A note on Jeeves In 1996 a search engine called Ask Jeeves let you type a question in plain English and got you a useful answer. It was the natural-language search engine before the term existed. Then Google won the keyword war, Ask Jeeves got bought and rebranded to nothing, and the idea of natural-language search went into hibernation for two decades until Anthropic and OpenAI pulled it back out of the freezer. We named our threat-intelligence MCP Jeevesus as a sly tip of t

In February of this year we published The Sberbank Pipeline, documenting a 5,000-euro wire from Jeffrey Epstein's executive assistant Lesley Groff to Sberbank Ekaterinburg in October 2017. The framing was that Epstein moved money to Russia after his conviction. The framing was correct. The chronology was incomplete. This morning, while chasing a separate thread on Aviloop LLC, we surfaced the rest of the lane. Between September 2008 and September 2014 — six years — Jeffrey Ep

This morning a user typed two words into our search bar: aviloop epstein. We had not heard of Aviloop. By tonight we had eleven hundred pages of receipts. Here is what the corpus actually says. The wire memo EFTA01580967 is a one-page memorandum on Jeffrey Epstein letterhead. Date: October 21, 2011. To: Janet at JP Morgan. From: Darren Indyke. The text reads: "Please wire Fifty Thousand Dollars and 00/100 ($50,000) from the above account to: Bank name: JP Morgan Chase, ABA #:

Kyle Hanslovan, CEO of Huntress, posted this morning about a campaign his team and Microsoft Threat Intelligence have been tracking. The crew calls themselves EvilTokens. They host on Railway, route victims through Cloudflare Workers and Vercel, and abuse the Microsoft device-code authentication flow to bypass MFA. Hundreds of orgs across five countries hit in weeks. Every IOC unique. Every domain trusted. Every cloud service reputable. Then Kyle named the thing the security

May 3, 2026 · DugganUSA LLC We shipped a new cron job today. It runs every thirty minutes, pulls the RSS feeds for Mandiant, Unit 42, CrowdStrike, Microsoft Security, Elastic Security Labs, Volexity, and Cisco Talos, regex-extracts indicators of compromise from every fresh post, and writes them straight into our STIX feed with the originating vendor and the post URL preserved as provenance. Three point two seconds, end to end. Six hundred ninety-five IOCs on the first run. He

May 3, 2026 · DugganUSA LLC Microsoft Security today published "Cross-tenant helpdesk impersonation to data exfiltration: A human-operated intrusion playbook." It is, line by line, the attack chain we documented in our outreach to Medtronic's Product Security team on March 19, 2026. Six weeks before Medtronic confirmed the breach. Our outreach went to [email protected] and got an autoresponder from [email protected] confirming delivery. We have the timesta

One Worker. 8,898 catches a day. Zero servers. Free to start. May 2, 2026 · Patrick Duggan, DugganUSA LLC In the last 24 hours, a single Cloudflare Worker we wrote — Edge Shield — caught 8,898 attacker probes and reported them home. That's six probes a minute, every minute, all day, never reaching our origin server. No SOC analyst on shift. No SIEM. No EDR license. No on-call rotation. It just runs. The Worker is open source. The same code runs on a $10/month side project and

Six weeks later, ShinyHunters disclosed 9 million records. May 2, 2026 · Patrick Duggan, DugganUSA LLC On April 17, the cybercrime collective ShinyHunters posted on a dark-web forum that they had breached Medtronic and exfiltrated records on roughly nine million people. Five days later, Medtronic confirmed an "unauthorized party" had accessed their corporate IT environment. By April 27, The Register was running a joint write-up about Medtronic and Itron both admitting digital

May 2, 2026 · DugganUSA LLC The Model Context Protocol now has 21,962 servers in the official registry. Tonight we ingested every one of them, hashed every tool description, and stamped a day-zero baseline. Tomorrow's snapshot will show us the first wave of rug-pulls — servers that quietly rotated their tool descriptions to inject new instructions into AI agents. We are going to be the people who notice. Why this matters now The MCP ecosystem went from prototype to enterprise

We posted earlier today about our exploit harvester catching three GitHub PoC repositories for CVE-2026-41940 — the cPanel and WHM authentication-bypass flaw. The harvester caught them overnight. Three rules, twelve hours, zero humans involved. We were proud of the catch. The story changed at 18:00 UTC. Help Net Security published a piece reporting the actual exploitation window: this CVE has been an active zero-day since at least February 23, 2026. Sixty-seven days before CI

We have posted three times today about the cluster on .bet TLD that PreCog Sweep flagged at 23:56 UTC on April 30. We named the botanical wing The Apothecary, surfaced the broader 32-wing umbrella, and traced the registration paper trail to PDR Ltd at IANA 303 with all 32 parents pushed through a single 24-hour batch registration window. The framework is ClearFake — URLhaus's classifier had 200 of 200 sampled IOCs tagged as malware_family equals ClearFake. ClearFake has been

We posted twelve hours ago about a fresh malware-delivery cluster on .bet TLD that our PreCog Sweep flagged at 23:56 UTC on April 30. Five botanical-themed subdomains. We named it The Apothecary. A few hours later, the URLhaus and SSLBL feeds caught up overnight and the cluster grew to thirty-two parent .bet domains hosting 184 subdomains, all serving the same path /software-distribution-dxnp2c7/meta-verify.index. We posted again about the broader umbrella, called the campaig

CISA added CVE-2026-41940 to the Known Exploited Vulnerabilities catalog on April 30, 2026. Critical missing-authentication flaw in WebPros cPanel and WHM, plus the WP Squared bundle. The kind of bug that turns the management plane of a shared-hosting environment into an unauthenticated console. Federal civilian agencies got a deadline. Everyone else got a problem. Most security teams woke up the next morning and started writing detection content from scratch. Our exploit har

April 21 through April 23, 2026. Forty-eight hours. Three independent supply-chain compromises hitting npm, PyPI, and Docker Hub. All three targeting secrets — API keys, cloud credentials, SSH keys, CI/CD tokens. GitGuardian framed it as "no off-season." They were right and the framing is worth our spin because the three campaigns each picked a different physics of attack. The wormable one is the one nobody is talking about loud enough. CanisterSprawl. April 21. A malicious v

April 22 to April 30, 2026. Eight days. Three independent supply-chain compromises, all targeting the machine-learning stack. xinference on PyPI — three consecutive releases on April 22 carrying a credential-stealing payload. SSH keys. AWS, Azure, and GCP credentials. Environment variables. Crypto wallets. intercom-client on PyPI — co-disclosed by Socket on April 30 in the same cluster as Lightning. Same JavaScript-payload pattern. PyTorch Lightning on PyPI — versions 2.6.2 a