The Three Evaluation Decisions for Every Vulnerability
This guide is based on the FedRAMP Consolidated Rules for 2026, version 2026.11.02.01, and the official schemas dated 2026-06-24. Rule identifiers are given in parentheses so you can check each statement.
These rules take effect on December 7, 2026, for obtaining and for keeping a certification. Providers have a grace period until March 7, 2027. Adoption has been optional since July 4, 2026. (VER info, VDR info)
What this guide covers
Every detected vulnerability gets three decisions from the provider. This guide explains each one, the definitions behind them, and one rule about automation. Deadlines are in guide 2. Report fields are in guide 3.
Terms are explained once, where they first appear. Rule identifiers are in parentheses after each statement.
The core requirement
Providers MUST evaluate detected vulnerabilities, “considering the context of the cloud service offering”. (VER-EVA-ELX, VER-EVA-EIR, VER-EVA-EPA)
Detection itself is covered by VDR-CSO-DET; this guide starts after a vulnerability has been detected. (VDR-CSO-DET)
Decision 1: likely exploitable or not
Providers MUST decide whether each detected vulnerability is a likely exploitable vulnerability (LEV). (VER-EVA-ELX)
The definition has three parts. A vulnerability is an LEV when all of these hold: (FRD: Likely Exploitable Vulnerability (LEV))
- It is not fully mitigated.
- A likely threat actor can reach it.
- A likely threat actor who knows about it would likely gain unauthorized access, cause harm, disrupt operations, or otherwise have an undesired adverse impact in the offering.
“Likely” means “a reasonable degree of probability based on context.” (FRD: Likely)
The definition has a floor. Any vulnerability that “an automated unauthenticated system can exploit over the internet” is an LEV. (FRD: Likely Exploitable Vulnerability (LEV), note)
The opposite is a Not Likely Exploitable Vulnerability (NLEV). (FRD: Likely Exploitable Vulnerability (LEV), note)
The rule’s notes say that exploitation of most scanner findings needs unrealistic circumstances. They also say that a provider that rates vulnerabilities as not likely exploitable “without careful consideration” is more likely to suffer an adverse impact, and that reckless or deliberate misrating has a negative impact on the provider’s certification. (VER-EVA-ELX, notes)
Decision 2: internet-reachable or not
Providers MUST decide whether each detected vulnerability is an internet-reachable vulnerability (IRV). (VER-EVA-EIR)
An IRV is a vulnerability in a machine-based information resource that might be exploited or triggered by a payload from a source on the public internet. (FRD: Internet-Reachable Vulnerability (IRV))
A machine-based information resource is one that relies mainly on computers to operate, such as hardware, software and cloud services. (FRD: Machine-Based (Information Resources))
Two points from the definition notes:
- A resource with no direct route to or from the internet still counts if it receives payloads, or acts on them, because of internet activity. (FRD: Internet-Reachable Vulnerability (IRV), notes)
- Reachability applies only to the specific vulnerable resource that processes the payload. (FRD: Internet-Reachable Vulnerability (IRV), notes)
The rule’s notes say that once triggering payloads are intercepted, filtered, rejected or otherwise deflected before the vulnerable resource processes them, the vulnerability “should no longer be considered” internet-reachable. (VER-EVA-EIR, notes)
The opposite is a Not Internet-Reachable Vulnerability (NIRV). (FRD: Internet-Reachable Vulnerability (IRV), notes)
Decision 3: the Potential Agency Impact rating
Providers MUST estimate the likely potential agency impact of exploitation on government customers and assign one N-rating, called the Potential Agency Impact N-rating (PAIN). (VER-EVA-EPA)
Potential Agency Impact is the estimated combined effect on all agencies using the service. (FRD: Potential Agency Impact)
| Rating | Meaning in the rule |
|---|---|
| N0 | Exploitation is extremely unlikely to have any adverse effects on agencies (VER-EVA-EPA) |
| N1 | Minimal customer effects on one or more agencies (VER-EVA-EPA) |
| N2 | Narrow customer effects on one or more agencies (VER-EVA-EPA) |
| N3 | Disruptive customer effect on one agency (VER-EVA-EPA) |
| N4 | Debilitating customer effect on one agency, OR disruptive customer effect on more than one agency (VER-EVA-EPA) |
| N5 | Debilitating customer effect on more than one agency (VER-EVA-EPA) |
The customer effects are defined terms: (FRD)
| Effect | Definition (shortened) |
|---|---|
| Minimal | Noticeable only by some users, such as reduced performance (FRD: Minimal Customer Effect) |
| Narrow | Interrupts use for some users for under 12 hours, or compromises an extremely limited amount and type of federal customer data (FRD: Narrow Customer Effect) |
| Disruptive | Interrupts use for many users for under 24 hours, or compromises large amounts or many types of federal customer data (FRD: Disruptive Customer Effect) |
| Debilitating | Interrupts use for most users, or compromises the integrity or confidentiality of most federal customer data (FRD: Debilitating Customer Effect) |
If the adverse effect is unknown, it “should be treated as if it is debilitating until proven otherwise.” (FRD: Debilitating Customer Effect)
The rating changes over time. Mitigation can lower it, and the reports track the current rating and each reduction. (VER-RPT-VDT)
Evaluation factors
Providers SHOULD consider at least eight factors: criticality, reachability, exploitability, detectability, prevalence, privilege, proximate vulnerabilities, and known threats. (VER-EVA-EFA)
Providers SHOULD also check for false positives and look for logical groupings of affected resources. (VER-EVA-EFP, VER-EVA-GRV)
The “assume automation” rule
Providers MUST assume that exploitation of vulnerabilities can be automated unless they have evidence proving otherwise. (VER-EVA-AIA)
Walk-through with two invented examples
Both examples are invented for this guide. They are not drawn from the rules. The rules leave the judgement calls to the provider, so the answers below show how the decisions fit together, not what a provider must conclude.
Example A (invented): file-parsing flaw
A Class C provider detects a flaw in a library used by a service that parses files uploaded through a public web form.
- Internet-reachable? A payload from the public internet reaches the service that processes it, so this is an IRV. (VER-EVA-EIR, FRD: Internet-Reachable Vulnerability (IRV))
- Likely exploitable? The provider has no evidence that exploitation cannot be automated, so it must assume it can be. (VER-EVA-AIA) If the flaw can be triggered by an automated unauthenticated upload, the floor in the definition makes it an LEV. (FRD: Likely Exploitable Vulnerability (LEV))
- Rating? Suppose the provider estimates that exploitation could debilitate one agency’s use: N4. (VER-EVA-EPA)
Result: IRV, LEV, N4. Guide 2 shows what the deadline tables say for this combination.
Example B (invented): admin tool flaw
A Class B provider detects a flaw in an internal reporting tool. A scan flags it, but the vulnerable code is installed and never loaded.
- Is it a false positive? A vulnerability that is not present in an exploitable state, including code that is installed but not running, can be a false positive. A fully mitigated or remediated vulnerability cannot also be a false positive. (FRD: False Positive Vulnerability; VER-EVA-EFP)
- If the provider instead finds the code is loaded but reachable only from a private network, it is not internet-reachable unless internet activity triggers it. (FRD: Internet-Reachable Vulnerability (IRV))
- If the provider judges that a likely threat actor could not reach the flaw or would not gain an adverse effect from it, it is not an LEV. (FRD: Likely Exploitable Vulnerability (LEV))
- Rating? The rating is a separate decision about the effect on agencies if exploitation did happen. Suppose the provider estimates narrow effects: N2. (VER-EVA-EPA)
Result: NIRV, NLEV, N2. Guide 2 shows the matching deadline.