<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>PentStark Blog</title>
    <link>https://pentstark.com/blog/</link>
    <description>CVE analyses, TTP walkthroughs, and methodology deep-dives from the PentStark research team.</description>
    <language>en-us</language>
    <lastBuildDate>Wed, 11 Nov 2026 09:34:56 GMT</lastBuildDate>
    <atom:link href="https://pentstark.com/blog/rss.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Containment that does not contain: what still works after you reset, reimage and revoke</title>
      <link>https://pentstark.com/blog/containment-that-does-not-contain-what-still-works-after-you-reset-reimage-and-revoke/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/containment-that-does-not-contain-what-still-works-after-you-reset-reimage-and-revoke/</guid>
      <pubDate>Wed, 11 Nov 2026 09:34:56 GMT</pubDate>
      <description>Issued certificates, Kerberos tickets, access tokens, refresh tokens and app credentials can outlive the usual response steps. A test plan for measuring time-to-effective-revocation in Microsoft identity environments.</description>
      <category>Identity</category>
    </item>
    <item>
      <title>Coding agents trust by location, and six October findings show the gap</title>
      <link>https://pentstark.com/blog/coding-agents-trust-by-location-and-six-october-findings-show-the-gap/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/coding-agents-trust-by-location-and-six-october-findings-show-the-gap/</guid>
      <pubDate>Tue, 10 Nov 2026 09:04:40 GMT</pubDate>
      <description>Repo settings, gateways, extensions, server-supplied URLs and rewritten config: six recent reports show coding agents trusting where something sits, with no CVE to flag it. A test list for developer environments.</description>
      <category>AI Security</category>
    </item>
    <item>
      <title>The vendor&apos;s app secret is your standing access: testing multi-tenant Entra apps</title>
      <link>https://pentstark.com/blog/the-vendors-app-secret-is-your-standing-access-testing-multi-tenant-entra-apps/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/the-vendors-app-secret-is-your-standing-access-testing-multi-tenant-entra-apps/</guid>
      <pubDate>Mon, 09 Nov 2026 17:34:33 GMT</pubDate>
      <description>The ConnectWise Cloud Backup incident shows why tenant reviews and pentests must treat consented third-party Entra apps as an identity class, with their own permissions, credentials and logging.</description>
      <category>Identity</category>
    </item>
    <item>
      <title>CERT-In&apos;s AI audit guidelines turn &quot;AI pentest&quot; into a tiered scope document</title>
      <link>https://pentstark.com/blog/cert-ins-ai-audit-guidelines-turn-ai-pentest-into-a-tiered-scope-document/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/cert-ins-ai-audit-guidelines-turn-ai-pentest-into-a-tiered-scope-document/</guid>
      <pubDate>Mon, 09 Nov 2026 17:34:19 GMT</pubDate>
      <description>India&apos;s CERT-In AI Security Auditing Guidelines v1.0 tie testing depth to a four-tier risk rating and list factors that push agentic systems up a tier. Here is what that means for scoping an AI penetration test.</description>
      <category>Compliance</category>
    </item>
    <item>
      <title>RC4 is no longer the Kerberos default, but Kerberoasting is not gone</title>
      <link>https://pentstark.com/blog/rc4-kerberoasting-after-default-removal/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/rc4-kerberoasting-after-default-removal/</guid>
      <pubDate>Fri, 06 Nov 2026 10:39:59 GMT</pubDate>
      <description>Microsoft&apos;s 2026 change removes RC4 as the assumed Kerberos default. The attack surface that remains is configuration debt: explicit RC4 exceptions, weak service passwords and domain controllers not yet enforcing.</description>
      <category>Active Directory</category>
    </item>
    <item>
      <title>Your repo&apos;s AI-agent and editor config is now an execution surface</title>
      <link>https://pentstark.com/blog/repo-agent-editor-config-execution-surface/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/repo-agent-editor-config-execution-surface/</guid>
      <pubDate>Thu, 05 Nov 2026 09:55:19 GMT</pubDate>
      <description>ChainDrop planted editor and assistant config files to restart its payload. Vendor docs show the guardrails vary by tool and launch mode, so these files belong in code review and developer-environment pentests.</description>
      <category>AI Security</category>
    </item>
    <item>
      <title>Trusted publishing moved the npm secret into an identity boundary</title>
      <link>https://pentstark.com/blog/trusted-publishing-identity-boundary-npm/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/trusted-publishing-identity-boundary-npm/</guid>
      <pubDate>Thu, 05 Nov 2026 09:54:25 GMT</pubDate>
      <description>Replacing npm tokens with OIDC trusted publishing shifts trust to the workflow identity and repo binding. ChainDrop and SlugFest show why release pipelines need testing, and why provenance is not a safety verdict.</description>
      <category>AppSec</category>
    </item>
    <item>
      <title>MCP servers are the new vendor risk: auditing the agent toolbox</title>
      <link>https://pentstark.com/blog/mcp-servers-are-the-new-vendor-risk/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/mcp-servers-are-the-new-vendor-risk/</guid>
      <pubDate>Wed, 22 Apr 2026 00:00:00 GMT</pubDate>
      <description>Every MCP server you plug into your agent is a trust boundary. Here&apos;s how to audit them like the supply-chain risk they are.</description>
      <category>AI Security</category>
    </item>
    <item>
      <title>Passkeys in the enterprise: 18 months in, here&apos;s what breaks</title>
      <link>https://pentstark.com/blog/passkeys-enterprise-18-months-in/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/passkeys-enterprise-18-months-in/</guid>
      <pubDate>Sat, 18 Apr 2026 00:00:00 GMT</pubDate>
      <description>Five recurring attack paths we&apos;ve found in enterprise passkey rollouts since they became table stakes.</description>
      <category>Identity</category>
    </item>
    <item>
      <title>EU AI Act red teaming: what a pentester actually delivers</title>
      <link>https://pentstark.com/blog/eu-ai-act-red-teaming/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/eu-ai-act-red-teaming/</guid>
      <pubDate>Wed, 15 Apr 2026 00:00:00 GMT</pubDate>
      <description>Article 15 and 54a say you need cybersecurity and adversarial testing. Here&apos;s the engagement shape the first audit cycle accepted.</description>
      <category>Compliance</category>
    </item>
    <item>
      <title>AD CS ESC1 in the wild: the certificate template you forgot about</title>
      <link>https://pentstark.com/blog/ad-cs-esc1-misconfiguration/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/ad-cs-esc1-misconfiguration/</guid>
      <pubDate>Fri, 10 Apr 2026 00:00:00 GMT</pubDate>
      <description>How a single ManageCA + supplied-subject template got us to domain admin in 22 minutes.</description>
      <category>Active Directory</category>
    </item>
    <item>
      <title>Your Kubernetes ingress is the new DMZ</title>
      <link>https://pentstark.com/blog/kubernetes-ingress-is-the-new-dmz/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/kubernetes-ingress-is-the-new-dmz/</guid>
      <pubDate>Fri, 03 Apr 2026 00:00:00 GMT</pubDate>
      <description>Ingress controllers parse untrusted input, reach every service, and run privileged. They are a DMZ, not a detail.</description>
      <category>Cloud Security</category>
    </item>
    <item>
      <title>LLM tool-use escalation: from prompt injection to arbitrary code</title>
      <link>https://pentstark.com/blog/llm-tool-use-escalation/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/llm-tool-use-escalation/</guid>
      <pubDate>Sat, 28 Mar 2026 00:00:00 GMT</pubDate>
      <description>A case study on bridging a benign prompt-injection primitive into full agent compromise.</description>
      <category>AI Security</category>
    </item>
    <item>
      <title>DORA ICT third-party testing: lessons from the first audit cycle</title>
      <link>https://pentstark.com/blog/dora-ict-testing-first-cycle-lessons/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/dora-ict-testing-first-cycle-lessons/</guid>
      <pubDate>Fri, 20 Mar 2026 00:00:00 GMT</pubDate>
      <description>What we learned running threat-led penetration tests under DORA Articles 26–27 across the first wave of 2026 audits.</description>
      <category>Compliance</category>
    </item>
    <item>
      <title>BOLA across microservices: when authorization is someone else&apos;s job</title>
      <link>https://pentstark.com/blog/bola-across-microservices/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/bola-across-microservices/</guid>
      <pubDate>Thu, 12 Mar 2026 00:00:00 GMT</pubDate>
      <description>Why BOLA is the most under-tested flaw class in modern SaaS - and how we find it.</description>
      <category>AppSec</category>
    </item>
    <item>
      <title>Lockfile injection: the npm supply-chain attack your SCA missed</title>
      <link>https://pentstark.com/blog/lockfile-injection-supply-chain/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/lockfile-injection-supply-chain/</guid>
      <pubDate>Thu, 05 Mar 2026 00:00:00 GMT</pubDate>
      <description>Pinning your lockfile is necessary but not sufficient. Here&apos;s the pattern attackers are using to slip past SCA tools.</description>
      <category>AppSec</category>
    </item>
    <item>
      <title>Ransomware-ready posture: the assume-breach scenario every board should fund</title>
      <link>https://pentstark.com/blog/ransomware-ready-posture-check/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/ransomware-ready-posture-check/</guid>
      <pubDate>Fri, 20 Feb 2026 00:00:00 GMT</pubDate>
      <description>A red-team scoping guide for boards asking &apos;are we ransomware-ready?&apos;</description>
      <category>Red Team</category>
    </item>
    <item>
      <title>SOC 2 without slowing down your release train</title>
      <link>https://pentstark.com/blog/soc2-without-slowing-down/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/soc2-without-slowing-down/</guid>
      <pubDate>Wed, 04 Feb 2026 00:00:00 GMT</pubDate>
      <description>Patterns for embedding evidence collection into your engineering workflow.</description>
      <category>Compliance</category>
    </item>
    <item>
      <title>Shift-left that actually ships: patterns from ten SSDLC rollouts</title>
      <link>https://pentstark.com/blog/ssdlc-shift-left-patterns/</link>
      <guid isPermaLink="true">https://pentstark.com/blog/ssdlc-shift-left-patterns/</guid>
      <pubDate>Thu, 22 Jan 2026 00:00:00 GMT</pubDate>
      <description>The three patterns that make product-security programs work - and the three that quietly kill them.</description>
      <category>Product Security</category>
    </item>
  </channel>
</rss>
