Summary

  • The catalog's 20 November 2026 release prints 1,790 entries with twelve fields each: two of them are dates, and none records whether the work was done.
  • BOD 26-04, dated 10 June 2026, revoked BOD 19-02 and BOD 22-01, computes each due date from four variables, starts the clock at whichever comes first — the regulator's listing or the agency's own enumeration — and prints no penalty for a missed timeline.

Start with a single row, because the row is not vague. The feed CISA released on 20 November 2026 carries an entry for CVE-2026-129448, and it prints a vendor project, a product, a vulnerability name, a short description, a required action, a ransomware flag, a forensic-triage flag, a weakness code, notes, the date the vulnerability was added — 2026-11-20 — and the date it must be remediated by: 2026-11-23. Twelve fields in all (JSON feed). Two of them are dates. None of the twelve records whether the remediation happened.

That is the whole instrument in one line. The deadline is exact to the calendar day; the outcome is somebody else's file. The same release prints its own header — catalogVersion 2026.11.20, dateReleased 2026-11-20T17:48:39Z, and a count of 1,790 live entries.

The interval is not a fixed window, and that is the first thing a reader arriving with old knowledge gets wrong. An entry added 2026-11-18 prints a due date of 2026-11-21 in one row and 2026-12-02 in another — three calendar days and fourteen calendar days from the same addition date, by simple subtraction of printed dates. Something computes per vulnerability. That something is a directive.

BOD 26-04, Prioritizing Security Updates Based on Risk, dated 10 June 2026, states it "supersedes and hereby revokes BOD 19-02 ... and BOD 22-01". Its operative verb is shall: Required Action 6 directs agencies to "Remediate each vulnerability as quickly as possible and no later than the timelines set forth in Table 1: Remediation Timelines". The date derives from four variables — Asset Exposure, KEV Status, Exploit Automation and Technical Impact — and Appendix A settles the unit: days are calendar days. The clock starts at whichever comes first: CISA adding the vulnerability to the catalog, or the agency identifying it on an asset and updating the CDM Agency and Federal Dashboard.

Table 1 carries the printed day counts, and in the directive text I read it arrives as an image, so I quote none of its cells. The number I can use is the one CISA prints in its own prose: publicly exposed, total technical impact and automatable yields a three-day deadline, with the agency making the final asset-exposure determination. The feed's 20 November row independently prints three days. The two agree without either being derived from the other.

The revoked directive matters because it is still what most people are describing. BOD 22-01, issued 3 November 2021, had required agencies to remediate each vulnerability "according to the timelines set forth in the CISA-managed vulnerability catalog", with default windows of six months for CVEs assigned before 2021 and two weeks for everything else. Its own page now opens with a notice: "This Directive has been revoked. It is superseded by BOD 26-04." The catalog agrees with the banner — recent entries' required-action text names BOD 26-04 and CISA's "Forensics Triage Requirements", not the 2021 directive. Anyone still quoting six months and two weeks is working from a superseded page.

Two CISA surfaces then describe the same act in two strengths. The same-day press release says the directive requires agencies to "assess and align their vulnerability management policies" around the four criteria, and describes it as harmonizing and improving the two earlier directives. The directive says it supersedes and hereby revokes them. Both sentences belong to CISA; the difference is what a reader should expect a revoked document to mean.

A stricter detail sits beside the softer one. The guidance is flat on waivers — "CISA does not issue waivers or exceptions for actions required in cyber directives" — and sends agencies with special use cases to CISA so it can suggest mitigation options and an estimated remediation timeframe. Actionable text, no exit stamped on the document.

Now the consequence, stated as a fact about the document rather than about anyone's conduct. The directive text I read prints no penalty, sanction or escalation for a missed timeline. What it prints instead is the statutory duty to comply under 44 U.S.C. § 3554(a)(1)(B)(ii), and Required Action 1(d), under which each agency must "Establish internal validation and enforcement procedures to ensure adherence with this Directive". Enforcement is delegated into the agency's own procedures. The DHS inspector general puts the architecture in one sentence: "although non-compliance may be escalated to OMB and reported as part of the annual Federal Information Security Modernization Act metrics cycle, there is no enforcement mechanism for BODs", and FISMA "does not give CISA the authority to enforce FCEB compliance". That report uses BOD 25-01 as its worked example, where 86 percent remained noncompliant as of June 2025 after notification — a different directive, a different cycle, predating any BOD 26-04 compliance period. It is cited for the architecture, not for the catalog.

So where does the receipt go? Required Action 3 requires automated KEV status reporting through the CDM Dashboard, with bi-weekly manual reporting for agencies that are not fully automated. CISA Actions item 9 requires CISA, by the end of each fiscal year, to report to the Secretary of Homeland Security, the OMB Director and the National Cyber Director on "cross-agency status and outstanding issues in the implementation of this Directive"; the text I read states no publication duty for that report. CISA's dashboard page describes the Federal Dashboard as giving CISA and OMB visibility across federal networks, the Agency Dashboard as an object-level view of one agency's posture — and nothing about release to the public. The nearest published artefact is a metric: the FY26/FY27 CIO FISMA set defines CIO-FY26-0121.cdm as the "CDM-Reported mean time to remediate Known Exploited Vulnerabilities (KEVs) in days", reported by agency CIOs through CyberScope. A mean in days, at agency level. Not a case. Whether individual submissions are ever published is not established by that document, which describes submission rather than disclosure.

Read carefully, that is a different claim from the one it is easy to write. The record exists, it is captured, and it travels to people with the authority to use it. It is not published. An unpublished record is not an absent one, and the case-level check has been cut off rather than skipped.

What the public does get is arithmetic on other stages. In January 2025 CISA wrote: "Since FY21, FCEB agencies have remediated an incredible 99% of internet facing KEVs identified by CISA" (post). Unit, population qualifier, window — and no denominator count. It measures remediation, not remediation by a printed date. In June 2026, reprinting a third party, CISA carried this: "Per Verizon's 2026 Data Breach Investigations Report, only 26% of vulnerabilities on CISA's Known Exploited Vulnerabilities (KEV) Catalog were fully remediated by organizations in 2025, a drop from the previous year's 38%. The median time for full resolution rose to 43 days." Organizations generally, not federal agencies; full remediation, not compliance with a due date; again no base count. The same post cites one large civilian agency where 1% of vulnerability instances fell into the three-day category and more than 60% were deferred to the next system upgrade — an anecdote about one agency, not a rate. And CISA's FY2024–FY2025 vulnerability review says "Most organizations continue to miss CISA's recommended remediation timelines, leaving systems unnecessarily exposed", while printing its own disclaimer that the product "is not a product of the DHS Office of Homeland Security Statistics or the CISA Statistical Official"; its timing evidence comes from scanning 8,852 enrolled organizations in FY2024 and 10,890 in FY2025, framed as exposure beyond 30 days.

Three populations, three units, no printed bridge between any of them and the catalog's own date. I do not build that bridge, because the reviewed documents do not print it. The same refusal applies to the volume figures: secondary reporting of 2023 congressional testimony records more than 12 million KEV instances remediated across 102 agencies over two years and a 72 percent drop in KEVs left exposed 45 days or more — a volume measure, from a second-hand account, with no due-date denominator anywhere in it.

Here is where the checking stops, and it is worth naming precisely. The human-facing catalog page is the entry point a reader would open first, and I could not read its text for this review, so no column heading is quoted from it; every field statement above comes from the machine-readable feed or from CISA's related pages. Table 1 of the directive is an image. The metrics document describes submission. The annual report goes to three officials with no printed publication requirement. CISA's KEV explainer confirms the binding: all FCEB agencies must remediate KEV entries within the prescribed timeframes under BOD 26-04, organizations outside the FCEB are not bound but strongly recommended to act. Nothing in that set lets a stranger test one row.

A single added field would. A completion date, or a met-or-missed flag, printed where the entry already lives, returned by the system that already receives the status. No new penalty is required for that to work; the strictness is already in the text and the monitoring already exists. What is missing is only the return path — the one that would let a reader holding 1,790 dated rows ask about one of them.

BTW has written about this agency's Cyber Safety Review Board work, about CIRCIA, and about this directive itself; none of that coverage has read the catalog's per-row due date against any published record of whether it was met. The agency's directory entry is the reference point for the rest.

A deadline printed 1,790 times, to the day, is not a soft instrument. It is a precise one with the audit trail turned inward. Until one row can be checked, the number to watch is not the percentage. It is whether a sixth field ever appears beside the two dates.