Vulnerabilities / guides
Is this CVE actively exploited?
"Is this CVE actually being exploited?" is the question that decides whether you patch tonight or next cycle. Three different signals answer it, they disagree more often than you would expect, and all three are queryable for free.
The three signals
CISA KEV - confirmed exploitation. The Known Exploited Vulnerabilities catalog lists CVEs where CISA has evidence of exploitation in the wild. It is the strongest signal and the slowest one: a CVE only lands there after exploitation is confirmed, which can be days or weeks into a campaign. Roughly 1,400 CVEs out of 250,000+ ever published make the list.
EPSS - predicted exploitation. FIRST's Exploit Prediction Scoring System gives every CVE a daily probability (0 to 1) of exploitation in the next 30 days. It is a prediction from a model, not an observation: a 0.97 EPSS means "very likely", not "confirmed". Useful for ranking a backlog; not proof.
Live reporting volume. Neither KEV nor EPSS tells you whether a campaign is running right now. Security vendors and researchers publishing about in-the-wild exploitation today is the earliest public signal there is - it is what KEV eventually confirms.
A CVE can be high-EPSS but absent from KEV (predicted, unconfirmed), in KEV with a low EPSS (old, confirmed, quiet), or all over this week's reporting before either catches up. Check all three.
Checking each signal
KEV is a public JSON file, no key:
curl -s https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json \ | jq '.vulnerabilities[] | select(.cveID == "CVE-2024-3400")'
EPSS is a public API, no key:
curl -s "https://api.first.org/data/v1/epss?cve=CVE-2024-3400"
# {"data":[{"cve":"CVE-2024-3400","epss":"0.99999","percentile":"1.0", ...}]} Live reporting is the awkward one - it means monitoring vendor blogs, advisories and news continuously. That is what threat-intelligence platforms are for.
All three signals in one call
ThreatCluster tracks CVEs against clustered security reporting from 28,000+ publishers, so one request returns KEV status, EPSS, public exploit availability and how much live reporting the CVE has right now. Free keys are 100 credits a day, no card (get one here):
curl -s "https://threatcluster.io/api/public/v1/vulnerabilities/CVE-2024-3400" \ -H "X-API-Key: tc_live_..."
{
"cve_id": "CVE-2024-3400",
"cvss_v3_score": 10.0,
"epss_score": 0.99999,
"in_kev": true,
"kev_added_date": "2024-04-12",
"ransomware_use": true,
"has_exploit": true,
"affected_vendors": ["paloaltonetworks"],
"...": "description, CWEs, exploit and reference URLs"
} To watch the stream instead of asking per CVE - "what became exploitable this week in my stack?":
curl -s "https://threatcluster.io/api/public/v1/vulnerabilities?days=7&kev_only=true&vendor=fortinet" \ -H "X-API-Key: tc_live_..."
For AI agents
If the thing asking the question is an agent rather than a script, the same data is an MCP server - Claude, Cursor, VS Code or anything MCP-capable gets get_vulnerability and exploited_vulnerabilities as tools:
npx -y threatcluster-mcp # or: uvx threatcluster-mcp This matters more than it looks: a bare LLM answers "is this exploited?" from training data that is months stale. On a 150-question benchmark about recent threats, a frontier model scored 0/150 ungrounded and 158/158 grounded (methodology).
Which signal should gate patching?
- Compliance-grade "known exploited": KEV. It is the regulatory anchor (BOD 22-01).
- Ranking a 500-CVE backlog: EPSS, with KEV membership as an override.
- "Is there a campaign right now?": live reporting volume.
- One answer for a ticket or an agent: all three in one call.
KEV and EPSS are free, authoritative and should be in every pipeline. The third signal is what tells you about the week between a campaign starting and the catalogs catching up - which is usually the week that matters.