[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"cve-CVE-2026-96962":3,"landing-trending":42,"landing-articles":127},{"id":4,"cve_id":5,"euvd_id":6,"summary":7,"published":8,"cvss_data":9,"is_remote":19,"cwes":20,"cpes":22,"technologies":23,"references":24,"score":26,"epss_score":27,"epss_percentile":28,"is_kev":29,"cisa_kev_date_added":30,"cisa_kev_due_date":30,"cisa_ssvc":31,"exploits":41},398917,"CVE-2026-96962","EUVD-2026-91949","The Pie Register  WordPress plugin before 3.8.4.14 does not restrict access to an invitation-code report, allowing unauthenticated visitors who know a valid invitation code to obtain the username and email address of every user who registered with that code.","2026-10-03T06:16:00Z",{"cvss_v3.1":10},{"scope":11,"version":12,"baseScore":13,"attackVector":14,"baseSeverity":15,"vectorString":16,"integrityImpact":17,"userInteraction":17,"attackComplexity":18,"availabilityImpact":17,"privilegesRequired":17,"confidentialityImpact":15},"UNCHANGED","3.1",3.7,"NETWORK","LOW","CVSS:3.1\u002FAV:N\u002FAC:H\u002FPR:N\u002FUI:N\u002FS:U\u002FC:L\u002FI:N\u002FA:N","NONE","HIGH",true,[21],"CWE-200",[],[],[25],"https:\u002F\u002Fwpscan.com\u002Fvulnerability\u002Fbf634ddf-277e-46a4-9e5b-7253b3e02979\u002F",0.39,0.00171,0.06169,false,null,{"timestamp":32,"automatable":33,"tech_impact":34,"exploitation":35,"cisa_decision":36,"cisa_remediation_timeline":40},"2026-10-03T14:59:52.889050Z","no","partial","none",{"vector":37,"decision":38,"priority":39},"CISAv1\u002FE:N\u002FA:N\u002FT:P\u002FM:H\u002F2026-10-24T06:30:48.535569\u002F","track_star","medium","60D",[],{"count":43,"next":44,"previous":30,"results":45},264,"http:\u002F\u002Fweb:8606\u002Fapi\u002Ftrending_attacks\u002F?page=2&sorted_by=-published_at",[46,56,60,68,76,85,96,105,111,120],{"id":43,"title":47,"summary":48,"published_at":49,"severity":50,"vendor":51,"products":52},"Oracle Java SE JSSE - Improper Access Control & Unauthorized Data Access & Unauthorized Data Modification","Patrowl scan limitation — this vulnerability cannot be detected by network scanning. The problem is not in a service exposed on the network, but in the list of trusted certificates built into Java — and nothing a scanner can see from the outside tells it which certificates a system accepts, or even whether the system is running Java at all. The only way to test for it from the network would be to use a fraudulent certificate during a secure connection, which is a real attack rather than a scan, and it would still miss most cases. The only reliable way to detect this vulnerability is to check locally which Java versions are installed and compare them with the list of affected versions.\r\n\r\nOracle Java SE is the Java runtime and development platform. Its JSSE (Java Secure Socket Extension) component provides TLS\u002FSSL for secure network communication. Oracle GraalVM is a JDK distribution based on the same platform.\r\n\r\nCVE-2026-91403\r\n\r\nA vulnerability in the JSSE component of Oracle Java SE, Oracle GraalVM for JDK and Oracle GraalVM Enterprise Edition can be exploited by an unauthenticated attacker with network access over multiple protocols. A successful attack can give unauthorized access to critical data and allow unauthorized creation, deletion or modification of data. It can also affect sandboxed Java Web Start applications and applets that run untrusted code. The record lists the weakness as CWE-284 (Improper Access Control) in the CISA ADP entry, while another database lists CWE-295. The record lists these affected versions:\r\n- Oracle Java SE: 8u511, 8u511-perf, 11.0.33, 17.0.21, 21.0.13, 25.0.5, 27.0.1.\r\n- Oracle GraalVM for JDK: 17.0.21, 21.0.13.\r\n- Oracle GraalVM Enterprise Edition: 21.3.20.\r\n\r\nOracle published an out-of-cycle Security Alert on October 21, 2026. One secondary source says the fix makes Java distrust eight compromised intermediate CA certificates shipped with the JDK. The Oracle-focused sources do not confirm this root cause, so check it against the Oracle advisory. CISA ADP SSVC data shows Exploitation: none, Automatable: yes, Technical Impact: total. The CVE is not in the CISA KEV catalog as of October 22, 2026, and its EPSS score is below 1%.","2026-10-23T14:12:32.871954Z",4,"Oracle",[53,54,55],"Oracle Java SE","Oracle GraalVM Enterprise Edition","Oracle GraalVM for JDK",{"id":57,"title":47,"summary":58,"published_at":49,"severity":50,"vendor":51,"products":59},263,"Oracle Java SE is the Java runtime and development platform. Its JSSE (Java Secure Socket Extension) component provides TLS\u002FSSL for secure network communication. Oracle GraalVM is a JDK distribution based on the same platform.\r\n\r\nCVE-2026-91403\r\n\r\nA vulnerability in the JSSE component of Oracle Java SE, Oracle GraalVM for JDK and Oracle GraalVM Enterprise Edition can be exploited by an unauthenticated attacker with network access over multiple protocols. A successful attack can give unauthorized access to critical data and allow unauthorized creation, deletion or modification of data. It can also affect sandboxed Java Web Start applications and applets that run untrusted code. The record lists the weakness as CWE-284 (Improper Access Control) in the CISA ADP entry, while another database lists CWE-295. The record lists these affected versions:\r\n- Oracle Java SE: 8u511, 8u511-perf, 11.0.33, 17.0.21, 21.0.13, 25.0.5, 27.0.1.\r\n- Oracle GraalVM for JDK: 17.0.21, 21.0.13.\r\n- Oracle GraalVM Enterprise Edition: 21.3.20.\r\n\r\nOracle published an out-of-cycle Security Alert on October 21, 2026. One secondary source says the fix makes Java distrust eight compromised intermediate CA certificates shipped with the JDK. The Oracle-focused sources do not confirm this root cause, so check it against the Oracle advisory. CISA ADP SSVC data shows Exploitation: none, Automatable: yes, Technical Impact: total. The CVE is not in the CISA KEV catalog as of October 22, 2026, and its EPSS score is below 1%.",[53,54,55],{"id":61,"title":62,"summary":63,"published_at":64,"severity":50,"vendor":65,"products":66},260,"XWiki Platform - Unauthenticated Remote Code Execution via SolrSearch Template Injection","XWiki Platform is an open-source enterprise wiki and application platform written in Java, offering runtime services for building collaborative applications on top of it.\r\n\r\nCVE-2025-24893\r\n\r\nInsufficient neutralization of directives in dynamically evaluated code (CWE-95, eval injection) in the SolrSearch macro allows any unauthenticated guest to perform arbitrary remote code execution through a specially crafted request to SolrSearch (e.g. \u003Chost>\u002Fxwiki\u002Fbin\u002Fget\u002FMain\u002FSolrSearch?media=rss&text=...), injecting and executing arbitrary Groovy code via template injection. This impacts the confidentiality, integrity, and availability of the entire XWiki installation. Affected versions are XWiki Platform from 5.3-milestone-2 up to (but excluding) 15.10.11, and from 16.0.0-rc-1 up to (but excluding) 16.4.1. The flaw was patched in XWiki 15.10.11, 16.4.1, and 16.5.0RC1; users unable to upgrade can mitigate it by editing Main.SolrSearchMacros (line 955 in SolrSearchMacros.xml) to match the rawResponse macro in macros.vm#L2824 with a content type of application\u002Fxml. The vulnerability has been actively exploited in the wild, including a two-stage attack chain delivering a coinminer, although it is not in the CISA KEV catalog.","2026-10-23T08:48:16.092785Z","XWiki",[67],"XWiki Platform",{"id":69,"title":70,"summary":71,"published_at":72,"severity":50,"vendor":73,"products":74},262,"Arista VeloCloud Orchestrator - Authenticated SQLi to Privilege Escalation","VeloCloud Orchestrator (VCO) is the central management and orchestration platform for Arista VeloCloud SD-WAN. It is used to configure and monitor edges, gateways, and enterprise tenants.\r\n\r\nCVE-2026-114625\r\n\r\nA security issue in VeloCloud Orchestrator On-Prem lets a remote, authenticated attacker reach privileged internal functionality and affect the VCO host. Any authenticated user can gain operator privileges and act as a local \"superuser operator\". The minimum role needed is Enterprise Read Only. The web interface is exposed by default and no configuration prevents the exposure. The weakness is mapped to CWE-89 (SQL Injection) and CAPEC-66. CISA ADP SSVC lists exploitation as \"none\", automatable as \"no\", and technical impact as \"total\". Secondary sources disagree on whether it is being exploited, so check Arista's advisory directly. Affected on-prem versions are 5.2.0 before 5.2.3.18, 6.1.0 before 6.1.3.10, and 6.4.0 before 6.4.2.12. Hosted and Dedicated VCO deployments were also affected but have already been patched. The fix is to upgrade to 5.2.3.18, 6.1.3.10, 6.4.2.12 or later. Until then, restrict web interface access to trusted administrative networks and monitor for suspicious activity.","2026-10-23T07:57:12.173417Z","Arista Networks",[75],"VeloCloud Orchestrator",{"id":77,"title":78,"summary":79,"published_at":80,"severity":50,"vendor":81,"products":82},261,"Portainer - Standard user can bypass the bind-mount restriction and take over the Docker host","Portainer is a lightweight service delivery platform for containerized applications, used to manage Docker, Swarm, Kubernetes and ACI environments through a web UI and an API proxy that can enforce per-environment security restrictions for non-administrator users.\r\n\r\nGHSA-j3mf-q9qf-5qf8\r\n\r\nImproper input validation in the Compose stack validator IsValidStackFile (api\u002Fstacks\u002Fstackutils\u002Fvalidation.go): when the per-environment setting \"Hide bind mounts for non-administrators\" is enabled, the validator rejected bind mounts declared in top-level volumes and each service's volumes, but never inspected a service's volumes_from field. By declaring volumes_from pointing at an existing container, any authenticated Standard user with stack-management access to a Docker standalone environment inherits all of that container's mounts — including host bind mounts. Since the documented Portainer Server\u002FAgent deployments mount the Docker socket, an attacker can gain full control of the host Docker daemon (and from there root on the host), or read Portainer's own database and stored credentials by inheriting the Server's data directory. Affected versions: ≥ 2.39.0 ≤ 2.39.8 and ≥ 2.40.0 ≤ 2.45.2 (both editions; Swarm stacks not affected). Fixed in 2.45.3, released 2026-10-22; no patch for 2.39.x or EOL 2.40.x–2.44.x — upgrade to 2.45.3. Interim mitigations: disable stack usage for non-administrators, restrict standard-user environment access, and audit existing non-administrator stacks for volumes_from. No CVE ID had been assigned at publication time.","2026-10-23T07:54:21.608453Z","Portainer",[83,84],"Portainer Community Edition","Portainer Business Edition",{"id":86,"title":87,"summary":88,"published_at":89,"severity":50,"vendor":90,"products":91},258,"Multiple Vulnerabilities in Cisco Unified Communications Products - Injection & Authentication Bypass","Cisco Unified Communications Manager and its companion products (Emergency Responder, Unity Connection, IM & Presence, SME, Prime Collaboration Deployment) form Cisco's enterprise VoIP and collaboration platform, providing call control, emergency call routing, voicemail, and presence services for enterprise telephony networks.\r\n\r\nCVE-2026-76473\r\n\r\nImproper control of a resource through its lifetime (CWE-74, injection-related) in multiple Cisco voice products, including Cisco Emergency Responder, Unified CM, Unified CM IM & Presence Service, and Unity Connection. It is remotely exploitable with no privileges or user interaction required, and can result in full compromise of confidentiality, integrity, and availability (CVSS 9.8). Affected releases include Emergency Responder, Unified CM, IM & Presence, and Unity Connection versions 14 and 15 (14 through 14SU6, and 15 through 15SU4 and their maintenance releases). Cisco addressed the flaw in its October 2026 Unified Communications Products Security Hardening Release, which groups several internally discovered vulnerabilities fixed during a comprehensive security review. (Strix CVE entry, NHS Digital alert)\r\n\r\nCVE-2026-76474\r\n\r\nImproper authentication (CWE-287) affecting the same set of Cisco voice products. The flaw covers authentication bypass and session-fixication-style weaknesses; a remote, unauthenticated attacker without user interaction can fully compromise the affected system (CVSS 9.8). The same product versions are affected (Emergency Responder, Unified CM, Unified CM IM & Presence Service, and Unity Connection releases 14.x and 15.x up to 14SU6 \u002F 15SU4). It was disclosed alongside the other flaws in Cisco's October 2026 security hardening release, and customers should upgrade to the fixed hardening releases listed in the vendor advisory. (Strix CVE entry, Security Online)","2026-10-22T14:31:16.035692Z","Cisco",[92,93,94,95],"Cisco Emergency Responder","Cisco Unified Communications Manager","Cisco Unity Connection","Cisco Unified CM IM & Presence Service (IM&P)",{"id":97,"title":98,"summary":99,"published_at":100,"severity":50,"vendor":101,"products":102},259,"Node.js (bundled Brotli) - Out-of-Bounds Write & Remote Code Execution","Node.js is a cross-platform JavaScript runtime built on V8, widely used for server-side web applications and APIs. Google Brotli is a lossless compression library whose decoder Node bundles in deps\u002Fbrotli and exposes through node:zlib; compressed HTTP responses (including Brotli Content-Encoding) are automatically decompressed by fetch() and frameworks such as Express 5.\r\n\r\nCVE-2026-111497\r\n\r\nAn out-of-bounds write in the Brotli decoder (c\u002Fdec\u002Fdecode.c) bundled with Node.js. When a Node.js process decodes a malformed Brotli-compressed stream from an attacker — for example an HTTP response body auto-decompressed via fetch(), Express 5, or node:zlib — the decoder fails to validate input boundaries and writes past buffer limits, which the Node.js project says can permit remote code execution (memory disclosure and arbitrary code execution). The same flaw exists in upstream Google Brotli versions 1.0.3 through 1.2.0. It affects the Node.js 22.x, 24.x, and 26.x release lines; end-of-life lines (e.g. 20.x, 25.x) receive no fix and are stated to be \"always affected\". Fixed in Node.js v22.23.4, v24.22.1, and v26.12.1 (October 21, 2026), based on upstream Brotli commit 0ed4268 (\"Detect and report invalid stream earlier\", Evgenii Kliuchnikov, Google). Upstream Brotli shipped fixes in 1.2.1, 1.1.1, and 1.0.10. No workaround is documented; users should upgrade Node and restart all Node processes, and rebuild any containers, single-executable apps, or appliances that bundle a Node binary. Distro packages linked against a system Brotli library depend on the distro's brotli package instead.","2026-10-22T14:08:04.199100Z","Node.js",[103,104],"Node.js runtime","Google Brotli",{"id":106,"title":107,"summary":108,"published_at":109,"severity":50,"vendor":81,"products":110},257,"Portainer - Query Parameter Confusion in Kubernetes Shell Leading to Arbitrary Command Execution in Any Pod","Portainer is a web-based container and Kubernetes management platform that provides GUI-driven administration of Docker, Swarm, and Kubernetes environments, including a browser-based kubectl shell for Kubernetes clusters managed through the Portainer Agent.\r\nGHSA-7423-r5wh-26x7\r\nImproper handling of extra parameters (CWE-235) combined with incorrect authorization (CWE-863) in the Kubernetes shell WebSocket handler (api\u002Fhttp\u002Fhandler\u002Fwebsocket\u002Fshell_pod.go). When a user opens the kubectl shell, the Portainer server creates a short-lived shell Pod and rewrites the WebSocket request to target it via the Agent's Pod-exec endpoint. The rewrite used Query().Add() to append the server-chosen namespace, podName, containerName, and command values instead of replacing them, while the Agent resolves each parameter by its first occurrence — so an authenticated standard-role user could supply duplicate query parameters and redirect the exec to any Pod and command of their choosing. On Agent-managed and Edge-Agent-managed Kubernetes environments the exec runs under the Agent's in-cluster service account, which the published manifest binds to cluster-admin, so the attacker obtains arbitrary command execution in any Pod in any namespace, full disclosure of cluster Secrets, and potential compromise of the Agent itself; Kubernetes RBAC provides no mitigation. Affected versions: >= 2.39.0 and \u003C 2.39.6 (2.39.x LTS) and all versions prior to 2.45.0, since the flawed behavior was introduced with the kubectl shell feature in 2.9.0. Fixed in 2.39.6 (LTS backport) and 2.45.0 (LTS); the 2.40–2.44 short-term support lines receive no dedicated patch. Local Kubernetes connections are not affected. Reported by D0HY30N on 2026-07-30; no CVE ID has been assigned as of publication. (Source: GitHub Security Advisory GHSA-7423-r5wh-26x7)","2026-10-22T09:12:41.197853Z",[81],{"id":112,"title":113,"summary":114,"published_at":115,"severity":50,"vendor":116,"products":117},256,"Multiple Vulnerabilities in GitLab CE\u002FEE - Authenticated Remote Code Execution via Regular Expression Handling in CI\u002FCD Configuration","GitLab is a web-based DevOps platform providing source code management (Git repositories), CI\u002FCD pipelines, issue tracking and collaboration, widely deployed as self-managed instances (the affected configuration) in addition to GitLab's SaaS offering.\r\n\r\nCVE-2026-89078\r\n\r\nDouble free issue in GitLab's regular expression parser, exploited by supplying a specially crafted regular expression in a CI\u002FCD configuration. Under certain conditions it could allow an authenticated user to execute arbitrary code on the self-managed GitLab server. It affects GitLab CE\u002FEE versions 19.2 before 19.2.7, 19.3 before 19.3.3, and 19.4 before 19.4.1, and was patched in the September 23, 2026 critical patch release (19.2.7, 19.3.3, 19.4.1), rated 9.9 CVSS by GitLab. Reported by the same researcher (joaxcar) as the related regex-compiler flaws. (Strix, Cibersafety, The Hacker News)\r\n\r\nCVE-2026-93577\r\n\r\nInteger overflow issue in the regular expression compiler used when processing CI\u002FCD configurations: an authenticated user who causes GitLab to compile a specially crafted regular expression in a CI\u002FCD configuration could, under certain conditions, execute arbitrary code on the server. It affects the same version ranges (19.2 before 19.2.7, 19.3 before 19.3.3, 19.4 before 19.4.1) and was fixed alongside the double-free flaw in the September 23, 2026 release (19.4.1 and backports), rated 9.9 CVSS. (Strix, Mallory, IT-Connect)","2026-10-22T07:54:49.013651Z","GitLab",[118,119],"GitLab Community Edition (CE)","GitLab Enterprise Edition (EE)",{"id":121,"title":122,"summary":123,"published_at":124,"severity":50,"vendor":125,"products":126},255,"GitLab - Unauthenticated Remote Code Execution","This control is the Patrowl response to CVE-2026-110264, a critical remote code execution vulnerability in GitLab Community Edition and Enterprise Edition.\r\n\r\nPatrowl identifies every asset exposing a self-managed GitLab instance. When the version can be read, it is compared with the affected range. An asset running an affected version is reported as impacted. An asset running GitLab with an unknown version is reported as a warning. A warning means potential exposure, not a confirmed vulnerability.\r\n\r\nAbout the vulnerability. GitLab rated the flaw CVSS 10.0 and released fixes on 21 October 2026. It is an integer overflow in the regular expression compiler. Under certain conditions, an unauthenticated remote attacker can run arbitrary code on the GitLab server. GitLab has not published these conditions. No authentication is required. All versions from 7.0 are affected, which covers more than ten years of releases. \r\n\r\nGitLab.com and GitLab Dedicated are already patched. Only self-managed instances are concerned.\r\n\r\nAffected versions:\r\n- 7.0 up to 18.11.12\r\n- 19.0 up to 19.0.9\r\n- 19.1 up to 19.1.8\r\n- 19.2 up to 19.2.8\r\n- 19.3 up to 19.3.5\r\n- 19.4 up to 19.4.2\r\n- 19.5.0\r\n\r\nFirst recommendation:\r\n\r\nUpdate GitLab (only real fix). Upgrade to 19.5.1, 19.4.3, 19.3.6, 19.2.9, 19.1.9, 19.0.10 or 18.11.13. Start with instances reachable from the internet. Versions older than 18.11 receive no fix. They require a migration to a supported branch.\r\n\r\nReduce the exposure until the update is deployed. GitLab does not publish a workaround. If the instance cannot be updated at once, restrict access to it. Allow only your VPN or a list of trusted IP addresses. Do not leave the sign-in page open to the internet.\r\n\r\nWatch for exploitation attempts. GitLab published two detections for self-managed instances: \"Onigmo native worker crash\" and \"Org mode render error burst\". Look for them in your logs. Unexplained crashes of Puma or Sidekiq workers deserve an investigation.","2026-10-21T22:02:26.003181Z","gitlab",[125],{"count":128,"next":129,"previous":30,"results":130},30,"http:\u002F\u002Fweb:8606\u002Fapi\u002Farticles\u002F?page=2&sorted_by=-published_at",[131,138,146,154,161,168,175,180,187,194],{"id":128,"title":132,"summary":133,"published_at":134,"type":135,"link":136,"image_url":137},"PHP2Shell: Two Critical RCE Flaws in PHP (CVSS 9.2)","Unauthenticated code execution demonstrated on WordPress, Drupal and Joomla, and reproduced in our lab. Patched versions and steps to take today.","2026-10-22T00:00:00Z","cve","https:\u002F\u002Fpatrowl.io\u002Fen\u002Fblog\u002Fphp2shell-2-critical-rce-flaws-php-cvss-9-2","https:\u002F\u002Fpatrowl.io\u002Fmedia\u002Fsite\u002Fc0615c4ee6-1792684469\u002Fphp2shell-2-critical-rce-flaws-php-cvss-9-2.png",{"id":139,"title":140,"summary":141,"published_at":142,"type":143,"link":144,"image_url":145},29,"Manage External Security Across All Your Entities","Centralize the management of entities, users and external security resources while maintaining strict data isolation with Patrowl.","2026-09-15T00:00:00Z","security tips","https:\u002F\u002Fpatrowl.io\u002Fen\u002Fblog\u002Fmulti-entity-external-security-how-structure-governance-group-level","https:\u002F\u002Fpatrowl.io\u002Fmedia\u002Fsite\u002F8b2c0cc5eb-1789551467\u002Fvisu_1_multi_tenancy_dark_1x.webp",{"id":147,"title":148,"summary":149,"published_at":150,"type":151,"link":152,"image_url":153},28,"Will AI Become the Next Hacker? Bruce Schneier Answers at DEF CON 34","Tax law, finance, legislation: Bruce Schneier shows how AI could hack human rule systems just as easily as it hacks software.","2026-09-14T00:00:00Z","hacks","https:\u002F\u002Fpatrowl.io\u002Fen\u002Fblog\u002Fbruce-schneier-def-con-34","https:\u002F\u002Fpatrowl.io\u002Fmedia\u002Fsite\u002F5c616d9db4-1789462598\u002Fai-hacker.en.png",{"id":155,"title":156,"summary":157,"published_at":158,"type":143,"link":159,"image_url":160},27,"Exposure Management: definition, stakes, and how it works","Vulnerabilities, misconfigurations, poorly managed identities: understand what Exposure Management is and why it matters in 2026.","2026-09-10T00:00:00Z","https:\u002F\u002Fpatrowl.io\u002Fen\u002Fblog\u002Fdefinition-exposure-management","https:\u002F\u002Fpatrowl.io\u002Fmedia\u002Fsite\u002F06170638a9-1789050834\u002Fexposure.management.definition.en.png",{"id":162,"title":163,"summary":164,"published_at":165,"type":151,"link":166,"image_url":167},26,"DGFiP Cyberattack 2026: Inside the Attacker","Two distinct attacks hit France","2026-08-17T00:00:00Z","https:\u002F\u002Fpatrowl.io\u002Fen\u002Fblog\u002Fdgfip-cyberattack-2026-inside-attacker-mind","https:\u002F\u002Fpatrowl.io\u002Fmedia\u002Fsite\u002Fdc38d33696-1787055371\u002Fdgfip-2026.png",{"id":169,"title":170,"summary":171,"published_at":172,"type":135,"link":173,"image_url":174},25,"CVE-2026-64638: XSS2Shell, a WordPress Login Vulnerability","Pre-auth XSS on wp-login.php, patched August 6. What is actually exploitable, and how to read it in your Patrowl report.","2026-08-10T00:00:00Z","https:\u002F\u002Fpatrowl.io\u002Fen\u002Fblog\u002Fxss2shell-critical-wordpress-login-vulnerability-2026","https:\u002F\u002Fpatrowl.io\u002Fmedia\u002Fsite\u002F904e61748b-1786369212\u002Fxss2shell.png",{"id":176,"title":177,"summary":177,"published_at":172,"type":135,"link":178,"image_url":179},24,"CVE-2026-64638 : XSS2Shell, faille du login WordPress","https:\u002F\u002Fpatrowl.io\u002Fen\u002Fblog\u002Fcve-2026-64638-xss2shell-faille-login-wordpress","",{"id":181,"title":182,"summary":183,"published_at":184,"type":135,"link":185,"image_url":186},23,"How Agentic AI Hunts Down CVEs Before Attackers Do","38 vulnerability sources monitored around the clock, pentester-validated proof of exploitability, an alert within hours. See how it works.","2026-08-05T00:00:00Z","https:\u002F\u002Fpatrowl.io\u002Fen\u002Fblog\u002Fproactive-threat-hunting-agentic-ai-cve","https:\u002F\u002Fpatrowl.io\u002Fmedia\u002Fsite\u002F804988ce7b-1785930160\u002Fproactive-threat-hunting-how-agentic-ai-tracks-cves.png",{"id":188,"title":189,"summary":190,"published_at":191,"type":151,"link":192,"image_url":193},22,"AI Agent Incident: OpenAI HuggingFace — Risks and Lessons","An autonomous AI agent escaped its sandbox via a JFrog Artifactory zero-day, compromised HuggingFace via privilege escalation and stolen credentials. Reconstructed timeline and security lessons.","2026-07-31T00:00:00Z","https:\u002F\u002Fpatrowl.io\u002Fen\u002Fblog\u002Fai-agent-incident-openai-huggingface-risks","[Image URL from Kirby]",{"id":195,"title":196,"summary":197,"published_at":198,"type":143,"link":199,"image_url":200},21,"SaaS Under Watch: What Your SDLC Doesn","Your pipeline is green. Your real exposure may have changed without you knowing it. The complete guide for software vendors.","2026-07-29T00:00:00Z","https:\u002F\u002Fpatrowl.io\u002Fen\u002Fblog\u002Fsecuring-saas-application-software-vendor","https:\u002F\u002Fpatrowl.io\u002Fmedia\u002Fsite\u002F20d6ec0cd0-1785331401\u002Fapplication-security.png"]