Get a demo →
A single decision brain that intercepts every package install — across developer laptops, CI/CD pipelines and Kubernetes clusters — and answers one question in milliseconds: is this safe to ship?
They score a CVE the moment one is filed. They don't see the package the day it gets compromised. They don't see typosquats, install-script malware, or maintainer takeovers. Most of the npm and PyPI supply-chain attacks of the last four years never received a CVE on the day they shipped — and the ones that did got it days or weeks late.
CVSS scores the technical impact of a single CVE in isolation. A CVSS 9.8
in a library you don't ship is noise. A CVSS 6.1 in axios — used
by half of npm — is an emergency. The score can't tell the difference.
EPSS estimates the probability a published CVE will be exploited in the next 30 days. It's a useful signal for triaging known vulns — but it has no opinion on the entire malware-on-the-registry problem, because that universe has no CVEs to begin with.
Of the most damaging npm supply-chain compromises of the last four years —
axios, event-stream, ua-parser-js,
coa, rc, colors, faker,
node-ipc, eslint-scope — none had a
meaningful CVSS or EPSS score on day zero. CyberXYZ catches them on signal,
not on paperwork.
We don't replace CVSS or EPSS — we absorb them, then add the four things they're missing: package centrality, behavioral drift, malware intel, and exploit availability. The result is a score you can act on.
axios@1.14.1, the day the maintainer was compromised.
CVSS shrugged. EPSS hadn't caught up. XYZ blocked it.
The classic 0–10 base score. We keep it — but we never trust it alone.
The 30-day exploitation forecast. Folded in as an exploit-likelihood input, not the verdict.
Monthly downloads, dependent count, GitHub stars, tier (1–4). A vuln in a Tier-1 package is 1000× more dangerous than the same vuln in a leaf.
Maintainer-set change, file-count drift > 30%, install-script appearance, ≥3-minor version jump. Catches compromise before any CVE is filed.
14,000+ MAL-* records (OSV MAL + 9 human-verified) + 16 attack campaigns mapped to MITRE ATT&CK TTPs. The corpus CVSS doesn't see.
ExploitDB + Packet Storm + GitHub PoCs cross-referenced. If it's weaponized, the score knows.
Every install request runs all four detection signals in parallel against two databases (1.1M vulnerabilities, 72K npm packages with behavioral baselines). We aggregate the strongest signal into one of four decisions.
Catches a brand-new transitive dependency that never existed in any prior version of the package. The classic compromise pattern.
quarantineFires when the new minor version is ≥3 ahead of the last known minor within the same major. Pre-release suffixes are stripped first.
quarantineCross-checks the human-verified attack corpus + 14K malware advisories + 16 attack campaigns mapped to MITRE ATT&CK TTPs. Hard block.
blockLive advisory match with semver-range evaluation. Critical severity triggers block, high triggers alert, otherwise informational.
block alert.tgz is buffered (≤5 MB), package.json
install scripts extracted, and 11 supply-chain patterns run against them
(eval, env exfil, base64+eval, SSH-key write, webhook exfil, reverse shell,
crypto mining…). Adds ~50–200 ms.
POST /api/v1/proxy/sbom accepts CycloneDX or SPDX and runs the
whole graph through the brain at once. Perfect for compliance gates.
A vulnerability in a Tier-1 package — one that millions of dependent packages pull in transitively — is a thousand times more dangerous than the same vulnerability in a leaf. We rank every package on the registry by downloads, dependent count, GitHub stars and graph centrality, and we feed that into every decision the brain makes.
The brain reasons over 1.1 million vulnerability records, 14,000+ malware advisories and 6.6 million dependency edges — refreshed continuously from the public feeds and our own proprietary intel: behavioral baselines, maintainer drift, and 16 hand-curated attack campaigns mapped to MITRE ATT&CK.
Vulnerability data is fragmented across dozens of feeds — different schemas, different IDs, different update cadences. We ingest them all, normalize into one unified schema, deduplicate, and build a live dependency graph with 6.6 million edges.
Refreshed daily via Cloud Run jobs. Dedup'd, semver-normalized, indexed for sub-100ms lookup.
Coverage across npm, PyPI, Maven, NuGet. Synced after every OSV pull. The corpus CVSS doesn't see.
Built in-house. Updated continuously. The reason we catch axios@1.14.1 in 18 seconds and not 6 days.
Wherever a developer, a build server or a container runtime asks the registry for a package, CyberXYZ is already there. Same brain, same verdict, three different entry points.
"dependencies": {
"axios": "1.14.1" ⛔ blocked · supply-chain attack,
"lodash": "4.17.21" ✓ clean,
"express": "4.18.2" ⚠ 2 medium
}
VSCode extension shows inline diagnostics on every
package.json save, with hover-cards exposing the full XYZ
risk score, tier, downloads, dependents and any known campaign link.
The xyz CLI covers scan, audit,
check and depalert — same verdict, terminal-native.
- name: XYZ Supply-Chain Gate
uses: cyberxyz/depalert-action@v1
with:
api-key: ${{ secrets.XYZ_API_KEY }}
files: package-lock.json, requirements.txt
fail-on: block,quarantine
Drop-in GitHub Action that gates every PR and push.
Reads package-lock.json, requirements.txt,
go.sum, pom.xml — calls the brain — exits with
0/1/2/3 for clean / block / quarantine / alert. Skips
Dependabot PRs automatically. Works with GitLab CI, CircleCI, Jenkins.
Drop-in HTTP proxy in front of registry.npmjs.org and
pypi.org/simple. Buffers every tarball, scans
preinstall/install/postinstall scripts against
11 supply-chain patterns, sanitizes manifests, and
returns 403/451/200 in milliseconds.
Cluster-wide enforcement, zero developer friction.
The maintainer account jasonsaayman was compromised. A malicious
version was pushed to npm. 43 minutes later it was downloaded
37,000 times. Here's exactly what the brain did, in order, in milliseconds.
Compromised maintainer pushes a new version to registry.npmjs.org.
No CVE filed. CVSS unknown. EPSS unknown.
Cloud Run job polling 189 Tier-1 packages every 15 minutes catches the
new version. Onboards it into version_compatibility.
A runtime dependency that has never existed in any prior version of axios is now present. Verdict: quarantine.
Last seen minor: 1.9. New minor: 1.14. Jump = 5. Threshold = 3. Verdict: quarantine.
Analyst confirms the RAT dropper. malicious_packages entry
marked verified_by_human=TRUE. is_malicious_version()
now hard-blocks 1.14.1 globally.
Across all CyberXYZ customers, every npm install axios@1.14.1
through the proxy returns 403 BLOCK. Every CI gate fails.
Every VSCode session shows the red badge. Every CLI exits 1.
Auto-tagged: T1195.002 Compromise Software Supply Chain,
T1059.007 JS Execution, T1567.002 Exfiltration to
Cloud Storage. SOC dashboards light up with a pre-built incident.
By then, the package had been installed millions of times. CyberXYZ blocked it in under 18 seconds.
Most "supply-chain security" products are vulnerability scanners with a different homepage. They tell you what's broken after it's in your tree. CyberXYZ is the only one that intercepts the install request itself, across CLI, CI/CD and runtime, with one decision brain.
axios@1.14.1 catchTwo database-native SQL functions that evaluate dep-injection and version-jump anomalies for every newly-published version of every Tier-1 package, every 15 minutes. Catches compromise before the CVE.
Most tools are advisory. We are in the data path. The tarball does
not reach the developer's node_modules if our brain says no.
Per-package maintainer set, file count, install-script presence and version-cadence baselines. We notice when something is suddenly different — not just suddenly broken.
One brain. CLI, IDE, CI/CD and the runtime proxy all hit the same
POST /api/v1/proxy/check. No drift. No "the IDE said one thing
but the pipeline said another".
16 mapped attack campaigns, 9 human-verified malware records with exact version ranges, full MITRE ATT&CK TTP mapping. Stuff you literally cannot get from NVD.
npm, PyPI, Maven, NuGet, Go and more — same brain, same verdict, same API. Not an "npm tool that bolted on Python last quarter".
Get a 15-minute walkthrough of the brain, the proxy, and the live attack feed. We'll plug it into your CI in under an hour.
Check your inbox — we'll reach out within 24 hours.