cve.bz

About cve.bz

cve.bz publishes a vulnerability database as its sources publish it: every CVE with the scores that decide what to fix first. No sign-in, no key.

On this page

Sources

NVD Descriptions, CVSS scores, CWEs and affected products

The NIST National Vulnerability Database: descriptions, CVSS 4.0 and 3.x vectors and scores, CWE ids, affected products and references. Checked every 15 minutes; updated 14 minutes ago.

CVE List New CVEs within minutes, with the CNA's scores and CISA's SSVC

The CVE Program's records: a CVE minutes after it is published, before the NVD has analyzed it, with the CNA's own description, scores and affected and fixed versions, and CISA's Vulnrichment: its SSVC assessment (exploitation, automatable, technical impact). Checked every 10 minutes; updated 3 minutes ago.

FIRST EPSS The probability of exploitation in the next 30 days

The Exploit Prediction Scoring System: for every CVE, the probability of exploitation in the next 30 days and its percentile, scored daily, with the scores of a week and 30 days before. Checked every 6 hours (FIRST publishes it once a day); updated 2 hours ago.

CISA KEV The CVEs known to be exploited

The Known Exploited Vulnerabilities catalog: the CVEs CISA has seen exploited, with the date added, the federal due date and known ransomware use. Checked every hour; updated 1 minute ago.

OSV Package advisories and the Linux distributions' fixes

Open Source Vulnerabilities: package advisories with affected version ranges and fixes, behind check your stack, and the Linux distributions' fixed versions per release (Debian, Ubuntu, Red Hat, SUSE, openSUSE, AlmaLinux, Rocky Linux, Alpine, Azure Linux and BellSoft's Alpaquita) with their advisories. Checked every 30 minutes; updated 14 minutes ago.

Red Hat Severity, statement and fix state per product

Red Hat Security Data: Red Hat's severity rating, statement and mitigation for a CVE and the fix state of each of its products' packages, with the advisories that fixed them. Checked every hour; updated 12 minutes ago.

CISA ICS advisories Industrial control system and medical device advisories

CISA's advisories for industrial control systems and medical devices, as CSAF: for a CVE, each advisory's id, title, date, critical infrastructure sectors and the products it lists as affected and fixed, linked to CISA's page. Checked every hour; updated 12 minutes ago.

Public exploits Links to Exploit-DB and Metasploit

Links to public exploit code for a CVE from two indexes: the Exploit Database (by OffSec) and Rapid7's Metasploit Framework modules (BSD-3-Clause). Only the link, title and date are stored; the code stays on their sites. Checked every 6 hours; updated 25 minutes ago.

Detection rules Nuclei templates, Emerging Threats and Snort rules

ProjectDiscovery's nuclei template (MIT) that checks a system for a CVE, and the intrusion detection rules that name it, from two rule sets: Emerging Threats Open (BSD) and the Snort community rules (GPL-2.0). Only a rule's id, name and date are stored, never its text; read with the exploit indexes.

GitHub PoC Repositories named after a CVE, with Harbor Check

Repositories on GitHub whose name carries a CVE id, found with GitHub's repository search, and, once Harbor Check has found their code to match, those the PoC-in-GitHub index (by nomi-sec, CC0) lists for a CVE: the repository's name, description, stars, language and dates, never its code. Harbor Check reads a repository's code against the CVE: one that implements the exploit carries a mark, and one found to be aimed at whoever runs it is left out. The code was read, not run. Listed apart from the public exploits and not counted as one.

MITRE CWE Weakness names, descriptions and attack patterns

The Common Weakness Enumeration: the name, common name and short description of each weakness a record names, as MITRE wrote them, with a link to MITRE's full entry, and the attack patterns MITRE's CAPEC relates to it (their names, likelihood and severity). Checked every week; updated 5 days ago.

Every source's notice, license and link are on the credits page.

Each source is checked on its own schedule for what changed since the last check, and a new record is served as soon as it is stored. A record's page shows the dates its sources carry and when it was last synced; the home says how long ago the freshest source synced, or, while the first import of the NVD or the CVE List runs, how many records are stored so far. Pages are cached for five minutes, so a page can be up to five minutes behind the database.

What this site is not

  • Not affiliated with the CVE Program, MITRE, NIST, FIRST, CISA, OSV, OffSec, Rapid7, ProjectDiscovery, Red Hat or the Linux distributions. The data is theirs, under their terms, with their notices and licenses on the credits page.
  • Not advice. A score is the source's score; what to fix first depends on where a product or package runs, which this site cannot know.
  • Not written by a model. Every description and score is the record as published; nothing here is generated, summarized or rewritten.

Who runs it

cve.bz is run by Tarnover, the company behind Flawharbor, a vulnerability management platform. cve.bz, its API, feeds and MCP server are covered by Flawharbor's Terms of Service and Privacy Policy.

In your browser

Type cve.bz/CVE-2024-3094 in the address bar to open a record. Browsers also offer cve.bz as a search engine (OpenSearch): in Chrome and Edge type cve.bz and press Tab, then a CVE id, product or keyword; in Firefox right-click the address bar and choose Add "cve.bz".

Or drag this bookmarklet to your bookmarks bar: Open in cve.bz. Select a CVE or GHSA id on any page and click it; with nothing selected it asks for one. Some sites' security policies block bookmarklets.

Feeds

The newest records as RSS 2.0 at /cve/feed.rss and as JSON Feed 1.1 at /cve/feed.json, 100 records each. The list's filters apply to both:

https://cve.bz/cve/feed.rss?kev=true&severity=critical

is the critical CVEs in CISA's KEV catalog.

API

The same records as JSON: no key, cacheable for five minutes, the sources' notices in every answer. Every route with its parameters is in the API reference, and as OpenAPI 3.1 at /api/public/openapi.json. The list takes the pages' filters (q, kev, exploit, epss, severity, product, cwe (such as CWE-79, the records naming exactly it), owasp (such as A05:2025, the records naming a CWE in that OWASP Top 10 category), limit up to 100) and answers a next cursor for the page after:

GET https://cve.bz/api/public/cves?severity=critical&limit=20

One record, with its CVSS vectors, EPSS score, KEV listing, public exploits, CWEs, affected products and references:

GET https://cve.bz/api/public/cves/CVE-2024-3094

The figures behind Insights (new CVEs per day by severity, top vendors, products and CWEs, the newest KEV additions and EPSS scores), counted again every 15 minutes:

GET https://cve.bz/api/public/insights

Check your stack as JSON: the lockfile or SBOM itself is the body, its file name the query:

curl --data-binary @package-lock.json "https://cve.bz/api/public/check-stack?name=package-lock.json"

MCP server

Coding agents can query the same data over the Model Context Protocol at https://cve.bz/mcp (Streamable HTTP, no key), with three tools: get_vulnerability (a CVE, OSV or GHSA id), list_feed (the newest records, with the list's filters) and check_stack (a lockfile or SBOM, stored nowhere). A stack check, here or as the check_stack tool, counts against one budget per address: 10 at once, then 30 a minute; over it the API answers 429 with Retry-After and the tool an error naming the seconds to wait. Add it to an agent's MCP configuration as:

{"type": "http", "url": "https://cve.bz/mcp"}

Install the skill

The same instructions come as an Agent Skill, cve-bz: download cve-bz.zip and unzip it into your agent's skills folder, such as ~/.claude/skills/ for Claude Code.

Verify the download

Before you unzip it, compare its SHA-256 with this one, also at /skills/cve-bz.zip.sha256:

6cbf63664840f2257b444cd9a473a64e7c1ef68a64bef2a5bcc0f1b0e51c7582
sha256sum cve-bz.zip      # shasum -a 256 cve-bz.zip on macOS

Each version is also a signed GitHub release of the same zip, with a Sigstore attestation from the repository's release workflow. To verify a download:

gh attestation verify cve-bz.zip --repo Flawharbor/cve-bz-skill

Or ask your agent to check it and install the skill:

Download https://cve.bz/skills/cve-bz.zip, check that its SHA-256 is 6cbf63664840f2257b444cd9a473a64e7c1ef68a64bef2a5bcc0f1b0e51c7582, and install the skill only if it matches.

In Claude Code it also installs as a plugin, from the Flawharbor/cve-bz-skill marketplace:

/plugin marketplace add Flawharbor/cve-bz-skill
/plugin install cve-bz@flawharbor

Check your stack

Check your stack takes a lockfile (package-lock.json, poetry.lock, Cargo.lock, go.sum, Gemfile.lock, composer.lock, Package.resolved) or a CycloneDX or SPDX SBOM and lists the packages the stored OSV advisories affect, with severity and fixed version. The file is matched in memory and discarded: no table, log line or cache holds it.