Vulnerability teams face tens of thousands of new CVEs a year but only a small fraction are ever exploited, and the CISA Known Exploited Vulnerabilities (KEV) catalogue is the most widely used public list of the ones that are. KEV started as the remediation list for US federal agencies under Binding Operational Directive (BOD) 22-01 and is now referenced by BOD 26-04, issued in June 2026. Organisations far outside government use it as a prioritisation signal, a scanner filter and an audit benchmark. This guide explains what qualifies a vulnerability for KEV, every field in the feed, how the federal deadlines work, how to consume the data automatically and how to build an internal service level around it. It is written for vulnerability management leads, SOC and threat intelligence analysts and engineers integrating KEV into tooling.
Key takeaways
- KEV is evidence of exploitation, not a severity score. An entry needs a CVE ID, reliable evidence of active exploitation and clear remediation guidance. CVSS does not decide inclusion, and many KEV entries are not rated critical.
- The directive behind it has changed. Entries added since 11 June 2026 cite BOD 26-04 "Prioritizing Security Updates Based on Risk", carry a new
forensicTriagefield and mostly have due dates three days after addition. Read the current directive text rather than relying on the older two-week rule. - Use the machine-readable feed. CISA publishes KEV as JSON and CSV with a JSON schema, mirrored on GitHub under a CC0 licence, and NVD exposes KEV data through the
hasKevAPI parameter. Poll it and diff it; do not rely on reading the web page. - KEV is a floor, not a ceiling. Inclusion lags real-world exploitation and is limited to vulnerabilities with CVE IDs and fixes. Combine KEV with EPSS, SSVC, vendor advisories and your own exposure data.
- Patching a KEV entry is often not the end. For exploited edge devices, session tokens, machine keys and credentials may already be stolen. The
forensicTriageflag and the required action text tell you when to look for compromise as well as patch.
What the KEV catalogue is
The catalogue is maintained by the Cybersecurity and Infrastructure Security Agency. The CISA criteria page sets three conditions, all of which must be met:
- Assigned CVE ID: the vulnerability has a CVE identifier.
- Active exploitation: there is reliable evidence that it has been exploited in the wild. CISA counts attempted exploitation (for example code executed against a honeypot) as well as successful exploitation. Scanning, security research and proof-of-concept code do not count, and a public proof of concept is not required for inclusion.
- Clear remediation guidance: there is a clear action to take, such as a vendor update.
As of catalogue version 2026.10.21 (released 21 October 2026) the catalogue holds 1,754 entries. Additions by year show how the list grew: 311 in 2021, when the catalogue launched with a large backlog, 555 in 2022, 187 in 2023, 186 in 2024, 245 in 2025 and 270 in 2026 up to 21 October. Microsoft (390 entries), Cisco (101), Apple (95), Adobe (82) and Google (75) are the most frequent vendors, which reflects both deployment scale and how often their products are targeted.
For the broader picture of zero-days and exploited flaws, see the Zero-Day and Exploited Vulnerabilities Guide. This guide stays on KEV itself.
The KEV data fields
The KEV JSON schema published by CISA defines the catalogue object (catalogVersion, dateReleased, count, vulnerabilities) and the following fields for each vulnerability. The first eight are required by the schema.
| Field | Meaning (from the CISA schema) | How to use it |
|---|---|---|
cveID | CVE ID; the number part can have more than four digits | Join key to scanner findings, NVD and your CMDB; validate with the schema pattern |
vendorProject, product | Vendor or project name, and product | Rough match to asset inventory; "Multiple Products" entries need manual mapping |
vulnerabilityName | Name of the vulnerability | Ticket title |
dateAdded | Date added to the catalogue | Start of your internal service level clock |
shortDescription | Short description | Context for asset owners |
requiredAction | The required action | Read every time: it may require more than patching, such as killing sessions, disconnecting servers or following directive guidance |
dueDate | Date the required action is due | Federal deadline; a useful benchmark for your own SLA |
knownRansomwareCampaignUse | "Known" if leveraged in a ransomware campaign; "Unknown" if CISA lacks confirmation | Raise priority for "Known"; never treat "Unknown" as "safe" |
forensicTriage | "Yes" if the vulnerability requires forensic triage per BOD 26-04, "No" if not | Trigger a compromise check on affected assets, not just a patch |
notes | Additional notes, usually vendor advisory, directive and NVD links | Source for mitigation detail and detection guidance |
cwes | Associated CWE codes | Trend analysis of weakness classes in your estate |
In the 21 October 2026 catalogue, 361 entries (about one in five) are marked "Known" for ransomware use. The most common CWEs are CWE-20 (improper input validation), CWE-78 (OS command injection), CWE-787 (out-of-bounds write) and CWE-416 (use after free); 175 entries have no CWE recorded.
Federal deadlines: BOD 22-01 and BOD 26-04
KEV deadlines are legally binding only on US Federal Civilian Executive Branch (FCEB) agencies. Everyone else uses them as a benchmark.
BOD 22-01 (November 2021)
BOD 22-01 "Reducing the Significant Risk of Known Exploited Vulnerabilities" created the catalogue. Its original timelines were two weeks for vulnerabilities with CVE IDs assigned in 2021 or later and six months for older ones, which is visible in the first batch of entries: those added on 3 November 2021 are due either on 17 November 2021 or 3 May 2022. Later additions mostly carried a 21-day due date, with shorter dates (14 or 3 days) for urgent cases. The required action for agencies was to apply updates per vendor instructions or remove end-of-life products that cannot be updated.
BOD 26-04 (June 2026)
CISA announced BOD 26-04 "Prioritizing Security Updates Based on Risk" in June 2026. CISA's own bulletins say it establishes vulnerability management requirements for FCEB agencies, directs them to act quickly on the most dangerous KEV vulnerabilities, specifically those on publicly exposed assets that would give an attacker full control, to defer action on lower-risk ones, and sets expectations for when agencies must check whether threat actors have already compromised a system. The KEV data shows the effect directly:
- New required action wording: from 11 June 2026, the
requiredActionfield references BOD 26-04 and CISA's "Forensics Triage Requirements", and asks stakeholders to evaluate each asset's internet exposure. - Shorter due dates: of the 147 entries added between 1 June and 21 October 2026, 116 are due three days after addition and 31 are due in 14 days.
- Forensic triage flag: 89 of those entries have
forensicTriageset to "Yes", all with three-day due dates.
The full BOD 26-04 decision matrix (how public exposure, automation and technical impact combine into deadlines, including for vulnerabilities not in KEV) is set out in the directive and its implementation guidance on cisa.gov. Read those pages directly before you copy any timeline into policy; this guide does not reproduce the matrix because the directive pages could not be verified at the time of writing. Note that the BOD 26-04 decision variables echo the SSVC decision points that CISA already publishes for CVEs through its Vulnrichment project (Exploitation, Automatable and Technical Impact). See the SSVC Guide.
Consuming KEV data
| Channel | What you get | Best for |
|---|---|---|
| cisa.gov KEV JSON and CSV | The canonical catalogue, with the JSON schema | Authoritative pulls; validate against the schema on ingest |
| GitHub cisagov/kev-data | Same JSON, CSV and schema, updated within minutes of cisa.gov, typically on weekdays in US Eastern business hours; CC0 licence | Automation, and the commit history, which is the easiest way to see what changed in each update |
| NVD CVE API 2.0 | hasKev filter, kevStartDate and kevEndDate; CVE records carry cisaExploitAdd, cisaActionDue, cisaRequiredAction and cisaVulnerabilityName | Joining KEV status to CVSS, CPE and references in one call |
| Scanner and VM platform integrations | A KEV flag on findings | Day-to-day prioritisation, provided you confirm how often the vendor refreshes it |
Practical integration rules:
- Poll and diff: fetch the JSON at least daily, compare with the previous copy, and raise a ticket for every new
cveIDand for changes todueDate,requiredAction,knownRansomwareCampaignUseorforensicTriage. Entries can change after they are added. - Validate before trusting: check the file against the published schema and the
countfield. A truncated download should fail loudly, not silently drop entries. - Match on more than CVE: scanners miss products they do not fingerprint, particularly appliances, SaaS-managed agents and embedded components. Use
vendorProjectandproductto search your inventory for assets the scanner cannot see. - Keep the clock honest: start your SLA from
dateAdded, or from your own detection date if later, and record both.
Building an internal KEV service level
Non-federal organisations should set their own targets based on exposure. A workable pattern, aligned with the direction of BOD 26-04 but not a copy of it, is:
| Asset context | Suggested target for a KEV entry | Extra action |
|---|---|---|
| Internet-facing edge device or application (VPN, gateway, file transfer, web server) | Mitigate within 72 hours or take offline | Compromise check when forensicTriage is "Yes" or the vendor reports exploitation before the fix; rotate secrets the device held |
| Internal server with high privilege or sensitive data | Within 14 days | Hunt for exploitation indicators from the vendor and CISA advisories |
| End-user devices and browsers | Within 14 days, through normal update rings accelerated | Confirm update compliance, not just deployment |
| Systems that cannot be patched (end of life, OT, vendor-controlled) | Compensating control within the same window, with documented risk acceptance | Isolation, virtual patching, removal plan. See the Virtual Patching and Compensating Controls Guide |
Decision guidance: small organisations with little patching capacity should treat "KEV and internet-facing" as the single most important queue and accept slower action elsewhere. Large organisations should automate the KEV join in their vulnerability platform, route tickets by asset owner, and report KEV exposure separately from overall vulnerability counts. Organisations in regulated sectors should expect auditors and insurers to ask how quickly KEV entries are closed.
How to roll it out
- Baseline: join the current catalogue to your scanner results and asset inventory and count open KEV findings by asset type and exposure. This is usually the most useful single number to show leadership.
- Close the internet-facing gap first: work through KEV entries on external attack surface, using external scanning and attack surface data to find assets the internal scanner misses. See the Attack Surface Management Guide.
- Automate ingestion: schedule the poll, diff and ticketing described above, with alerts to the vulnerability and SOC teams for new entries affecting your products.
- Write the SLA into policy: add KEV targets and an exception process to your vulnerability management policy. See the Vulnerability Management Policy Template.
- Connect to incident response: for edge devices and entries flagged for forensic triage, define who performs the compromise check, what evidence they collect and when the asset may return to service.
- Test and review: sample closed KEV tickets each quarter to confirm the fix is actually present on the asset (rescan, version check) and that compromise checks were done when required.
Where identity fits
Many KEV entries on edge devices are valuable to attackers because of the identities and secrets they expose: session tokens, cached credentials, service account passwords used for LDAP binds, and keys stored on the appliance. CitrixBleed 2 (CVE-2025-5777) was added to KEV on 10 July 2025, and patching alone did not end the exposure because leaked session tokens stayed valid until sessions were terminated; see the CitrixBleed 2 (CVE-2025-5777) breach analysis. In the ToolShell SharePoint campaign, CVE-2025-53770 was added to KEV on 20 July 2025 and attackers who had stolen ASP.NET machine keys could regain access after patching unless the keys were rotated; see the ToolShell SharePoint exploitation analysis. When a KEV entry affects a system that holds credentials, add credential and key rotation to the required action. See the Secrets Rotation Runbook Template.
Detection and measurement
- Open KEV exposure: count of open KEV findings, split by internet-facing and internal, trended weekly.
- Time to remediate KEV: median and 90th percentile days from
dateAdded(or detection) to verified fix, compared with your SLA and with the federaldueDate. - SLA compliance rate: share of KEV findings closed within target, by business unit.
- Ingestion latency: hours between a KEV update and a ticket in your system.
- Blind spot count: KEV entries that match a product in your inventory but produce no scanner finding, which indicates either a clean estate or a coverage gap that needs a manual check.
- Compromise checks completed: share of
forensicTriage"Yes" or edge-device KEV findings with a documented compromise assessment.
For detection, watch for exploitation of public-facing applications (T1190) and abuse of external remote services (T1133) on assets with open KEV findings, and use the vendor and CISA advisories in the notes field for indicators.
Common failure patterns
- Treating KEV as the whole list. KEV covers only vulnerabilities with CVE IDs, confirmed exploitation and a fix, and is added after exploitation is seen. Exploited vulnerabilities in your sector may appear in vendor advisories days before KEV.
- Ignoring the required action text. Teams apply the patch and close the ticket when the
requiredActionalso said to kill sessions, disconnect servers or follow specific guidance. The CitrixBleed (CVE-2023-4966) entry, for example, required killing all active and persistent sessions as well as applying mitigations. See the CitrixBleed 2023 breach analysis. - Reading "Unknown" ransomware use as low risk. "Unknown" means CISA lacks confirmation, not that ransomware groups do not use it.
- Scanner-only matching. Edge appliances and unmanaged products are often missing from scanner coverage, which is exactly where many KEV entries sit.
- Copying an outdated federal timeline. Policies that still say "two weeks for all KEV" do not reflect BOD 26-04's shorter, exposure-based deadlines.
- Patching without a compromise check. If the vulnerability was exploited before you patched, the attacker may already have persistence or credentials.
Framework mapping
| Framework | Reference | How KEV supports it |
|---|---|---|
| NIST SP 800-53 Rev. 5 | RA-5 Vulnerability Monitoring and Scanning; SI-2 Flaw Remediation; SI-2(2) Automated Flaw Remediation Status | Prioritising and tracking remediation of exploited flaws |
| NIST SP 800-53 Rev. 5 | SI-5 Security Alerts, Advisories, and Directives; CM-8 System Component Inventory | Receiving KEV updates and mapping them to inventory |
| NIST CSF 2.0 | ID.RA-01, ID.RA-02, ID.RA-05, PR.PS-02, ID.AM-02 | Identifying vulnerabilities, receiving threat intelligence, prioritising risk response, maintaining software, inventory |
| CIS Controls v8.1 | Control 7 Continuous Vulnerability Management | Risk-based remediation process. See the CIS Control 7 Continuous Vulnerability Management Guide |
| ISO/IEC 27001:2022 Annex A | A.8.8 Management of technical vulnerabilities | Timely evaluation and remediation of exploited vulnerabilities |
| MITRE ATT&CK v20 | T1190, T1133 | Initial access techniques most KEV edge-device entries enable |
Practitioner checklist
- Ingest the KEV JSON at least daily, validate it against the CISA schema and diff it against the previous copy.
- Join KEV to scanner findings, external attack surface data and asset inventory, including products the scanner cannot see.
- Set an internal KEV SLA by exposure, with internet-facing assets first, and write it into policy with an exception process.
- Read and act on the full
requiredActiontext, not only the patch. - Trigger a compromise check for
forensicTriage"Yes" entries and for exploited edge devices, and rotate credentials and keys they held. - Review your policy against the current BOD 26-04 text and implementation guidance on cisa.gov.
- Combine KEV with EPSS and SSVC for vulnerabilities not yet in the catalogue.
- Report open KEV exposure and time to remediate KEV to leadership monthly.
Standards and references
- CISA: Reducing the Significant Risk of Known Exploited Vulnerabilities (KEV criteria)
- CISA: KEV catalogue update bulletin referencing BOD 26-04
- CISA: kev-data repository and JSON schema
- CISA: Vulnrichment (SSVC decision points for CVEs)
- NIST NVD: CVE API 2.0
- NIST: SP 800-53 Rev. 5 Security and Privacy Controls
- NIST: Cybersecurity Framework 2.0
Related NHI Mgmt Group resources: Zero-Day and Exploited Vulnerabilities Guide · EPSS Guide · Risk-Based Vulnerability Management Guide · Emergency Out-of-Band Patching Playbook