Remediation
Remediation is the findings queue aggregated by fix, and beside it the same work seen through the model. The page has three tabs: Work items, the queue this page describes first; Remediation options, what to fix next ranked by the projected change of a modeled figure; and Remediation scenarios, what was proposed and whether it verified.
A single missing patch can produce findings on forty hosts. The findings queue shows forty items, correctly - each is fixed and confirmed on its own. The remediation queue shows one piece of work with forty instances behind it, which is how the person doing it thinks about the job.
Same underlying data, grouped for a different question:
| Queue | Answers |
|---|---|
| Findings | What is wrong, and where? |
| Remediation | What should we do, and how much does each thing clear? |
Findings and vulnerabilities, one project
A remediation project consolidates the work behind one fix across both findings and vulnerabilities that share its root cause. The count on each project is the combined total, with the finding and vulnerability shares shown beneath it, so a project that clears twelve findings and eight not-yet-triaged vulnerabilities reads as twenty pieces of work, not two separate numbers.
A vulnerability that has been promoted to a finding is counted once, as a finding, so the total never double counts the same issue. Both records stay linked to the project, so you can trace a project back to every finding and vulnerability it addresses.
Each project is matched to the issues it applies to from structured criteria the project carries, not from its title. The same matching serves findings, vulnerabilities, and future exposure types from one shared catalog, and a new project added through the catalog is matched for qualifying issues immediately, with no release required.
When nothing fits
Not every issue maps cleanly onto a project. When the evidence is not specific enough to assign one confidently, CyberOptix CTEM Platform does not guess: the issue is surfaced for review rather than attached to an unrelated project. A recurring gap there is the signal that a new remediation project is genuinely needed.
Why work from this view
It surfaces leverage. Sorting by how many findings a remediation clears puts the highest-value work at the top, and it is frequently not what severity ordering would have chosen. One dependency upgrade closing sixty findings beats one configuration change closing one, even if the second is a Critical.
It matches how fixes actually happen. Nobody patches a library forty times. They patch it once and roll it out.
It makes estimating possible. "Upgrade this library across twelve services" is a thing you can size. "Forty findings" is not.
Working an item
A remediation carries what to do and which findings it resolves. The sequence:
- Pick by leverage, not only by severity.
- Check the instances. Forty findings is not always one job - some assets differ enough to need separate handling, and that is visible before you commit.
- Move the findings to a remediation status. See Findings and triage.
- Apply the fix.
- Let a scan confirm it. Findings close when a subsequent scan stops seeing them, not when somebody says the work is done.
That last step is the one to respect. A remediation marked complete on trust is indistinguishable from one that silently failed on eight of forty hosts.
Partial completion
A rollout that reaches most instances but not all is the normal case, not an exception.
Because each finding closes independently as scans confirm it, this resolves itself: findings on fixed hosts close, findings on missed hosts stay open, and what remains is visible rather than buried under a remediation someone ticked off. The stragglers are usually hosts that were down, out of scope, or drifted from their configuration - all worth knowing about separately.
Remediation and SLAs
SLA deadlines attach to findings, not to remediations. A remediation clearing forty findings clears forty separate deadlines, each computed from when its finding was opened.
So a remediation is not overdue; its findings are. If you are prioritising by deadline pressure, sort the findings and work back to the remediation that clears the most of them. See Remediation and SLAs.
Remediation options
The Remediation options tab projects, one exposure at a time, what would change if that exposure were remediated, for the exposures that stand on a path. It ranks them by the change of one figure you choose: breach probability, lateral reach, crown jewels, or financial exposure. Every other figure's change is shown beside the ranked one, with the asset, the weakness, how many of its detections are open, its SSVC remediation priority, the remediation catalog entries linked to its findings, the crown jewels it reaches, and whether a compensating control is stated on it. State control opens the Risk Posture page on that exposure to state one.
Choosing the figure matters. Ranked by breach probability, a weakness that only an attacker already inside could use shows no change; ranked by lateral reach or crown jewels, the same weakness can be the largest change on the list. Neither ranking is wrong; they answer different questions.
A projection is a projection. It assumes the action is fully effective and that nothing outside the relationships the platform knows is affected, and every projection says so; a compensating control you have stated is part of the figures on both sides. The changes of two rows are never added together: remediating both is a different projection, not the sum. Where a build was cut by a bound, the comparison is marked as qualified. Both sides of a projection are calculated the same way: where the figures before the action could not be combined exactly within the budget, the figures after it are combined by the same procedure, and the row says the comparison is approximate. A change is never the difference between an exact figure and an approximate one.
Remediation scenarios
Propose on a remediation option records a scenario on the Remediation scenarios tab: the action, the exposures it resolves to, the findings behind them, and the projection against the figures at that moment. Proposing stores the projection; it changes nothing about the environment and nothing about the figures.
A scenario's status is derived from what happened, never set by hand:
| Status | Means |
|---|---|
| Proposed | Recorded with its projection |
| Implemented | Every finding the scenario names has reached closed in the findings workflow. A finding reopened returns the scenario to proposed |
| Verified | Every exposure the scenario names is resolved on evidence the platform recognises, listed below |
| Superseded | Set aside by a person; everything it recorded stays. A verified scenario is history and cannot be superseded |
Implemented is a fact about the workflow. Completing the work does not establish that the exposure is gone: a closed finding whose scanner still reports the weakness keeps the exposure open and the scenario unverified. The scenario then says why, per exposure: the weakness is still reported and was re-observed since the proposal; the weakness is still listed but nobody has looked again; the finding was closed without a retest; the relationship has not been revoked; nothing has tried the hop since the revocation. An endpoint security detection counts as re-observed when the vendor first reported it, or reopened it, after the proposal; one that has stayed open is still listed, whatever else changed on its record.
The evidence that verifies:
| Evidence | Meaning |
|---|---|
| Retest closed the finding | A tester or an automated validation retested and closed the finding |
| Scanner ran the check again and no longer reports it | The Scanner marked the weakness fixed after the proposal, which it does only when it ran the check again |
| Endpoint security closed the detection | The endpoint security vendor closed the detection after the proposal. The vendor closes a detection when its sensor no longer finds the vulnerable software; a suppression is not a closure, and a detection the vendor merely stopped returning is not one either |
| A validation attempt after the revocation failed | For a revoked relationship, an attempt to use it after the revocation failed |
Verification is scoped to the exposures and relationships the scenario named. It does not establish that every attack path they touched is gone, nor anything about the business consequences.
Projected, observed, verified
Three different things, kept apart on the page:
- Projected is what the scenario's projection said would change. It is never an outcome.
- Observed is the difference between the figures at the scenario's baseline and the figures at the snapshot that verified it. The page lists everything that moved the figures in between (rebuilds from new telemetry, inputs of yours, model changes, other scenarios verified in the window) and attributes the change to all of them. It never attributes the whole change to the scenario.
- Verified names the exposures that were shown to be gone, with the evidence. It is not a number, and it is not a claim about effectiveness beyond those exposures.
The latest verified scenario is shown with its evidence, its observed figures before and after, what else moved them, and the model versions on each side. When the two snapshots are under different model versions, or were calculated by different methods, the page says the figures are not reliably comparable and shows them side by side; no difference between them is then read as an improvement or a deterioration.
Export
The queue exports to XLSX, which is the usual way this reaches people who do not use the platform - a patching team, a third-party supplier, a change advisory board.
API
The endpoints behind this section are in the API reference.
Next
- Findings and triage - the per-instance view
- Remediation and SLAs - how deadlines work
- Remediation SLA policy - setting them