<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>Research on Mystic Cyber Research</title><link>https://research.mysticcyber.com/</link><description>Recent content in Research on Mystic Cyber Research</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Fri, 07 Aug 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://research.mysticcyber.com/index.xml" rel="self" type="application/rss+xml"/><item><title>Threat Hunts for Shell Command Obfuscation on VMware ESX</title><link>https://research.mysticcyber.com/threat-hunts-shell-obfuscation-vmware-esx/</link><pubDate>Fri, 07 Aug 2026 00:00:00 +0000</pubDate><guid>https://research.mysticcyber.com/threat-hunts-shell-obfuscation-vmware-esx/</guid><description>VMware ESX systems are a recurring target in ransomware campaigns. Threat groups including SCATTERED SPIDER, BlackBasta, Royal (aka BlackSuit) and Akira lean on the same handful of obfuscation tricks — here is how to hunt them.</description></item><item><title>Expanding AI Benchmarks in Cybersecurity Beyond Vulnerability Discovery</title><link>https://research.mysticcyber.com/expanding-ai-benchmarks-beyond-vulnerability-discovery/</link><pubDate>Thu, 06 Aug 2026 00:00:00 +0000</pubDate><guid>https://research.mysticcyber.com/expanding-ai-benchmarks-beyond-vulnerability-discovery/</guid><description>The conversation about AI in cybersecurity has centered on capabilities like vulnerability discovery, exploit generation and automated proof-of-concept development. That framing leaves most of the defensive surface unmeasured.</description></item><item><title>Secure Agent Harness Execution: Preventing Privilege Escalation</title><link>https://research.mysticcyber.com/secure-agent-harness-execution/</link><pubDate>Wed, 05 Aug 2026 00:00:00 +0000</pubDate><guid>https://research.mysticcyber.com/secure-agent-harness-execution/</guid><description>As autonomous agentic systems take on more critical operations inside corporate boundaries, securing the execution harness against untrusted agent actions is the new frontier of enterprise risk mitigation.</description></item><item><title>What a Board Actually Wants From a Quarterly Risk Report</title><link>https://research.mysticcyber.com/what-a-board-wants-from-a-risk-report/</link><pubDate>Wed, 29 Jul 2026 00:00:00 +0000</pubDate><guid>https://research.mysticcyber.com/what-a-board-wants-from-a-risk-report/</guid><description>Most quarterly risk reporting optimizes for completeness. Boards optimize for decisions. Three structural changes that close the gap without adding a reporting cycle.</description></item><item><title>Reducing Alert Toil With Response Playbooks That Survive Contact</title><link>https://research.mysticcyber.com/response-playbooks-that-survive-contact/</link><pubDate>Wed, 22 Jul 2026 00:00:00 +0000</pubDate><guid>https://research.mysticcyber.com/response-playbooks-that-survive-contact/</guid><description>We instrumented eighteen months of tier-one response across four mid-market SOCs. Automation cut volume, but only the playbooks with explicit rollback paths held up under real incidents.</description></item><item><title>Pharmaceutical Client: Segmenting a Flat Network Without Downtime</title><link>https://research.mysticcyber.com/pharma-segmenting-a-flat-network/</link><pubDate>Wed, 15 Jul 2026 00:00:00 +0000</pubDate><guid>https://research.mysticcyber.com/pharma-segmenting-a-flat-network/</guid><description>A regulated manufacturer with a decade-old flat network needed segmentation before an audit. The phased path we ran, the two things that went wrong, and what it cost.</description></item><item><title>Model Context Protocol Servers Are a New Trust Boundary</title><link>https://research.mysticcyber.com/mcp-servers-new-trust-boundary/</link><pubDate>Wed, 08 Jul 2026 00:00:00 +0000</pubDate><guid>https://research.mysticcyber.com/mcp-servers-new-trust-boundary/</guid><description>MCP servers sit between an assistant and your data with very little between them. Treat each one as an unauthenticated integration until proven otherwise.</description></item><item><title>Detection Engineering as a Product Practice</title><link>https://research.mysticcyber.com/detection-engineering-as-product-practice/</link><pubDate>Tue, 30 Jun 2026 00:00:00 +0000</pubDate><guid>https://research.mysticcyber.com/detection-engineering-as-product-practice/</guid><description>Detections decay. Teams that version, test and deprecate them like software keep their true-positive rate flat as the estate grows; teams that do not, do not.</description></item><item><title>Third-Party Risk When the Third Party Ships Agents</title><link>https://research.mysticcyber.com/third-party-risk-when-vendors-ship-agents/</link><pubDate>Sun, 21 Jun 2026 00:00:00 +0000</pubDate><guid>https://research.mysticcyber.com/third-party-risk-when-vendors-ship-agents/</guid><description>Vendor questionnaires assume software behaves the same way twice. Agentic vendors break that assumption — a revised diligence framework for procurement and security to run together.</description></item><item><title>Higher Education: Visibility Across Forty Thousand Unmanaged Devices</title><link>https://research.mysticcyber.com/higher-ed-visibility-unmanaged-devices/</link><pubDate>Fri, 12 Jun 2026 00:00:00 +0000</pubDate><guid>https://research.mysticcyber.com/higher-ed-visibility-unmanaged-devices/</guid><description>A constant mix of student and faculty devices made the attack surface unknowable. Passive discovery plus enrollment nudges got coverage from an estimated 40% to 94% in one semester.</description></item></channel></rss>