shaw276/kevscope-cve-kev-epss-daily

Kevscope: CVE + CISA KEV + EPSS, joined

Which vulnerabilities should you patch first? Kevscope puts the three public signals side by side, already joined on the CVE ID:

• CISA KEV: the US government's catalog of vulnerabilities known to be exploited in the wild, with due dates and ransomware use;
• EPSS (FIRST): the probability that a CVE will be exploited in the next 30 days, for every published CVE;
• NVD: CVSS severity (v4.0 → v3.1 → v3.0 → v2, best available), CWE weaknesses, affected vendor:product pairs and the description.

This build (see meta.json): 1,721 KEV entries (catalog 2026.09.23; 361 with known ransomware use, 46 added in the last 30 days), 378,156 EPSS scores (model v2026.06.15, scores of 2026-09-23) and 14,363 CVEs published in the last 30 days (2026-08-25 to 2026-09-24; 1,665 rated CRITICAL, 24 already in KEV).

One join in, one join out: no API keys, rate limits or three different formats.

Configs

config	rows	one row per
kev (default)	1,721	CISA KEV entry, with EPSS and NVD CVSS added
recent	14,363	CVE published in the last 30 days (NVD), with EPSS and a KEV flag
epss	378,156	CVE with an EPSS score, with a KEV flag

Use it

from datasets import load_dataset
kev = load_dataset("shaw276/kevscope-cve-kev-epss-daily", "kev", split="train").to_pandas()
recent = load_dataset("shaw276/kevscope-cve-kev-epss-daily", "recent", split="train").to_pandas()

# Exploited in the wild AND used by ransomware, newest first
print(kev[kev.known_ransomware_use == "Known"][["cve_id", "vendor_project", "product", "date_added", "epss"]].head(20))

# New this month, critical, not yet in KEV, highest exploitation probability
hot = recent[(recent.cvss_severity == "CRITICAL") & (~recent.in_kev)].sort_values("epss", ascending=False)
print(hot[["cve_id", "cvss_score", "epss", "vendor_products"]].head(20))

Or straight from DuckDB: SELECT * FROM 'hf://datasets/shaw276/kevscope-cve-kev-epss-daily/data/kev.parquet' WHERE vendor_project = 'Microsoft'.

Columns

kev: cve_id, vendor_project, product, vulnerability_name, date_added (to KEV), due_date (US federal remediation deadline), known_ransomware_use (Known/Unknown), short_description, required_action, cwes (;-separated), epss, epss_percentile, cvss_version, cvss_score, cvss_severity, cvss_vector, nvd_published, nvd_status, days_published_to_kev (days between NVD publication and the KEV listing).

recent: cve_id, published, last_modified, nvd_status (Received, Awaiting Analysis, Analyzed...), source (the CNA that assigned it), description, cvss_version, cvss_score, cvss_severity, cvss_vector, cvss_source (NVD's own Primary score preferred over the CNA's), cwes, vendor_products (vendor:product from NVD CPE data, first 10; empty until NVD analyses the CVE), reference_count, epss, epss_percentile, in_kev, kev_date_added, nvd_url. Rejected CVEs are left out.

epss: cve_id, epss (0–1 probability of exploitation activity in the next 30 days), epss_percentile, in_kev, score_date.

Notes: brand-new CVEs often have no EPSS score yet (350 of the 30-day rows) and no CVSS until NVD or the CNA scores them (997 rows); those fields are null, never guessed. EPSS is a probability, not a verdict; KEV is evidence of exploitation. Use both.

Refresh

Rebuilt from the three official feeds (meta.json has the exact build time, the KEV catalog version and the EPSS score date). The build is a plain Python script with direct downloads; no scraping.

Licence and attribution

Each part keeps its source's terms:

• CISA KEV: Creative Commons CC0 1.0 (per CISA's KEV licence). No CISA logo or endorsement implied.
• NVD data (CVSS, CWE, CPE): a US government work, public domain in the US. CVE descriptions are © The MITRE Corporation, used under the CVE Program Terms of Use, which allow reproduction and redistribution.
• EPSS: scores from the Exploit Prediction Scoring System (EPSS), FIRST.org, https://www.first.org/epss/. FIRST publishes them free to use and asks for attribution, which this line provides. Model and score date are in meta.json.

This dataset: compiled and joined by CyberMax. Not affiliated with CISA, NIST, MITRE or FIRST. Provided as is, without warranty; check the official sources before acting on a single row.

Also from CyberMax

• DomainDNA: tech stack, email security & domain age API: find out which vendors and products a company's website runs on, then match them against the KEV table above.
• More tools for AI agents and data teams: https://apify.com/cybermax