Markdown Editor

Vulnerability Management Policy Template (CVSS 4.0, EPSS, KEV, SSVC)

Enterprise vulnerability policy: asset coverage, scan cadence, CVSS 4.0, EPSS, KEV and SSVC prioritisation, BOD 26-04 factors, SLAs, exceptions, metrics

Sicherheits-EngineeringVollständigCVSS v4.0; EPSS; CISA KEV; CISA SSVC guide; CISA BOD 26-04; NIST SP 800-40 Rev. 4vulnerabilitymanagementpolicy

Anleitung: Use it when you set how your organisation finds, prioritises and fixes vulnerabilities in systems it runs. Fill asset tiers and the prioritisation model first, then the SLA matrix. The main pitfall is SLAs driven only by CVSS base scores instead of exploitation evidence and exposure.

Vorschau

Vulnerability Management Policy: [Organisation name]

Enterprise policy for finding, prioritising, remediating and reporting vulnerabilities in the technology an organisation runs: scope and asset coverage, scanning cadence, risk-based prioritisation, remediation SLAs, exceptions and risk acceptance, metrics and reporting. Prioritisation combines FIRST CVSS v4.0 (November 2023), the FIRST Exploit Prediction Scoring System (EPSS), the CISA Known Exploited Vulnerabilities (KEV) catalog and a decision model in the style of the CISA Stakeholder-Specific Vulnerability Categorization Guide (November 2022: Track, Track*, Attend, Act). It draws on NIST SP 800-40 Rev. 4 Guide to Enterprise Patch Management Planning (April 2022) and describes the factors of CISA Binding Operational Directive 26-04 (June 2026) as a reference model. This is the consumer-side policy; product security response (producer side) is covered by PSIRT Vulnerability Handling (security/psirt-vulnerability-handling) and EU CRA duties by CRA Vulnerability Handling and Reporting Procedure (security/cra-vulnerability-handling). Replace every [bracketed] value and every row marked Example. Worked examples use Corvid Commerce, a fictional SaaS company. This is a documentation format, not legal advice.

Document control

FieldValue
Document ID[SEC-POL-VULN-001]
Version[1.0]
Status[Draft / In review / Approved / Superseded]
Owner[@vulnerability-management-lead]
Approvers / reviewers[@ciso], [@cio], [@head-of-platform], [@head-of-it-operations], [@risk-manager]
Last updatedYYYY-MM-DD
ClassificationInternal
Effective dateYYYY-MM-DD
Parent policy[Information security policy, section [x]]
Obligations[Example: customer contracts, PCI DSS for the payment zone, NIS2 national measures, US federal customers referencing CISA directives]
Review cycleAnnually, and when a directive or major threat change requires it
Next reviewYYYY-MM-DD

Approvals

RoleNameDecisionDate
Chief information security officer[@name][Approved / Changes requested]YYYY-MM-DD
Chief information officer[@name][Decision]YYYY-MM-DD
Head of platform engineering[@name][Decision]YYYY-MM-DD
Enterprise risk manager[@name][Decision]YYYY-MM-DD

Revision history

VersionDateAuthorChange summary
0.1YYYY-MM-DD[@name]First draft
0.9YYYY-MM-DD[@name]Example: SLA matrix tested against last two quarters of data
1.0YYYY-MM-DD[@name]Example: Approved; moved from CVSS-only SLAs to decision-based SLAs

Contents

1. Purpose

This policy sets how [Organisation name] identifies vulnerabilities in the systems it operates, decides which to fix first, fixes or mitigates them within defined times, and accepts risk when it cannot. Its objectives are:

  1. Reduce the window in which known exploited vulnerabilities are reachable on [Organisation name] systems.
  2. Spend remediation effort where exploitation is likely and impact is high, not where scores are merely high.
  3. Make every unremediated vulnerability past its SLA a visible, owned and time-limited decision.
  4. Give leaders and customers reliable evidence of vulnerability exposure and trend.

2. Scope

2.1 In scope

  • All hardware, operating systems, firmware, applications, containers, cloud services, network and security appliances, and third-party components that [Organisation name] owns, operates or configures, in all environments that hold production data or connect to production.
  • Open source and commercial dependencies in software that [Organisation name] builds and runs (consumer view of software composition analysis findings).
  • Cloud configuration weaknesses with a CVE or vendor advisory. General misconfigurations are handled by Cloud Security Posture Review (security/cloud-security-posture-review).
  • Suppliers that operate systems on behalf of [Organisation name], through contract clauses requiring equivalent SLAs.

2.2 Out of scope

TopicWhere it is covered
Receiving and fixing vulnerabilities in products that [Organisation name] ships to customersPSIRT Vulnerability Handling (security/psirt-vulnerability-handling)
Public coordinated vulnerability disclosure policyVulnerability Disclosure Policy (security/vulnerability-disclosure-policy)
EU Cyber Resilience Act manufacturer dutiesCRA Vulnerability Handling and Reporting Procedure Template (security/cra-vulnerability-handling)
Patch packaging, testing rings, maintenance windows, routine patch cadencePatch Management Policy (ops/patch-management-policy)
Penetration testing methodPenetration Test Report Template (PTES, OWASP WSTG) (security/pentest-report)

3. Roles and responsibilities

ActivityVulnerability management teamAsset or service ownerPlatform and IT operationsSOC and threat intelligenceCISORisk manager
Maintain scanning tools and coverageA/RCR (agents, credentials)III
Keep asset inventory and ownership accurateCA/RRIII
Enrich and prioritise findingsA/RCCR (exploitation intelligence)II
Remediate or mitigateCARIII
Verify closureA/RCCIII
Emergency response to actively exploited vulnerabilitiesR (coordination)RRRAI
Approve exceptionsR (assess)R (request)CCA (critical)C
Report metricsA/RIIIII

R = responsible, A = accountable, C = consulted, I = informed.

4. Asset coverage

4.1 Asset tiers

TierDefinitionExamples
T1 Internet-facing criticalReachable from the internet, or controls access to many systemsEdge VPN and remote access gateways, public load balancers, identity provider, CI/CD control plane
T2 ProductionRuns production workloads or holds production data, not directly internet-facingApplication servers, Kubernetes nodes, databases, message brokers
T3 Internal businessSupports internal business processesWorkforce endpoints, internal tools, file servers
T4 Non-production and isolatedNo production data and no production connectivityIsolated test labs, sandboxes

4.2 Assessment methods per asset class

Asset classMethodTool or sourceCoverage target (Example)
Servers and virtual machinesAuthenticated agent or credentialed network scan[Scanner]98% of inventory
Containers and imagesRegistry and admission-time image scanning; runtime inventory[Scanner]100% of images deployed to production
Kubernetes clustersNode image versions, control plane version, cluster add-ons[Tool]; see Kubernetes Cluster Design (ops/kubernetes-cluster-design)100% of clusters
Cloud servicesProvider security findings and CSPM, mapped to CVEs where available[CSPM]100% of accounts
Application dependenciesSoftware composition analysis in CI and on deployed SBOMs[SCA tool]100% of production services
Network and security appliancesVendor advisory monitoring plus external scanning[Advisory feed], [external scanner]100% of T1 appliances
EndpointsEndpoint management or EDR vulnerability module[Tool]95% of active endpoints
SaaSVendor advisories, shared responsibility reviewSupplier registerAll SaaS holding confidential data
External attack surfaceUnauthenticated external discovery and scanning[Tool]All registered domains and IP ranges

5. Discovery and scanning cadence

ActivityT1T2T3T4
Authenticated or agent-based scan (Example)Continuous (agent) or dailyDailyWeeklyMonthly
External attack surface scan (Example)Dailyn/an/an/a
Image and dependency scan (Example)Every build and daily on deployed artefactsEvery build and dailyEvery buildEvery build
KEV catalog and vendor advisory check (Example)Daily, automatedDailyDailyWeekly
Targeted scan after a critical advisory (Example)Within 24 hours of detection content being availableWithin 48 hoursWithin 5 daysNext cycle

Intake sources: scanners, SCA, the CISA KEV catalog, vendor advisories, CSAF and VEX documents from suppliers, threat intelligence, bug bounty and penetration test reports, and internal reports. Where a supplier publishes a VEX statement that a product is not affected, record it and close the finding with that evidence (see SBOM and VEX Policy, security/sbom-vex-policy).

6. Policy statements

IDStatementLevel
VM-01Every in-scope asset MUST be in the authoritative inventory with an owner and a tier.MUST
VM-02In-scope assets MUST be assessed at the cadence in section 5; coverage gaps MUST be reported.MUST
VM-03Every finding MUST be prioritised with the model in section 7 within [1] business day (Example) of detection.MUST
VM-04Vulnerabilities listed in the CISA KEV catalog MUST be treated as having active exploitation.MUST
VM-05Findings MUST be remediated or mitigated within the SLA in section 8, measured from the detection date.MUST
VM-06A finding that will miss its SLA MUST have an approved exception before the SLA expires (section 10).MUST
VM-07Closure MUST be verified by rescan or equivalent evidence (section 9).MUST
VM-08Software and hardware past vendor end of support MUST be removed, isolated or covered by an exception with a retirement date.MUST
VM-09For an Act decision on a T1 asset, and for any finding the directive overlay (section 7.3, overlay 6) places in a directive's fastest tier, owners SHOULD apply a mitigation within hours if the full fix needs longer, and MUST check for signs of compromise (section 9.3).MUST / SHOULD
VM-10Suppliers operating in-scope systems MUST meet equivalent SLAs by contract and report status on request.MUST
VM-11Vulnerability data MUST be classified Confidential and shared only on a need-to-know basis.MUST
VM-12Teams SHOULD fix classes of vulnerabilities at the source (base images, golden AMIs, dependency update automation) rather than per host.SHOULD

7. Risk-based prioritisation

7.1 Inputs

InputSourceHow it is used
CVSS v4.0 scoreFIRST CVSS v4.0; use the label that matches the metrics applied: CVSS-B (base), CVSS-BT (base and threat), CVSS-BE (base and environmental), CVSS-BTE (all)Technical severity and tie-breaker; record the vector string. Where only a CVSS v3.x score exists, record it and its version
Exploitation statusCISA KEV catalog, threat intelligence, vendor advisories, CVSS v4.0 Exploit MaturityDecision point: None, Public proof of concept, Active
EPSS probabilityFIRST EPSS daily scores (record the score date and the model version shown in the data feed)Overlay that raises priority when exploitation is likely in the next 30 days
AutomatableAnalyst judgement using the CISA SSVC guide definition (can an attacker reliably automate the steps to exploitation)Decision point: Yes, No
Technical impactAnalyst judgement: does successful exploitation give total control of the affected component or only partialDecision point: Partial, Total
ExposureAsset tier (section 4.1) and the internet-reachable flag on the asset record, set from external attack surface scanningTier selects the SLA column; the reachable flag drives overlays 3 and 6 and the directive mapping in 7.4
Mission and well-beingBusiness criticality of the asset and potential harm to customers or the publicDecision point: Low, Medium, High

Decision point names follow the CISA SSVC guide (Exploitation, Technical Impact, Automatable, Mission Prevalence, Public Well-being Impact, and the combined Mission and Well-being). The guide contains the full CISA decision tree; the table in 7.2 is an organisation-specific simplification in original wording.

7.2 Decision model (Example)

Evaluate the rules in order; the first match sets the decision.

RuleExploitationAutomatableTechnical impactMission and well-beingDecision
D1ActiveAnyAnyHighAct
D2ActiveYesTotalLow or MediumAct
D3ActiveAny other combinationLow or MediumAttend
D4Public PoCYesTotalHighAttend
D5Public PoCAny other combinationMedium or HighTrack*
D6NoneAnyTotalHighTrack*
D7Any remaining combinationTrack

Decision meanings, paraphrasing the CISA guide:

  • Track: no action outside the standard remediation cycle.
  • Track*: remediate in the standard cycle but monitor closely for changes, such as new exploitation.
  • Attend: remediate sooner than standard timelines; supervisory-level staff are involved, and the issue may be notified internally or externally.
  • Act: remediate as soon as possible with supervisory and leadership attention; coordinate internally and, where needed, externally.

7.3 Overlays (Example)

  1. KEV overlay. A KEV listing sets Exploitation to Active (VM-04).
  2. EPSS overlay. If the EPSS probability is at or above [0.10] (Example), raise Track to Track*, and Track* to Attend when the asset is T1.
  3. Exposure overlay. A T1 asset raises Track* to Attend when Technical impact is Total.
  4. Compensating control downgrade. A verified control that removes the attack path (for example the vulnerable service is disabled) can lower the decision by one level with a recorded reason; it does not close the finding.
  5. Missing inputs. If decision inputs cannot be established within one business day, use CVSS-BTE: 9.0 or above is treated as Attend, 7.0 to 8.9 as Track*, otherwise Track, until the analysis is complete. A KEV-listed finding is never handled under this fallback: it starts at Attend or above.
  6. Regulatory and directive overlay. Where a binding directive, regulation or contract sets shorter timelines for a system, those timelines override this internal model. For US federal systems, and for supplier systems where a contract imposes it, CISA BOD 26-04 applies: take the timeline for each combination of factors from Table 1 of the directive at cisa.gov, and use it whenever it is shorter than the SLA in section 8. A KEV-listed finding on an exposed asset (T1 or internet-reachable flag set) MUST never fall into the slowest band: its decision is never lower than Attend, overlay 4 cannot lower it further, and when its technical impact is Total it is handled as Act with the compromise check in section 9.3.

7.4 Reference model: CISA BOD 26-04

CISA Binding Operational Directive 26-04, Prioritizing Security Updates Based on Risk (10 June 2026), applies to US federal civilian executive branch agencies and replaces BOD 22-01 and BOD 19-02. It sets remediation timelines from a combination of four factors: whether the affected asset is publicly exposed, whether the vulnerability is in the KEV catalog, whether exploitation is automatable, and the technical impact of exploitation. The day counts for each combination are set in the directive's table and its implementation guidance, which CISA updates; they are not reproduced here. Organisations that must comply, or want to align, should read the directive and implementation guidance at https://www.cisa.gov/news-events/directives/bod-26-04-prioritizing-security-updates-based-risk and set the SLA matrix in section 8 to be no slower than the directive requires for the systems it covers.

Mapping of this policy's inputs to the directive's factors:

BOD 26-04 factorInput in this policy
Public exposureReachability evidence (internet-reachable flag on the asset record), independent of tier
KEV listingExploitation = Active through the KEV overlay
AutomatableAutomatable decision point
Technical impactTechnical impact decision point

7.5 Worked prioritisation (Example)

Finding IDVulnerabilityAsset tierExploitationAuto.ImpactMissionEPSSCVSS-BTEDecision
VMF-101 (Example)Remote code execution in edge VPN appliance (CVE-[YYYY-NNNNN]), KEV-listedT1ActiveYesTotalHigh0.949.3Act (D1)
VMF-102 (Example)Privilege escalation in Linux kernel on Kubernetes nodes, public PoCT2PoCNoTotalHigh0.067.3Track* (D5)
VMF-103 (Example)Deserialisation flaw in a Java library used by the checkout service, public PoCT1PoCYesTotalHigh0.218.7Attend (D4)
VMF-104 (Example)Information disclosure in internal wiki plug-in, no known exploitT3NoneNoPartialLow0.015.1Track (D7)
VMF-105 (Example)Denial of service in internal message broker, no known exploitT2NoneYesPartialMedium0.126.2Track* (D7, raised by EPSS overlay)

8. Remediation SLAs

All values are Example values measured in calendar days from detection.

DecisionT1 Internet-facing criticalT2 ProductionT3 Internal businessT4 Non-production
ActMitigate within 24 hours; remediate within 3 daysRemediate within 7 daysRemediate within 14 daysRemediate within 30 days
AttendRemediate within 14 daysRemediate within 21 daysRemediate within 30 daysRemediate within 60 days
Track*Remediate within 30 daysRemediate within 45 daysRemediate within 60 daysNext standard cycle
TrackRemediate within 60 daysNext standard patch cycle, at most 90 daysNext standard patch cycle, at most 120 daysNext standard cycle, at most 180 days

Rules:

  1. "Next standard patch cycle" refers to the routine cadence in Patch Management Policy (ops/patch-management-policy).

  2. When a regulation, directive or contract sets a shorter time for a system, the shorter time applies (overlay 6 in section 7.3). Record those timelines here so analysts do not have to look them up per finding:

    Directive or contractSystems coveredTimelines
    CISA BOD 26-04 (if it applies)[Systems in scope of the directive or of a contract that imposes it][Copy the current Table 1 from the directive at cisa.gov, with the date you copied it]
    [Customer contract or regulation][Systems][Timelines]
  3. If no fix exists, apply a vendor-recommended mitigation within the SLA and track the fix as a dependent action; a mitigation without a fix keeps the finding open with status "Mitigated".

  4. A finding's SLA is recalculated from the date its decision rises (for example when it is added to KEV), and never extended when its decision falls.

  5. End-of-support software follows VM-08 regardless of individual finding SLAs.

9. Remediation, verification and closure

9.1 Remediation options, in order of preference

  1. Apply the vendor fix or upgrade to a fixed version.
  2. Remove the vulnerable component or decommission the asset.
  3. Apply a configuration change or vendor mitigation that removes the attack path.
  4. Apply a compensating control (network isolation, WAF rule, feature disable) with an exception if the SLA will be missed.

9.2 Verification and closure

StatusMeaningEvidence
OpenDetected and prioritisedScanner finding with decision and SLA
In remediationOwner has a change scheduledChange or ticket reference
MitigatedAttack path blocked, fix pendingMitigation evidence; dependent action
ExceptedApproved exception in forceException ID (section 10)
Closed: fixedFix appliedRescan shows the finding absent, or package inventory shows a fixed version
Closed: not affectedSupplier VEX or analysis shows not exploitable in this contextVEX statement or analyst note, reviewed
Closed: asset retiredAsset removedInventory record

A finding that reappears within 90 days (Example) reopens with its original detection date for SLA measurement.

9.3 Act decisions on exposed assets

For an Act decision on a T1 asset, and for any finding the directive overlay (section 7.3, overlay 6) places in a directive's fastest tier, such as a KEV-listed vulnerability with total technical impact:

  1. Open a security incident or emergency change per Incident Severity Matrix (ops/incident-severity-matrix).
  2. Apply the mitigation or fix and verify by rescan.
  3. Ask the SOC to hunt for indicators of exploitation from the earliest public exploitation date, using vendor and CISA guidance, and to preserve evidence if any are found.
  4. Record the outcome of the compromise check on the finding.

10. Exceptions and risk acceptance

10.1 Approval levels (Example)

Decision of the findingApproverMaximum duration
ActCISO, with the business owner of the service30 days
AttendHead of security engineering or deputy CISO90 days
Track*Vulnerability management lead180 days
TrackAsset owner, recorded in the tool365 days

10.2 Exception record

FieldExample (Example)
Exception IDVMX-042
FindingsVMF-102 on 36 Kubernetes nodes
StatementVM-05
ReasonKernel fix requires a node image rebuild; the managed node image with the fix is released after the SLA date
Compensating controlsPod security admission set to restricted for untrusted workloads; no shell access to nodes; runtime detection rule for privilege escalation enabled
Residual riskMedium
Approver[@vulnerability-management-lead] (Track* level per section 10.1)
ExpiryYYYY-MM-DD
Review triggerExploitation becomes Active, or EPSS rises above the overlay threshold

Rules: exceptions are requested before the SLA expires (VM-06); expired exceptions revert the finding to overdue; a Medium or High residual risk is recorded in the enterprise risk register (Risk Register Template, compliance/risk-register).

11. Metrics

MetricFormulaTarget (Example)
Scan coverageassets assessed in cadence / in-scope assets in the inventory98%
Mean time to remediate (MTTR), per decisionsum of (closure date minus detection date) for findings closed in period / findings closed in periodAct on T1: 3 days or fewer
SLA adherencefindings closed within SLA in period / findings closed in period90% overall; 100% for Act
KEV backlogcount of open KEV-listed findings past their original SLA at period end, including those under exception0
KEV exposure timemedian days from KEV listing (or detection, if later) to closure for T1 assetsTrend down
Overdue findings by decisioncount of open findings past SLA without exception, per decision0 for Act and Attend
Exception rateopen findings with an exception / open findings5% or lower
Recurrence ratefindings reopened within 90 days / findings closed3% or lower
End-of-support assetscount of in-scope assets running unsupported softwareTrend down to 0

Example (Corvid Commerce, Q3): 4,410 of 4,600 inventoried assets assessed in cadence: coverage = 4,410 / 4,600 = 95.9%. 1,180 findings closed, 1,062 within SLA: adherence = 1,062 / 1,180 = 90.0%. KEV backlog at quarter end: 1 (VMF-088, under exception VMX-039).

12. Reporting and governance

ReportAudienceFrequencyContent
Operational dashboardAsset owners, engineering managersContinuousOpen findings by decision and SLA status, per team
Act and Attend reviewVulnerability management, platform, SOCWeeklyNew Act and Attend findings, mitigations, compromise checks
Monthly risk reportCISO, CIO, heads of engineeringMonthlyMetrics in section 11, top overdue items, exceptions expiring
Quarterly governanceRisk committeeQuarterlyTrend, structural causes, resourcing decisions, policy changes
Customer or regulator evidenceAs contractedOn requestCoverage, SLA adherence, KEV handling statement

13. Compliance measurement and review

  1. The vulnerability management team checks compliance monthly using the metrics in section 11 and the inventory reconciliation.
  2. Internal audit tests a sample of closed and excepted findings at least annually for evidence quality.
  3. Repeated non-compliance by a team is escalated to its engineering director, then to the CISO.
  4. The policy is reviewed annually and also when: a new directive or regulation changes timelines (for example BOD 26-04 implementation guidance updates), a major incident involves a known vulnerability, or SLA adherence stays below target for two quarters.

14. Related documents

DocumentRelationship
Patch Management Policy (ops/patch-management-policy)Patch cadences, testing rings and maintenance windows used to meet these SLAs
PSIRT Vulnerability Handling (security/psirt-vulnerability-handling)Producer-side handling for products [Organisation name] ships
Vulnerability Disclosure Policy (security/vulnerability-disclosure-policy)Public reporting channel
CRA Vulnerability Handling and Reporting Procedure Template (security/cra-vulnerability-handling)EU CRA manufacturer duties
Software Supply Chain Security Assessment Template (SLSA) (security/supply-chain-security)Dependency and VEX practices
SBOM and VEX Policy (security/sbom-vex-policy)SBOM and VEX evidence used to close findings
Cloud Security Posture Review (security/cloud-security-posture-review)Cloud configuration findings
Penetration Test Report Template (PTES, OWASP WSTG) (security/pentest-report)Source of findings
Incident Severity Matrix (ops/incident-severity-matrix)Severity of exploitation incidents
Security Logging Standard (security/security-logging-standard)Logs needed for compromise checks

15. Glossary and references

15.1 Glossary

TermMeaning
VulnerabilityA weakness that an attacker could exploit to affect confidentiality, integrity or availability
FindingOne vulnerability on one asset, as recorded in the vulnerability management tool
CVSSCommon Vulnerability Scoring System, maintained by FIRST; version 4.0 has base, threat, environmental and supplemental metric groups
EPSSExploit Prediction Scoring System: a daily probability that a published CVE will be exploited in the wild in the next 30 days
KEVCISA Known Exploited Vulnerabilities catalog
SSVCStakeholder-Specific Vulnerability Categorization: a decision-tree approach whose CISA version ends in Track, Track*, Attend or Act
VEXVulnerability Exploitability eXchange: a statement of whether a product is affected by a vulnerability
MitigationA change that blocks or reduces an attack path without fixing the flaw
MTTRMean time to remediate
SLAThe maximum time allowed to remediate a finding, set by decision and asset tier

15.2 References

Standard and licence note

Based on NIST SP 800-40 Rev. 4 (April 2022), NIST SP 800-53 Rev. 5, the CISA Stakeholder-Specific Vulnerability Categorization Guide (November 2022), the CISA KEV catalog and CISA BOD 26-04 (June 2026), which are works of the United States Government. SSVC decision point and outcome names are cited from the CISA guide; the decision rules in section 7.2 are an original simplification, not the CISA tree. CVSS and EPSS are FIRST specifications cited by name and notation only; no FIRST text is reproduced. BOD 26-04 day counts are deliberately not reproduced: read the directive and its implementation guidance. Check the current official versions at cisa.gov, nist.gov and first.org before relying on this policy, as updates may have been published. CVSS and EPSS are marks of FIRST; this template is not affiliated with or endorsed by FIRST or CISA. Original text written for this collection. Template licence: MIT.

Ähnliche Vorlagen