Tool guideSecurity & Testing

CVE Search – Complete Guide to CVE Records, CVSS Scores, CWE and Vendor Vulnerabilities

What CVE, NVD, CNAs, CVSS, CWE, CPE, CISA KEV and EPSS are, how the NetTools.tech CVE search combines NVD and CVE.org records without merging them, how to use the dashboard, search, vendor and CWE views, and how to read every field.

By NetTools.tech · · 13 min read

A vendor advisory lands in your inbox, a scanner report lists forty CVE IDs, or a colleague asks whether last week's headline vulnerability affects the version you run. Answering that means reading the CVE record, the NVD analysis, the affected version ranges and the references, which live in different places and do not always agree.

CVE Search & Vulnerability Intelligence brings these sources together: the latest CVEs, search by CVE ID, keyword, vendor, product, CWE, severity and date, vendor overviews and weakness (CWE) listings. Each record shows NVD and CNA scores side by side, the CVSS breakdown, weaknesses, affected products and versions, remediation from the sources, references, CISA KEV and EPSS, always with the source of each value.

In simple terms

A CVE (Common Vulnerabilities and Exposures) ID such as CVE-2021-44228 is a unique name for one publicly disclosed vulnerability. A CVE search looks up those records: what the problem is, which products and versions are affected, how severe it is, and where the fix is published. NetTools.tech reads this from the official public databases; it adds no vulnerabilities of its own.

Technical background

Knowing who produces what explains most of the confusing parts of a CVE record:

TermWhat it isWho produces it
CVEThe identifier and the official record (CVE JSON 5) with description, affected products and referencesThe CVE Program, published through CVE Services (CVE.org)
CNACVE Numbering Authority: the organization that assigned the ID, often the vendor (for Log4Shell, Apache)Vendors, CERTs, open-source foundations
ADPAuthorized Data Publisher: adds data to a record after the CNA, e.g. CISA's VulnrichmentCISA and others
NVDNational Vulnerability Database: adds its own CVSS scores, CWE mapping and CPE configurationsNIST (U.S.)
CVSSCommon Vulnerability Scoring System: base-metric vector and a 0–10 scoreSpecification by FIRST; scores by NVD, CNAs, ADPs
CWECommon Weakness Enumeration: the kind of mistake, e.g. CWE-79 cross-site scriptingMITRE
CPECommon Platform Enumeration: product names like cpe:2.3:a:apache:log4j:*:...NVD CPE dictionary
KEVKnown Exploited Vulnerabilities catalog: CVEs with evidence of exploitation in the wildCISA (U.S.)
EPSSExploit Prediction Scoring System: probability of exploitation in the next 30 daysFIRST

CVSS v3.1 vs v4.0. CVSS v3.x uses Attack Vector, Attack Complexity, Privileges Required, User Interaction, Scope and the impact on Confidentiality, Integrity and Availability. CVSS v4.0 drops Scope, adds Attack Requirements (conditions in the target environment), splits User Interaction into None, Passive and Active, and scores impact separately for the vulnerable system (VC/VI/VA) and subsequent systems (SC/SI/SA). Scores from different versions are not comparable: the same vulnerability can be Medium under v3.1 and Low under v4.0. Severity bands are the same for v3.x and v4.0 (Low 0.1–3.9, Medium 4.0–6.9, High 7.0–8.9, Critical 9.0–10.0); CVSS v2 only has Low, Medium and High. See the specifications at FIRST.

Why NVD and CNA scores differ. The CNA knows the product; NVD scores from public information, often assuming the most exposed configuration, possibly in another CVSS version. Neither is "the correct" score, so the tool keeps them separate.

  • Triage advisories quickly: severity, attack vector, KEV status and fixed versions on one page.
  • Check your version: compare it with the CNA's version list and NVD's CPE ranges, shown side by side.
  • Watch what is new: CVEs published in the last seven days or modified in the last 24 hours, by severity, with known-exploited ones separated.
  • Assess a vendor, product or weakness class: counts, severities, weakness types and the newest CVEs.
  • Document decisions: CSV, JSON or print exports that include source and fetch time.

Who should use it?

System administrators and patch managers deciding what to update first, security analysts triaging scanner output and advisories, developers checking dependencies named in a CVE, network engineers following advisories for routers, firewalls and VPN gateways, and auditors who need sourced severity data.

How the CVE search works

Your browser only talks to the NetTools.tech API. The NetTools server queries four fixed public sources; no URL you type is ever fetched:

  • NVD API 2.0 for search, NVD CVSS scores, CWE mapping, CPE configurations, reference tags and the CISA KEV fields NVD carries; the NVD CPE dictionary supplies vendor and product suggestions.
  • CVE.org CVE Services for the CNA record: title, description, affected products with version status, solutions, workarounds, references, CWE and the CNA's CVSS scores. Its organization names are also used to label scores that NVD identifies only by UUID (for example CISA-ADP).
  • MITRE CWE API for weakness names and descriptions, including the names in the "Top weaknesses" charts.
  • FIRST EPSS API for the exploitation probability and percentile.

Values stay attributed to their source; a CNA score with exactly the same vector as an NVD entry is not listed twice. Rejected CVEs are excluded from all searches.

Paging. NVD returns results oldest first, so the server asks for the number of matches and then requests the last slice to show the newest first. Search and CWE pages hold 25 CVEs, up to page 400. The dashboard takes the newest up to 2,000 CVEs published in the last 7 days (Latest) or modified in the last 24 hours (Recently updated); its statistics are computed from exactly those CVEs.

Rate limits. NVD allows 5 requests per 30 seconds without an API key and 50 with one. The server paces NVD requests in a sliding window: 5 per 30 seconds without a key, 45 when the operator has configured an NVD_API_KEY. If a slot would take more than about 25 seconds to free up, the request fails with a "busy" message. Each upstream request has a 25-second timeout and up to three attempts on timeouts, HTTP 429 and server errors.

Caching. Results are cached on the server, and identical concurrent requests share one upstream call:

DataCache time
Dashboard, search pages, CWE lists, related CVEs, result counts30 minutes
"Not found" answers from NVD or CVE.org30 minutes
Single CVE record, vendor overview6 hours
EPSS12 hours
CPE name suggestions24 hours
CWE names and descriptions30 days

Open CVE Search & Vulnerability Intelligence and pick a tab.

  1. Dashboard (default): switch lists with Latest, Critical, High, Medium, Low, Known exploited and Recently updated. The search box stays available above it.
  2. Search: type a CVE ID or keyword. A CVE ID opens the record directly (case and spaces such as cve 2021 44228 are accepted); other text searches CVE descriptions in NVD. Under Filters:
    • Vendor and Product take NVD CPE names (microsoft, http_server), with suggestions after three letters. A product without a vendor is looked up in the CPE dictionary; if several vendors have it, you pick one.
    • CWE, Severity and Severity by (CVSS v3.x or v4.0: which NVD severity field is filtered, and which version the cards headline).
    • CVSS score range, Published and Modified dates (start and end required, at most 120 days apart), Status, Order and Only known exploited (CISA KEV).
  3. Vendor CVEs: enter a Vendor and optionally a Product, or click an example (Microsoft, Cisco, Apache, Oracle, Apple, Google, Linux, Fortinet). Names are normalised to CPE keys: "Apache Software Foundation" becomes apache, "Windows Server" becomes windows_server.
  4. Weakness (CWE): enter CWE-79, cwe79 or 79, or click an example (CWE-79, -89, -787, -22, -78, -502).
  5. Click any CVE, vendor, product or CWE to drill down. Each view has its own URL, e.g. /cve?cve=CVE-2021-44228, and back/forward work.
  6. Export with CSV, JSON, Copy all or Print; on a record, Download normalised record (JSON) saves the combined record with source metadata.

Tip

Score range and Status are not NVD search parameters. They filter only the 25 CVEs on the fetched page, so a page can show fewer results than the total. Use Severity for a server-side filter.

Understanding the results

List views

ResultMeaningWhat to look for
CVE cardID, headline score and severity, CVSS version, shortened description, attack facts, CWE and product chips, dates, NVD status"Known exploited" tag and attack vector
Severity / CVSS score distributionCounts per severity and in the bands 0–3.9, 4–6.9, 7–8.9, 9–10Many "Not scored" means NVD has not analysed recent CVEs yet
Top weaknesses / vendorsMost frequent CWEs (with MITRE names) and CPE vendors in the returned setVendors are CPE names, not company names
TimelineCVEs per publication dayIn Recently updated, modified CVEs are grouped by publication day
Vendor KPIsTotal CVEs in NVD for the CPE match and per-severity countsWithout an NVD key, severity counts cover the newest 200; with a key, totals by CVSS v3.x severity
Source lineEach source as "live", "cached, fetched … ago" or "unavailable"Cached data is not real-time

Record view

ResultMeaningWhat to look for
Score boxHeadline score, severity, CVSS version and scorerA CNA v4.0 score can headline over NVD's v3.1
StatusNVD status: Received, Awaiting Analysis, Undergoing Analysis, Analyzed, Modified, DeferredAnything but Analyzed or Modified means NVD data may be incomplete
Assigner (CNA), CISA KEV, EPSSWho assigned the ID; KEV date added and federal due date; EPSS probability, percentile and dateKEV means exploited in the wild; EPSS is a model estimate
Severity (CVSS)One block per score: version, type (Primary, Secondary, CNA), source, vector, each metric colored by risk, NVD sub-scoresCompare NVD and CNA vectors metric by metric
Attack vector visualisationAccess → Complexity → (Attack requirements, v4.0) → Privileges → User interaction → ImpactBase metrics only; not an exploit chain
Weakness (CWE)CWE ID, MITRE name and description, who assigned itNVD-CWE-noinfo and NVD-CWE-Other are NVD placeholders
Affected productsTrees from the CNA record and NVD CPE configurations: vendor → product → version or range with status. The primary vendor (the CNA's own, else the one with most vulnerable CPEs) is listed first and open; others are collapsed with countsNVD status platform marks a required platform or configuration that is not vulnerable itself
Advisory & remediationCNA solutions and workarounds, fixed versions (from version-like CNA lessThan and NVD versionEndExcluding values; NVD lines primary vendor first, at most 10 plus "N more…"), patch, advisory and mitigation references, CISA's required actionOnly shown when a source provides it; the NetTools general recommendation is labelled separately
Related CVEsOn demand: CVE IDs mentioned in the record, plus newest CVEs for the primary vendor's product and the same CWEThe last two are similarity, not an official relationship
References, SourcesOfficial (CVE.org, NVD, CISA, or tagged Vendor Advisory, Patch, Release Notes, Mitigation, US Government Resource) vs external; each source's fetch time or errorA failed source does not hide the others

Example: reading CVE-2021-44228 (Log4Shell)

Search for CVE-2021-44228 in CVE Search. The values below are from the NVD and CVE.org records saved for the NetTools.tech test suite in October 2026; live records keep changing.

Overview. CNA: Apache. Title: "Apache Log4j2 JNDI features do not protect against attacker controlled LDAP and other JNDI related endpoints". NVD status Analyzed, published 10 December 2021.

Severity. Headline 10.0 Critical, CVSS 3.1, NVD Primary:

CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H
Attack Vector: Network · Attack Complexity: Low · Privileges Required: None
User Interaction: None · Scope: Changed · C/I/A: High / High / High
Exploitability sub-score 3.9 · Impact sub-score 6

A Secondary CVSS 3.1 score with the same vector comes from CISA-ADP, and NVD's CVSS 2.0 score is 9.3 High (AV:N/AC:M/Au:N/C:C/I:C/A:C). Apache's CNA record has no CVSS score (only a non-CVSS "critical" rating, which is not displayed).

KEV. Listed as "Apache Log4j2 Remote Code Execution Vulnerability", added 2021-12-10, federal due date 2021-12-24; CISA's required action (apply updates or remove affected assets) appears under remediation.

Weaknesses. The CNA assigned CWE-502 (Deserialization of Untrusted Data), CWE-400 (Uncontrolled Resource Consumption) and CWE-20 (Improper Input Validation); NVD added CWE-917. The "Assigned by" line shows who said what.

Affected products. Apache is the primary vendor and comes first. NVD's ranges for Log4j are "from 2.0.1 before 2.3.1", "from 2.4.0 before 2.12.2" and "from 2.13.0 before 2.15.0", plus single versions such as 2.0 beta9 and 2.0 rc1. Products of eleven other CPE vendors that ship Log4j, including Siemens, Cisco, NetApp, Intel and Debian, appear in collapsed groups. The CNA describes the versions with a custom scheme, shown as published ("2.0-beta9 – < log4j-core*"); because that is not a version number, it produces no fixed-version line. After CISA's required action, the NVD fixed-version lines start with the three Apache Log4j upgrade targets (2.3.1, 2.12.2, 2.15.0) before other vendors' products.

Decision. Compare your exact Log4j version with the NVD ranges and follow the Apache advisory in the official references. KEV plus a network vector without privileges or user interaction puts it at the top of the queue.

Common use cases

  • Morning check: review Critical and Known exploited on the dashboard.
  • Scanner triage: look up each CVE and attach the record JSON to the ticket.
  • Vendor patch day: review a vendor's Latest and Recently updated lists.
  • Training: list CVEs for CWE-79 or CWE-89 as real examples.

Troubleshooting

Keyword search matches NVD descriptions, so a product name the text does not mention is not found. Use fewer filters, another spelling, or the CPE name in Product (http_server instead of "Apache HTTP Server"). Score range and status only filter the fetched page.

Vendor name vs CPE name

"No CVEs in NVD match the CPE vendor name …" means the normalised key is not in NVD's CPE data; company and CPE names often differ (paloaltonetworks, redhat). The tool lists matching names from the CPE dictionary to pick from.

"Awaiting Analysis" and no products or score

NVD has not analysed the CVE yet (Received, Awaiting Analysis, Undergoing Analysis or Deferred), and the Affected products section says so. Use the CNA tree and score if published, or the vendor advisory. Vendor search cannot find such CVEs.

"The vulnerability database is busy" or a source "is not responding"

The server stays within NVD's rate limit and refuses requests that would wait too long; "is not responding right now" means a source failed after three attempts. Retry after a minute. On a record page a failed source is marked "unavailable" and the others are still shown.

Published and modified ranges are limited to 120 days by NVD, and need both dates. Split longer periods.

The data looks out of date

The source line shows "cached, fetched … ago" when the result came from the cache (30 minutes for lists, 6 hours for records). A CVE missing from NVD is re-checked after 30 minutes. For the very latest state, use the NVD and CVE.org links under Sources.

"CVE-… was not found in NVD or the CVE List"

The ID may be reserved, rejected or mistyped. CVE IDs are CVE-YYYY- followed by four to seven digits.

Common mistakes

  • Treating CVSS as risk. The base score describes the vulnerability in general; your risk depends on affected versions, exposure, controls, KEV and EPSS.
  • Comparing scores across CVSS versions. v4.0 and v3.1 scores for one CVE can land in different bands.
  • Ignoring version ranges. "Apache Log4j" being listed does not mean every version is affected; read the exact range.
  • Reading a platform entry as vulnerable. NVD entries marked "platform" are the required environment, not the vulnerable component.
  • Assuming inferred related CVEs are official. Only IDs mentioned in the record come from the source; same-product and same-weakness lists are NetTools similarities.
  • Reading "no remediation listed" as harmless. It only means the sources have published no fix yet.

Limitations

  • Only as good as the sources: NVD analysis lags behind publication, CNAs differ in detail, records change.
  • Vendor views depend on NVD CPE data, so unanalysed CVEs are not found.
  • Statistics describe the returned set: up to 2,000 CVEs on the dashboard, the newest 200 for a vendor, the current page elsewhere.
  • Severity filters use one CVSS version (v3.x unless you choose v4.0).
  • Base metrics only; CNA version data is shown as published, including custom schemes.
  • No asset matching and not real-time (see the cache times above).

Security & privacy

The CVE search is read-only intelligence: it never runs exploit code, scans hosts or connects to affected systems, and it links to references rather than loading them. Search terms go to the NetTools.tech API, which forwards the relevant parts (CVE ID, keyword, CPE name, CWE, filters) to NVD, CVE.org, MITRE CWE or FIRST EPSS from the server; your browser does not contact those services. The optional NVD API key is set by the site operator on the server and never reaches the browser. There is nothing secret to enter, so do not paste internal hostnames, configuration or credentials into the search. See the privacy policy.

Summary

The CVE search combines NVD's analysis, the CNA's official record, MITRE CWE names and FIRST EPSS into one view that keeps every value attributed to its source. Use the dashboard for what is new, Search and the vendor and CWE views for what concerns you, and the record page for versions, fixes and exploitation signals. Treat CVSS as one input and set priorities with your own inventory, exposure, KEV and EPSS.

Frequently asked questions

What is the difference between CVE and NVD?

CVE is the naming system run by the CVE Program: a CNA assigns the ID and publishes the official record with its description and affected products. NVD is a separate U.S. government database that adds its own analysis on top, including CVSS scores, CWE mapping and CPE product configurations.

Why do NVD and the CNA show different CVSS scores?

They are independent assessments. The CNA usually knows the product best, while NVD analysts score from the public information and may assume a worst case; they can also use different CVSS versions. The tool shows every score with its source and version instead of picking one silently.

Which score does the tool show in the headline?

Normally the newest CVSS version available, in the order 4.0, 3.1, 3.0, 2.0, preferring NVD's Primary score within that version. When a severity filter is active, the result cards headline the version the filter uses (CVSS v3.x by default, v4.0 if chosen), so a Critical search does not show cards labelled with another version's severity.

Is a CVSS 10.0 vulnerability always my top priority?

Not necessarily. CVSS describes how severe a vulnerability is in general, not how likely it is to be exploited or how exposed your systems are. Combine it with whether you run an affected version, whether the service is reachable, the CISA KEV flag and the EPSS probability.

Why does a new CVE have no score or affected products?

NVD analyses CVEs after they are published, and records waiting in its queue show statuses such as Received, Awaiting Analysis or Deferred. Until then only the CNA's data from CVE.org may be available, and if the CNA published no CVSS score the record is shown as not scored.

Why does a vendor search find nothing for a company I know?

Vendor search matches NVD CPE vendor names, which are short lower-case keys such as microsoft or apache, and only CVEs that NVD has analysed with CPE configurations can match. If nothing is found, the tool lists matching names from the NVD CPE dictionary to pick from.

What do CISA KEV and EPSS add?

The KEV flag means CISA lists the CVE in its Known Exploited Vulnerabilities catalog, so exploitation in the wild has been observed; the tool shows the date added, the federal due date and the required action. EPSS from FIRST is a modelled probability that the CVE will be exploited in the next 30 days, with a percentile.

How current is the data?

Dashboard, search, CWE and related-CVE lists are cached on the NetTools server for up to 30 minutes, vendor overviews and single records for up to 6 hours, EPSS for 12 hours and CWE names for 30 days. A record that was not found is only cached for 30 minutes, and every view says when it was fetched and whether it came from the cache.

Does the tool scan my systems for these vulnerabilities?

No. It is read-only vulnerability intelligence: it queries public databases and never connects to your systems, runs exploit code or checks installed versions. Matching CVEs to your inventory is up to you.