched, record the rollback method (snapshot revert, previous firmware partition, package downgrade, configuration restore) and who can authorise it. Rolling back reintroduces the vulnerability, so a rollback must be paired with the mitigation in 4.1 or with disconnection.
5. Eradication
- Patch in order: internet-facing and identity-critical assets first, then internal servers, then endpoints. Use the patch tools' targeted deployment rather than waiting for the normal schedule.
- Restart where required: a patch that needs a service restart or reboot has not taken effect until that happens.
- Invalidate what the flaw could have exposed: patching stops new exploitation but does not undo theft that already happened. After CitrixBleed, Citrix advised killing all active sessions after upgrading because stolen session cookies stayed valid, and Xfinity was accessed after it had patched. After ToolShell, stolen ASP.NET machine keys gave attackers continued access until the keys were rotated and IIS restarted.
- Rotate machine credentials on exposed hosts: session tokens, API keys, service account passwords, certificates and signing keys held by or stored on the affected product. ED 24-01 required agencies to revoke and reissue certificates, keys and passwords, including stored API keys and service account passwords. See the Leaked Credential Response Playbook.
- Remove temporary mitigations only after verification in section 6, following the vendor's removal instructions.
6. Recovery
6.1 Verify the fix. SP 800-40 Rev. 4 separates verification (the patch is installed and has taken effect) from monitoring (the patch stays installed). Verify with at least two methods:
- Version evidence: the build number reported by the product, the package manager or endpoint management inventory.
- Authenticated scan: rescan affected assets with the scanner's check for the CVE; authenticated scans find more than unauthenticated ones.
- External check: for internet-facing assets, rescan from outside to confirm no vulnerable instance remains, including forgotten ones.
- Integrity check: where the vendor provides an integrity or compromise-assessment tool, run it after patching.
6.2 Monitor for regression. Keep a watch for 30 days for assets that return to a vulnerable build through snapshot restore, rebuild from an old image, or factory reset, which SP 800-40 Rev. 4 calls out explicitly. Update golden images and templates the same day.
6.3 Close exceptions. Any asset not patched gets an exception with owner, compensating control and expiry, recorded in the risk register. NCSC guidance is clear that a decision not to update is a senior-level risk decision. See the Remediation SLAs and Vulnerability Exceptions Guide.
7. Communication and notification
- Internal start notice: within one hour of the decision, tell service owners and the service desk what is changing, expected disruption and who to call.
- Status cadence: update stakeholders every [4] hours with assets patched, mitigated and outstanding, as counts and percentages per maintenance group.
- Customers and suppliers: if customer-facing services are affected or customers ask, use pre-approved wording that states the action taken without speculating about compromise.
- Regulatory: patching on its own is not normally a notifiable event, but evidence of exploitation may be. Hand over to the incident response plan's notification procedure if section 3 found indicators.
- Directives: where [Organisation] is subject to a government directive, file inventory and status reports on the directive's schedule. ED 24-01, for example, required a full inventory of affected instances within a week.
8. Post-incident review
Within [10] working days, the emergency patch lead runs a review and records:
- Timeline: trigger received, decision made, mitigation applied, first and last asset patched, verification complete.
- Metrics: time to decision; time to mitigate exposed assets; percentage of affected assets patched within target; mean and median time to patch by maintenance group (the measures SP 800-40 Rev. 4 Table 1 uses); number of rollbacks; assets found during the emergency that were missing from inventory.
- What slowed us: inventory gaps, unclear ownership, approval delays, vendor access, test environments.
- Actions: owners and dates, including asking vendors at renewal about emergency mitigations and rollback, as SP 800-40 Rev. 4 suggests for procurement.
Implementation guidance
Framework mapping. The playbook supports the following controls (confirm the mapping in your own control library):
| Framework | Reference | How the playbook supports it |
|---|---|---|
| NIST CSF 2.0 | ID.RA-01, PR.PS-02, RS.MI-02 | Identify and record vulnerabilities; maintain software commensurate with risk; eradicate where exploitation is found |
| NIST SP 800-53 Rev. 5 | SI-2, SI-2(2), SI-2(3), CM-3, CM-4, RA-5 | Flaw remediation, automated status, time-to-remediate benchmarks, change control and impact analysis, scanning for verification |
| CIS Controls v8.1 | 7.3, 7.4, 7.7 | Automated OS and application patch management; remediate detected vulnerabilities |
| ISO/IEC 27001:2022 Annex A | A.8.8, A.8.32 | Management of technical vulnerabilities; change management |
| MITRE ATT&CK | T1190, T1133 | The techniques emergency releases most often address: exploitation of public-facing applications and external remote services |
Rollout.
- Build maintenance groups and name owners for each (week 1 to 2).
- Agree triggers, targets and the abbreviated testing table with the CAB and risk owner (week 2 to 3).
- Pre-stage the tooling: targeted deployment collections, canary groups, emergency change ticket template, scanner checks (week 3 to 4).
- Run a tabletop with a real past release such as ToolShell, and measure time to decision (week 5).
- Run a live drill on a low-risk out-of-band update and measure end-to-end time to verified fix (quarter 2).
Small organisations can merge the patch lead and CAB approver roles, rely on vendor auto-update for appliances and SaaS, and use a managed service provider's emergency process, provided the contract states response times. Large organisations should automate exposure lookup from the KEV feed into the asset inventory so triage starts with a list, not a search.
Common failure patterns. Patching without invalidating sessions and keys; trusting a scanner that cannot see the appliance; forgetting high-availability peers and disaster recovery instances; leaving temporary mitigations in place indefinitely; and restoring an old snapshot weeks later. The Zero-Day and Exploited Vulnerabilities Guide and Patch Management Guide cover the wider programme.
Related NHI Mgmt Group resources: Zero-Day Exploitation Playbook · CISA KEV Catalogue Guide · Patch Management Guide · Vulnerability Prioritisation Guide