CVE, EPSS & KEV API with Monitoring

Ask for one CVE or search many. A CVE record carries the CVSS score, the EPSS exploitation probability and the CISA KEV status. The tools are free without an API key. They work over REST and as MCP tools, for developers and AI agents. With a Pro key you can also monitor CVEs and read their change events.

What you can ask

GET /cve_lookup/{cve_id} MCP cve_lookup. One CVE with CVSS, EPSS, KEV status, CWE, references, CISA SSVC and Nuclei templates. API docs
POST /cve_lookup_bulk MCP cve_lookup_bulk. Many CVEs in one request. API docs
GET /cve_products/{cve_id} MCP cve_products. The products a CVE affects, one page at a time. API docs
GET /cve_search MCP cve_search. Search CVEs by vendor, product, severity, KEV, EPSS and more. API docs
GET /cve_triage MCP cve_triage. Rank the CVEs of a vendor or product by risk score, to see what to patch first. API docs
POST /cve_triage_bulk MCP cve_triage_bulk. Rank a list of CVE IDs by risk score. API docs
GET /cve_risk_score/{cve_id} MCP cve_risk_score. A risk score from 0 to 100 for one CVE, with a label. API docs
GET /cve_leading MCP cve_leading. The CVEs most likely to be exploited, ranked by EPSS. API docs
GET /kev_lookup/{cve_id} MCP kev_lookup. The CISA KEV entry of a CVE: due date, required action, ransomware use. API docs
GET /cve_exploit/{cve_id} MCP cve_exploit. Public exploit data: GitHub PoCs, Metasploit modules, NVD and CNA exploit references, ExploitDB. API docs
GET /cwe_lookup/{cwe_id} MCP cwe_lookup. A MITRE CWE weakness with its mitigations. API docs
GET /product_exposure MCP product_exposure. The CVE exposure of one vendor and product, with its end of life dates from endoflife.date. API docs
GET /cpe_search MCP cpe_search. Find the vendor and product names that CVE filters accept. API docs
GET /cvss_calculator MCP cvss_calculator. Decode a CVSS v3 vector and recompute its scores. API docs

Example

Look up one CVE:

bashcurl https://api.contrastcyber.com/cve_lookup/CVE-2021-44228

Example response, trimmed; the values are an example:

json{ "cve_id": "CVE-2021-44228", "severity": "CRITICAL", "cvss_v3": 10.0, "published_time": "2021-12-10T00:00:00Z", "epss_score": 0.97566, "epss_percentile": 0.99970, "cwe_id": "CWE-20", "in_kev": true, "kev_date_added": "2021-12-10", "poc_count": 11, "has_metasploit_module": true, "ransomware_campaign": "Known" }

From an agent

Your agent calls these MCP tools: cve_lookup, cve_lookup_bulk, cve_products, cve_search, cve_leading, cve_triage, cve_triage_bulk, cve_risk_score, kev_lookup, cve_exploit, cwe_lookup, product_exposure, cpe_search and cvss_calculator. Connect once with the MCP setup, then ask in plain words.

Limits

Without a key you get 30 credits per hour. A Pro key gets 500 credits per hour. cve_lookup_bulk takes up to 50 CVE IDs per request. cve_triage_bulk takes up to 50.

Monitor it

A Pro key can watch CVEs. Add an item with watchlist_edit. Then read its change events with watchlist_events. You pull the events when you want them.

Each item watches one endpoint. The endpoint sets the events the item can emit:

Watched endpointEvents
cve_lookupcve_added kev_added kev_removed epss_level_up epss_level_down cvss_added cvss_risk_band_changed exploit_published product_added product_removed ransomware_linked vendor_advisory_published patch_published cve_disputed cve_rejected metasploit_added nuclei_template_added exploitdb_added ssvc_added ssvc_changed github_poc_added description_changed cvss_score_changed cwe_changed kev_updated ghsa_linked
product_exposurecve_added kev_added kev_removed epss_level_up epss_level_down cvss_added cvss_risk_band_changed exploit_published product_added product_removed ransomware_linked vendor_advisory_published patch_published cve_disputed cve_rejected metasploit_added nuclei_template_added exploitdb_added ssvc_added ssvc_changed github_poc_added description_changed cvss_score_changed cwe_changed kev_updated ghsa_linked eol_approaching eol_reached eol_date_changed
cve_searchcve_added kev_added kev_removed epss_level_up epss_level_down cvss_added cvss_risk_band_changed exploit_published product_added product_removed ransomware_linked vendor_advisory_published patch_published cve_disputed cve_rejected metasploit_added nuclei_template_added exploitdb_added ssvc_added ssvc_changed github_poc_added description_changed cvss_score_changed cwe_changed kev_updated ghsa_linked epss_above_min epss_below_min
kev_lookupkev_added kev_removed ransomware_linked kev_updated
cve_exploitexploit_published metasploit_added exploitdb_added github_poc_added

The EPSS levels are 0.05, 0.1, 0.2, 0.3, 0.5, 0.7 and 0.9. An epss_level_up event comes when the EPSS score reaches a level above the one the CVE holds. An epss_level_down event comes when the score falls under 0.8 times a level the CVE holds. A cve_search item that sets min_epss also hears epss_above_min, when the score rises from under min_epss to min_epss or above, and epss_below_min, when it falls from min_epss or above to under it. min_epss takes at most three decimals there. A cve_search item must set at least one filter. A product_exposure item takes a vendor and a product, and may take one version. The version makes it a separate item from the same pair without one. It does not narrow the CVE events. It picks the release cycles of the three end-of-life events. See pricing for the Pro key. The security monitoring page lists every watched endpoint and its events.

Data sources

To read CVEs from the CISA KEV catalog as web pages, open the Known Exploited Vulnerabilities list on contrastcyber.com.

FAQ

Is the CVE API free?
Yes. Every CVE tool works without a key, within the hourly credit limit. Only monitoring needs a Pro key.
Does it show whether a CVE is in the CISA KEV catalog?
Yes. cve_lookup returns in_kev and the KEV date. kev_lookup returns the due date, the required action and the ransomware flag.
How do I rank CVEs by how likely they are to be exploited?
Use cve_leading for the top EPSS scores. Use cve_triage or cve_triage_bulk to rank your own vendors, products or CVE IDs by risk score.
Can an AI agent use it?
Yes. Each tool is also an MCP tool with the same name, so Claude, Cursor and other MCP clients can call it.

Docs and source