Socket

Primary category: software-supply-chain.

One-liner — Supply-chain security that catches malicious open-source packages — backdoors, typosquats, install-script abuse, obfuscated code — before they land in your codebase, rather than just flagging known CVEs.

What it does

Socket inspects the actual behavior and metadata of open-source packages (npm, PyPI, and other ecosystems) to detect active threats: malware, hidden backdoors, typo-squatted package names, suspicious install scripts, credential/token exfiltration, and risky permission changes between versions. This is a different job from CVE-based scanning: Socket is built for the novel, malicious-package attack — the npm/PyPI compromises that don’t have a CVE yet. It claims to block 100+ supply-chain attacks per week and is frequently the first public discloser of major npm campaigns (e.g. the 2025 “Shai-Hulud” worm, the “Qix” maintainer phishing compromise, and “SANDWORM_MODE”), several later echoed by CISA.

Where it sits in the stack

software-supply-chain, foundation layer. Relevance to AI-generated code: this is exactly the untrusted-input vector that AI coding assistants amplify. LLMs hallucinate plausible-but-nonexistent package names (“slopsquatting”), which attackers pre-register with malware; LLMs also confidently suggest installing obscure packages. Socket’s behavioral, malicious-package focus is the control aimed at that threat — complementary to CVE scanners like snyk and semgrep, which would not flag a brand-new malicious package.

Deployment & architecture

SaaS. Primary surface is a GitHub App that posts pull-request comments/alerts when a dependency change introduces risk, plus a CLI (socket) for local/CI use and an API. Also offers IDE feedback and registry/firewall-style controls to block bad packages at install time. Lightweight to adopt — no need to instrument the whole codebase; it gates dependency changes.

Positioning & differentiators

The category leader for malicious-package / active-supply-chain-attack detection, distinct from the known-CVE SCA done by snyk, semgrep, and endor-labs. Where those answer “does this dependency have a published vulnerability?”, Socket answers “is this dependency trying to attack me?” Founder/CEO Feross Aboukhadijeh is a well-known open-source author (WebTorrent, StandardJS), which underpins its credibility and ecosystem reach. Best used alongside a CVE scanner, not instead of one. Versus aikido-security (which bundles many scanner types for SMBs), Socket is the specialist on the malicious-package leg.

Ownership, funding & M&A

Independent, private, venture-backed. Founded ~2020; HQ San Francisco; founder/CEO Feross Aboukhadijeh. Total funding ~$65M; most recent round a $40M Series B in October 2024, led by Abstract Ventures with participation from Elad Gil, Andreessen Horowitz (a16z), and angels including Bret Taylor. No M&A; no IPO. Reported scale: 7,500+ organizations and 300,000+ GitHub repos protected. (ownership_confidence: high. Founding year ~2020 is approximate — not confirmed against an incorporation record here.)

CTO / hedge-fund lens

Day-2 baseline, Day-1 if your devs ship AI-generated code — and arguably the single highest-leverage add precisely because AI assistants increase the odds of pulling in a hallucinated or malicious package. It is cheap to deploy (a GitHub App on your repos) and low-friction, so even a small fund (sub-50 eng) can adopt it. The recommended posture is Socket for malicious packages + snyk or semgrep for known CVEs and SAST — they cover different halves of the supply-chain problem. Model-risk relevance is light/indirect: it doesn’t touch SR 11-7 model governance, but it directly mitigates a fast-growing AI-coding-era attack surface.

Competitors / alternatives

snyk, semgrep, endor-labs, aikido-security; GitHub Dependabot (CVE-only, different scope; not yet a page).

Open questions / to verify

• Exact incorporation/founding year (using ~2020; not confirmed against a primary registration record).
• Whether the “firewall”/install-time blocking is GA across all supported ecosystems vs. detection-only in some.

Sources

• Socket secures $40M Series B — Socket blog — fetched 2026-06-28 — supports: $40M Series B Oct 2024 (Abstract Ventures, Elad Gil, a16z), $65M total, Feross Aboukhadijeh CEO, SF HQ, malicious-package detection, GitHub App/CLI, 7,500+ orgs; confidence: high (vendor primary)
• Widespread Supply Chain Compromise Impacting npm — CISA + Socket disclosures — fetched 2026-06-28 — supports: Socket as first-responder/discloser of major 2025 npm attacks (Shai-Hulud, Qix, SANDWORM_MODE); confidence: high

History

• [2026-06-28] Stub created from seed registry.
• [2026-06-28] Researched; established independent/private VC-backed, founded ~2020 (SF HQ, CEO Feross Aboukhadijeh), ~$65M raised, Series B $40M Oct 2024 (Abstract Ventures/Elad Gil/a16z); specialist in malicious open-source package detection (npm/PyPI), distinct from CVE scanners; ownership_confidence high. hedge_fund_fit medium (Day-1 if shipping AI-generated code — directly mitigates slopsquatting).