Picture a school with six hundred classrooms and one caretaker. Every morning, a stack of notes appears on the caretaker’s desk: a window latch is loose in building C, a fire door sticks in building F, a ceiling tile is stained in building A. The caretaker cannot fix everything on the same day, and treating every note as equally urgent would mean the sticking fire door waits behind a stained tile. What separates a well-run school from a chaotic one is not how hard the caretaker works but how the notes are sorted, who gets which job, and how the caretaker knows a repair is truly finished. That, in plain language, is VulnerabilityOps. Software has its own loose latches, and this article explains the idea without jargon, for students and career-changers.
The notes on the desk
In software, the notes are published weaknesses, each with a public ID. There are a lot of them. Counting only the public CVE list, the yearly total grew from 18,375 in 2020 to 48,244 by 2025, more than two and a half times larger. An organisation does not need to care about all of them, only the ones in software it actually runs, but the first job is knowing what it runs. A school caretaker who does not know the building plan cannot even tell which wing a note refers to. The software version of the building plan is called an asset inventory, and it is the least glamorous and most useful tool in the field.
Which job first?
Our caretaker would never fix things in alphabetical order or by the colour of the note. They would ask three questions. Has this actually caused trouble somewhere already? How likely is trouble soon? And how bad could it be if it happened? Those three questions map to three real tools in the software world. The first corresponds to a government list of flaws known to be exploited. The second corresponds to a probability score that estimates exploitation over the next thirty days. The third corresponds to the familiar severity score from zero to ten.
The order of trust matters: actual trouble first, probability second, severity third. A teacher’s complaint about a fire door that has already jammed in a drill outranks a leaky tap that might, in theory, flood a room. Keep this in mind when somebody says “fix everything above nine first”. That rule is easy to remember, but it ignores whether anyone is using the flaw.
Where to find a formal description
The name is new, and nobody awards a certificate for it. If you want one reference page that spells out the loop, the Ops Atlas explanation of VulnerabilityOps describes it as tiering, dated owner queues, verified closure and flow measurement, and draws a clear fence around what belongs inside the topic and what belongs to neighbours.
Who does the job, and by when
A good caretaker does not carry every ladder personally. Notes about electrics go to the electrician, plumbing to the plumber, and each note carries a day. Software works the same way: a finding is routed to the team that owns the system, with a due date that depends on how exposed the system is. A system facing the public internet is like a ground-floor window beside the playground; it gets a short deadline. A lab machine on an internal network behind several locked doors gets a longer one.
In mid-2026 a US federal directive turned this idea into a formal rule for federal civilian agencies. Its four inputs are exposure of the system, a spot on the exploited list, whether an attacker could script the attack, and the amount of technical harm possible. The shortest deadline is three days. You do not have to be a federal agency to use the same four questions.
How do we know it is really fixed?
Here our caretaker earns respect. When the plumber says the tap is repaired, the caretaker turns it on and checks. They do not rely on the plumber’s slip. Software teams need the same habit. A ticket marked “done” is only a slip. Real proof is a fresh scan that logs in to the system and no longer finds the flaw, a test of the configuration, a newly built package with a different fingerprint, or a formal decision that the finding was a false alarm.
Some repairs need extra checking. A well-known 2023 case involved a network gateway where attackers had already hijacked logged-in sessions. Installing the update did not remove those stolen sessions, so the instructions included ending all active ones. A caretaker would say: changing the lock does not help if someone already has a copy of the key.
The broken-torch problem
Imagine the caretaker’s torch dies halfway through the evening inspection of the cellar. With no light, they see nothing wrong, tick the page and go home. The report reads “no problems found”. That is exactly what happens when a vulnerability scanner loses its login. It sees less, reports less, and the dashboard glows green. The Atlas material gives an example where a routine change of keys quietly locked the scanner out of 80 machines in one go. A careful team counts how many machines were checked with a working login during the last seven days and raises an alert when that count slips.
Counting the flow, not the pile
At the end of term, the headmistress asks how the caretaker is doing. The wrong answer is “there are 340 open notes”. A better answer compares notes arriving with repairs verified each week. If 100 arrive and 80 are verified, the pile grows by twenty a week, however hard everyone works. The Atlas calls this ratio saturation, and the target is to keep it below one for every tier.
There is a neat bit of arithmetic here as well. If the pile holds 1,300 notes and you finish 600 a month, clearing it takes about 2.2 months, even if no new notes appear. Should your promise to the school be a 60-day clock, you are already at risk. This is the kind of back-of-the-envelope calculation, sometimes called Little’s Law, that you can do in your head and use in any interview.
What happens when two notes disagree
Sometimes the electrician says a flagged fault is harmless because the wire in question is disconnected. A sensible caretaker does not simply believe or dismiss that; they ask for it in writing, have someone else look, and record a date to check again. Software teams call this a dispute or an exception. The finding stays visible, the claim is tested, and an independent person decides. The rule worth borrowing is that nobody marks their own homework: the person who would benefit from the exception is not the one who approves it.
Likewise, the caretaker’s list should never shrink because someone demoted a note quietly. In the Atlas sample policy, analysts may raise the urgency of a finding but never lower it without going through the exception route. It is a small rule, and it keeps the list honest.
Beginner’s reading path
If this explainer made you curious, continue with our pieces on building critical thinking for operations work and on the gaps learners should close early. Both reinforce the same mindset: ask what is true, gather evidence, and write down how you will recognise success.
Five lines to remember
- Know the building plan: keep an inventory with owners.
- Sort by evidence of real trouble, then probability, then severity.
- Give every note a worker and a date.
- Check the repair yourself; slips do not count.
- Compare arrivals with verified repairs, not just the size of the pile.
Once these five lines feel natural, the technical vocabulary is just labels on ideas you already own.

