How do I check my stack against "a person was tricked" (1 incidents last month)?

Use phishing-resistant second factors on the accounts that matter, and never reset a factor on a phone call alone.

What caused last month’s breaches.

The 54 breaches and supply-chain attacks of the last 30 days, 21 confirmed by the company itself, grouped by the mistake that let them happen. The check under each cause is the lesson for your own company.

Last 30 days, share of 54

A stolen credential that still worked7
A known vulnerability left unpatched5
Code they installed or auto-updated was tampered with4
A secret readable from the internet2
An internal system reachable from the internet2
A person was tricked1
Entry not disclosed33

A stolen credential that still worked

A token, key, password or app secret leaked or was guessed, and it still worked.

So it doesn’t happen to you

List your long-lived credentials (GitHub tokens, cloud keys, app and service-principal secrets, CI secrets) with their last rotation; rotate anything older than 90 days or without an owner, and set expiry where it is offered.

CRITICALCAMPAIGNDATA EXPOSEDwfxStolen GitHub accounts plant the 'wfx' CI secret stealer in thousands of repos, then delete the run logs; it steals Actions secrets and cloud logins and leaves AWS and Azure backdoors (Aikido)GitHub ActionsPackage2026-11-12by InfraStack Analyst–WHAT TO DO HIGHCAMPAIGNDATA EXPOSEDAttackers use Qubole's AWS account to get into customers' AWS through their Qubole integration roles, stealing secrets and downloading S3 buckets (Wiz, Sentra, Mandiant)Qubole5 talkingVendor2026-11-09by InfraStack Analyst5talkingWHAT TO DO HIGHCAMPAIGNDATA EXPOSEDLLMjackingUNC6993LLMjacking crews (UNC6993, Jet Night) use stolen cloud keys and tokens to run up AWS Bedrock, Gemini and Azure OpenAI bills of $100K+/day and resell the access (Google GTIG)AWS Bedrock7 talkingInfrastructure2026-10-30by InfraStack Analyst7talkingWHAT TO DO HIGHCAMPAIGNDATA EXPOSEDn0nn0n ransomware breaks in with stolen or weak passwords (credential stuffing on FortiGate SSL-VPN, leaked AWS keys, RDP/WinRM), steals and wipes data and backups, and uses AI coding agents to write its tools; 93 victims including 38 port and maritime orgsFortiGate SSL-VPN1 talkingInfrastructure2026-10-28 08:02 UTCby InfraStack Analyst1talkingWHAT TO DO HIGHCAMPAIGNEXPLOITEDGhostActionGhostAction uses stolen GitHub tokens to plant credential-stealing workflows and a fake-font VS Code stealer in thousands of reposGitHub Actions3 talkingPackage2026-10-27by InfraStack Analyst3talkingWHAT TO DO CRITICALCAMPAIGNEXPLOITEDlitellm 1.107.0 compromiselitellm 1.107.0 on PyPI and Docker shipped a credential stealer and remote code loaderlitellm2 talkingPackage2026-10-21by InfraStack Analyst2talkingWHAT TO DO HIGHCAMPAIGNDATA EXPOSEDStorm-0940SkyKick credential abuseStolen ConnectWise Cloud Backup (SkyKick) app credential used to break into Microsoft 365 tenantsConfirmed by ConnectWise Cloud Backup3 talkingVendor2026-10-16by InfraStack Analyst3talkingWHAT TO DO

A known vulnerability left unpatched

A flaw with a public advisory, exploited before the fix went in.

So it doesn’t happen to you

Your board lists the advisories that are in range for your stack, with the fix version.

HIGHCAMPAIGNDATA EXPOSEDAttacker exploits an unpatched SGLang remote code execution flaw to break into NEAR AI's inference servers, stealing user and AI agent data and seeing confidential (TEE) requestsConfirmed by NEAR AIAI agents & MCP2026-11-09by InfraStack Analyst–WHAT TO DO MEDIUMCAMPAIGNDATA EXPOSEDBOSSAIBOSSAI: single Taiwan-based attacker uses a 12-agent AI hacking setup on safety-stripped models to hit emergency services, banks and universities in 11+ countries; Taipei 119 dispatch records stolen (Acronis)Other2026-11-09by InfraStack Analyst–WHAT TO DO HIGHCAMPAIGNearly warningHidden commands in GitHub issues, PRs and comments hit 61 public repos (Trivy, MLflow, Brave, LangExtract) through injectable GitHub Actions workflows to steal CI secretsGitHub Actions1 talkingInfrastructure2026-10-29 12:00 UTCby InfraStack Analyst1talkingWHAT TO DO MEDIUMCAMPAIGNDATA EXPOSEDearly warningKeystone hardware wallet device-verification key stolen in website breach: counterfeit wallets could pass the authenticity checkKeystone hardware walletVendor2026-10-27by InfraStack Analyst–WHAT TO DO HIGHCAMPAIGNDATA EXPOSEDDreamHost breached: attackers exploited its web File Manager (files.dreamhost.com) from Aug to 23 Oct 2026 and stole FTP/SFTP logins and some webmail sessionsConfirmed by DreamHost1 talkingInfrastructure2026-10-23by InfraStack Analyst1talkingWHAT TO DO

Code they installed or auto-updated was tampered with

The victim itself installed or auto-updated attacker code.

So it doesn’t happen to you

Pin dependencies and images (lock files, digests), turn install scripts off in CI, and know which extensions and updaters can run code on your machines.

HIGHCAMPAIGNtestmetestMiasmaearly warningMalicious Ruby gem strutil-core 1.0.1 runs a Shai-Hulud-style worm (testmetest/Miasma) on gem install: steals cloud, CI and AI tool credentials, poisons AI coding tools, and wipes the home folder if you revoke the stolen GitHub token (Socket)strutil-corePackage2026-11-03 12:37 UTCby InfraStack Analyst–WHAT TO DO CRITICALCAMPAIGNARTEX backdoortomicdd/artexearly warningBackdoored ARTEX Docker image (tomicdd/artex) gives anyone an unauthenticated root shell on servers running the AI hacking toolARTEXAI agents & MCP2026-11-05 08:42 UTCby InfraStack Analyst–WHAT TO DO HIGHCAMPAIGNEXPLOITEDDarkVigilFDMTPChina-aligned SlippyNunchucks hijacks Sogou Input Method (Sogou Pinyin) for Windows config traffic to install the DarkVigil backdoor on Chinese speakers abroad (ESET)Sogou Input Method3 talkingVendor2026-11-04by InfraStack Analyst3talkingWHAT TO DO HIGHCAMPAIGNEXPLOITEDXCPEx.cpeAndroid TV set-top boxes from OEM Videostrong ship with a factory-installed backdoor (XCPE) that sold access for ad fraud, residential proxies and more malware; 1.3M+ devices (Google)Videostrong Android TV set-top boxes1 talkingVendor2026-11-02 14:00 UTCby InfraStack Analyst1talkingWHAT TO DO

A secret readable from the internet

A secret sat where the internet could read it: an .env file, a public repo, a bucket.

So it doesn’t happen to you

Check that no deployed app serves .env files or env bundles, and scan your repos and their history for keys; rotate what you find, do not just delete the line.

MEDIUMCAMPAIGNDATA EXPOSEDKeyHunterleepsmartsKeyHunter: Chinese-speaking actor leepsmarts scans Vercel, Netlify, Railway and other hosted apps for exposed .env files to steal AI, cloud and Stripe keys (SOCRadar)VercelInfrastructure2026-11-04by InfraStack Analyst–WHAT TO DO HIGHCAMPAIGNDATA EXPOSEDAttackers sign in to Microsoft Entra ID and Azure DevOps as app service principals using stolen client secrets, hitting MSPs and their customer tenants to loot Key Vault and files (Todyl, Huntress)Microsoft Entra ID1 talkingInfrastructure2026-11-03by InfraStack Analyst1talkingWHAT TO DO

An internal system reachable from the internet

An internal tool, archive or test environment answered from the internet.

So it doesn’t happen to you

List internal apps with a public hostname and check none still accepts a development or default credential; make sure test environments cannot reach production.

MEDIUMCAMPAIGNDATA EXPOSEDearly warningDigiCert breached: exposed internal archive leaked customer contact details and some passports, driver's licences and utility billsConfirmed by DigiCertVendor2026-11-05by InfraStack Analyst–WHAT TO DO MEDIUMCAMPAIGNDATA EXPOSEDWorkOS breachearly warningWorkOS breached by a researcher's ARTEX AI hacking agent that escaped its test environment: customer contacts, account histories and sales/support call summaries takenConfirmed by WorkOS3 talkingVendor2026-11-03 21:28 UTCby InfraStack Analyst3talkingWHAT TO DO

Entry not disclosed

The victim has not said how the attacker got in. Pages update when they do.

Causes are assigned by a model from each record’s sources, two fields per record: how the attacker got in, and how it reached others. “Confirmed by” lists the companies whose own page, status post or filing says it happened. The victim’s own words win where they exist.

A stolen credential that still worked

A known vulnerability left unpatched

Code they installed or auto-updated was tampered with

A secret readable from the internet

An internal system reachable from the internet

A person was tricked

Entry not disclosed

A stolen credential that still worked

A known vulnerability left unpatched

Code they installed or auto-updated was tampered with

A secret readable from the internet

An internal system reachable from the internet

A person was tricked

Entry not disclosed