Summary
- Binding Operational Directive 26-04, dated 10 June 2026, replaced a single remediation timeline with a tier computed from four variables — asset exposure, KEV status, exploit automation and technical impact — and every entry in the 20 November 2026 catalog release now cites it.
- The clock starts at whichever comes first: CISA listing the flaw, or the agency finding it on one of its own assets and updating its own dashboard. The exposure finding that selects the fastest tier is the agency's own, asset by asset.
- The public per-vulnerability record prints twelve fields, including a new forensicTriage flag, and no field at all for completion — while CISA states plainly that it does not issue waivers or exceptions for actions required in cyber directives.
Three days is what CVE-2026-129448 got. The entry for an Apache Roller deserialization flaw, CWE-502, was added to the Known Exploited Vulnerabilities catalog on 20 November 2026 and carries a due date of 23 November 2026. Its ransomware field reads Unknown. Its forensicTriage field reads Yes. Three other rows in the same release show the spread: CVE-2025-42884, SAP NetWeaver, added 20 November, due 23 November; CVE-2024-10441, Synology DSM, added 18 November, due 21 November; CVE-2026-40217, BerriAI LiteLLM, added 18 November, due 2 December (catalog feed).
Same catalog, same release, three days and fourteen days. Something is choosing between them, and it is not the vendor's severity score.
The directive that is actually in force
The choosing is done by Binding Operational Directive 26-04: Prioritizing Security Updates Based on Risk, dated 10 June 2026 (directive page). Its predecessor is gone in CISA's own words. The BOD 22-01 page is now marked Revoked and carries the sentence "This Directive has been revoked. It is superseded by BOD 26-04." (revoked page). The implementation guidance adds BOD 19-02 to the superseded list (implementation guidance).
Scope first, because scope is where directives quietly narrow. BOD 26-04 is compulsory direction to federal executive branch departments and agencies and binds Federal Civilian Executive Branch systems. It excludes statutorily defined national security systems and certain intelligence-community systems by reference to 44 U.S.C. § 3553. It reaches agency assets in any federal information system as Circular A-130 defines the term, including systems another entity operates on an agency's behalf — and it does not bind contractors unless a procurement contract says so. The directive tells agencies to go and read those contracts.
Four variables, one of them the agency's own
The tier is computed, not assigned. Four risk variables set the timeline: asset exposure, KEV status, exploit automation, and technical impact, the last split into partial and total control of the system. CISA publishes the resulting matrix as Table 1, and Table 1 is an image. That is the first place the public record thins out: a picture of a table is not a table.
The row values used here therefore come from CERT/CC's implementation model, which reproduces the sixteen combinations as text and which is a third party, not CISA (CERT/CC model). In that model the outcomes take five values: fix on system upgrade, 60 days, 14 days, 3 days, and 3 days with forensic investigation. CERT/CC writes forensic investigation where CISA writes forensic triage; the difference is terminology, and the model is the carrier, not the source.
Two rows earn the most patience: not in KEV, not publicly exposed, not automatable — partial impact in row 0, total in row 4. Everything else gets a number. The instructive pair is row 8 against row 11. A KEV-listed vulnerability with total technical impact, sitting on an asset that is not publicly exposed and whose exploitation is not automatable, gets fourteen days. A KEV-listed vulnerability that is publicly exposed and automatable, with only partial technical impact, gets three. Exposure and automation beat totality of impact. Read the tiers as a severity ranking and you will invert the priority on the two cases where it matters most.
Then there is the input the agency writes. The implementation guidance states that the final asset-exposure determination rests with the agency, asset by asset. The same guidance calls the timelines maximums, invites agencies with a lower risk tolerance to set shorter ones, and exempts items marked fix-on-system-upgrade from being prioritised at all. So the variable that pushes a vulnerability into the three-day row is decided by the organisation that has to meet the three days.
Two ways the clock starts
The interval does not begin at disclosure, and it does not begin at a severity judgement. It begins at whichever of two events occurs first: CISA adds the vulnerability to the KEV Catalog, or the agency enumerates or identifies the vulnerable product on an asset and updates the CDM Agency Dashboard and Federal Dashboard under BOD 23-01.
Read that as arithmetic on the agency's own behaviour. Enumeration is the agency's act. It can start the clock earlier than CISA would have, and it cannot start it later, because a catalog listing runs the interval regardless. The second start button exists so that a flaw a patch process is already looking at does not sit in a queue until Washington lists it.
Which makes the asset inventory a compliance instrument rather than an inventory. An agency that cannot say where a product runs cannot start its own clock, and will meet every deadline CISA starts for it.
No waiver, and one exit
Every regulated entity eventually asks whether there is a way out when the patch breaks something. CISA answers without ornament: "CISA does not issue waivers or exceptions for actions required in cyber directives." Agencies with a special use case are directed to a CISA mailbox so the agency can understand the challenge, suggest options for mitigation and give an estimated timeframe for remediation. Advice and a timeframe — not an exemption.
The exit the record does name is more expensive than a waiver. The catalog's requiredAction field tells readers to apply mitigations in accordance with vendor instructions and to follow BOD 26-04, or to discontinue use of the product if mitigations are unavailable. For a federal system, discontinuing use of a product is a service decision taken on a vulnerability deadline.
Twelve fields, and the missing one
Here is what the catalog prints for one vulnerability: cveID, vendorProject, product, vulnerabilityName, dateAdded, shortDescription, requiredAction, dueDate, knownRansomwareCampaignUse, forensicTriage, notes, cwes. Twelve fields.
forensicTriage is the new one: an input, a flag saying triage work is owed as well as a patch. Among those twelve there is no status, no patched, no compliant, no verified, no closed. The dates the record carries look forward — the day it was added, the day it is due. Nothing prints the day it was fixed, or the day it was not. This reading rests on the entries inspected in the current release and on the shape of the per-entry object, not on a field-by-field count across all 1,790 records.
That is the asymmetry worth naming. The record gained a field describing more work and no field describing whether any work happened. A count of 1,790 catalogued vulnerabilities measures the size of the obligation; it cannot measure how much of it was met.
The date the directive does not print
Three phases structure the rollout. Phase I applies immediately. Phase II sits within 60 days of issuance. Phase III sits within 180 days of issuance, with remediation required no later than the Table 1 timelines and externally reachable assets tagged.
180 days after 10 June 2026 is 7 December 2026. That date is arithmetic on two printed numbers. CISA does not print it in the directive, so treat it as the day the transition closes rather than as a date the instrument names.
What this adds
On 19 November 2026 I wrote about this catalog's arithmetic: two flaws added the same day, one due in three days and one in fourteen. That piece stopped at the computation, and it is worth reading alongside this one. This article reads the machine underneath — which directive is in force, who it binds, which two events start the clock, who writes the exposure finding, what happens when the patch turns out to break something, and what the public record does with the answer. The agency's own file is here: Cybersecurity and Infrastructure Security Agency.
The arithmetic is public, granular and dated; anyone can pull the feed and check a due date against the day a flaw was listed. The two judgements that decide which number applies — exposure and impact, made by the party under the clock — and the one fact that would close the loop, whether the work was done, stay where they were: inside the agency, and off the record.
Member Briefing
Read the full trend briefing
Members read the full briefing with the source notes behind every finding.
Strategic Circle
A daily briefing in your language, built from changes we have verified.
Join for USD 20/monthCancel anytime.
Members also get the full research library, a daily briefing on the networks they follow and the private forum.






