18 vulnerabilities across 13 products scored HIGH or above on August 27, 2026.

  • 🟠 HIGH: 18

Exploit Status Upgrades

The following CVEs from previous bulletins have been upgraded based on new exploit intelligence:

  • [UPGRADED] CVE-2026-65641 (veeam/one) — F1: exploitable → functional, AAS: 10.4 → 12.4 (HIGH → CRITICAL). Originally in 2026-08-26 bulletin.
  • [UPGRADED] CVE-2026-76904 (geotools/geotools) — F1: exploitable → functional, AAS: 10.6 → 12.6 (HIGH → CRITICAL). Originally in 2026-08-21 bulletin.
  • [UPGRADED] CVE-2026-72843 (evershopcommerce/evershop) — F1: exploitable → functional, AAS: 10.2 → 12.2 (HIGH → CRITICAL). Originally in 2026-08-20 bulletin.

🟠 [HIGH] redis/redis

1 CVE | CVSS 4.0: 9.2 | AAS 11.2

  • cpe:2.3:a:redis:redis:*:*:*:*:*:*:*:*

Redis — Critical TLS Use-After-Free (1 CVE)

A high-severity use-after-free vulnerability in Redis affects deployments configured with TLS support. CVE-2026-81934 exists in the tlsProcessPendingData() function and can be exploited by a remote, unauthenticated attacker to execute arbitrary commands with the privileges of the Redis server process, making this a critical risk for any internet-facing or network-accessible Redis instance using TLS. Security teams running affected versions should prioritize upgrading immediately to Redis 8.2.9, 8.4.6, 8.6.6, 8.8.2, or 8.10.1, and should review their environments for any signs of exploitation given that this vulnerability is considered exploitable.

Vendor Advisory

🟠 [HIGH] triliumnext/trilium

6 CVEs | CVSS 4.0: 9.3 | AAS 10.7

  • cpe:2.3:a:triliumnext:trilium:*:*:*:*:*:*:*:* (< 0.104.0)

TriliumNext Trilium — Multiple HTML Sanitization Bypasses (6 CVEs)

Six vulnerabilities have been disclosed in TriliumNext Trilium, the open-source hierarchical note-taking application, affecting versions through 0.103.0, including multiple exploitable flaws that allow attackers to bypass the “Safe import” HTML sanitization filter and achieve script execution through crafted import archives. The most severe issue carries a CVSS 4.0 score of 9.3 and exploits the fact that book-type notes are excluded from sanitization yet rendered as HTML, enabling cross-site scripting via jQuery’s html method. Organizations and individuals using Trilium for note-taking should update beyond version 0.103.0 immediately and review any previously imported archives from untrusted sources for signs of embedded payloads.

Vendor Advisory

🟠 [HIGH] themeum/tutor_lms

1 CVE | CVSS 3.1: 9.8 | AAS 10.6

  • cpe:2.3:a:themeum:tutor_lms:*:*:*:*:*:*:*:*

Themeum Tutor LMS — Unauthenticated Arbitrary PHP Function Call (1 CVE)

A critical vulnerability in the Tutor LMS WordPress plugin before version 4.0.6 allows unauthenticated attackers to overwrite internal template variables via request data, enabling them to invoke arbitrary zero-argument PHP functions and retrieve their output. With a CVSS 3.1 score of 9.8, this flaw requires no authentication and is considered exploitable, posing a severe risk to any WordPress site running the affected plugin. Administrators should update Tutor LMS to version 4.0.6 or later immediately and audit their sites for any indicators of compromise, particularly unexpected function calls or anomalous request patterns targeting LMS template endpoints.

Vendor Advisory

🟠 [HIGH] mims-harvard/tooluniverse

1 CVE | CVSS 4.0: 9.3 | AAS 9.7

  • cpe:2.3:a:mims-harvard:tooluniverse:*:*:*:*:*:*:*:*

MIMS-Harvard ToolUniverse — Unauthenticated Python Sandbox Escape to RCE (1 CVE)

A critical vulnerability in ToolUniverse, the Harvard MIMS research tool framework, allows remote unauthenticated attackers to escape the Python code execution sandbox and achieve arbitrary code execution on the host server. CVE-2026-81096 exploits an incomplete denylist in the python_code_executor tool, where attackers can traverse Python’s class hierarchy via dunder attributes to access restricted builtins, bypassing the sandbox on a server that enforces no authentication. Organizations running ToolUniverse instances should take them offline or place them behind authentication immediately, review the project’s GitHub repository for patches, and audit server logs for any signs of unauthorized code execution.

Vendor Advisory

🟠 [HIGH] paolo/geodirectory

1 CVE | CVSS 3.1: 8.8 | AAS 9.6

  • cpe:2.3:a:paolo:geodirectory:*:*:*:*:*:*:*:*

GeoDirectory WordPress Plugin — Unauthenticated CSRF (1 CVE)

A high-severity cross-site request forgery vulnerability affects the GeoDirectory WordPress plugin in versions 2.8.176 and earlier, allowing attackers to trick authenticated users into performing unintended actions by visiting a malicious page. Rated CVSS 3.1 8.8, CVE-2026-81271 requires no authentication to exploit and poses a significant risk to any WordPress site using this popular business directory plugin. Site administrators should update GeoDirectory beyond version 2.8.176 immediately and review recent administrative actions for any unauthorized changes that may indicate prior exploitation.

Vendor Advisory

🟠 [HIGH] mcp-use/mcp-use

1 CVE | CVSS 4.0: 8.7 | AAS 9.6

  • cpe:2.3:a:mcp-use:mcp-use:*:*:*:*:*:*:*:*

mcp-use Inspector — Server-Side Request Forgery via Unrestricted Proxy (1 CVE)

A high-severity SSRF vulnerability in mcp-use’s inspector proxy middleware allows attackers to reach internal network resources through the application server. CVE-2026-81091 exists because the proxy in mountMcpProxy accepts caller-controlled destination URLs via the X-Target-URL header or __mcp_target parameter without validating the target host, permitting requests to loopback, link-local, and private addresses, including through DNS rebinding and open redirects. Teams running mcp-use inspector instances should restrict or disable the proxy component immediately, monitor for signs of internal network reconnaissance, and check the project’s GitHub repository for a patched release.

Vendor Advisory

🟠 [HIGH] speechbrain/speechbrain

1 CVE | CVSS 4.0: 8.7 | AAS 9.6

  • cpe:2.3:a:speechbrain:speechbrain:*:*:*:*:*:*:*:* (< 1.1.1)

SpeechBrain — Arbitrary Code Execution via Malicious Checkpoint YAML (1 CVE)

A high-severity arbitrary code execution vulnerability in SpeechBrain before version 1.1.1 allows attackers to execute arbitrary code by placing a crafted CKPT.yaml file in a checkpoint directory. CVE-2026-10036 exploits the use of PyYAML’s unsafe loader during candidate enumeration in Checkpointer.recover_if_possible(), where malicious Python object construction tags such as !!python/object/apply are processed automatically during checkpoint discovery, even before a checkpoint is explicitly loaded. Teams using SpeechBrain for speech and audio ML workflows should upgrade to version 1.1.1 or later immediately and audit any checkpoint directories sourced from untrusted origins for suspicious YAML content.

Vendor Advisory

🟠 [HIGH] weptile/mobile_app_for_woocommerce

1 CVE | CVSS 3.1: 8.6 | AAS 9.4

  • cpe:2.3:a:weptile:mobile_app_for_woocommerce:*:*:*:*:*:*:*:*

Weptile Mobile App for WooCommerce — Unauthenticated Broken Access Control (1 CVE)

A high-severity broken access control vulnerability in the Mobile App for WooCommerce WordPress plugin affects versions 0.4.62 and earlier, allowing unauthenticated attackers to access restricted functionality or data without valid credentials. Rated CVSS 3.1 8.6, CVE-2026-27330 is particularly concerning for WooCommerce store operators as unauthorized access could expose customer data, order information, or administrative functions through the mobile app integration layer. Site administrators should update the plugin beyond version 0.4.62 immediately, review access logs for unauthorized API activity, and audit their WooCommerce stores for any signs of data exposure or unauthorized modifications.

Vendor Advisory

🟠 [HIGH] craftcms/cms

1 CVE | CVSS 4.0: 8.7 | AAS 9.3

  • cpe:2.3:a:craftcms:cms:*:*:*:*:*:*:*:*

Craft CMS — Authenticated RCE via Twig Sandbox Bypass (1 CVE)

A high-severity vulnerability in Craft CMS allows authenticated users to achieve remote code execution by exploiting dangerous Yii framework functionality exposed through the Twig template sandbox. CVE-2026-79988, rated CVSS 4.0 8.7, follows a pattern similar to previously disclosed Craft CMS sandbox escape flaws, meaning organizations that patched earlier variants should treat this as a continued risk requiring attention. Craft CMS administrators should monitor the project’s GitHub repository for a patched release, restrict template editing permissions to trusted users only, and review audit logs for any suspicious template modifications that may indicate exploitation attempts.

Vendor Advisory

🟠 [HIGH] villatheme/suggestion_engine_for_woocommerce

1 CVE | CVSS 3.1: 8.5 | AAS 9.3

  • cpe:2.3:a:villatheme:suggestion_engine_for_woocommerce:*:*:*:*:*:*:*:*

VillaTheme Suggestion Engine for WooCommerce — SQL Injection (1 CVE)

A high-severity SQL injection vulnerability affects the Suggestion Engine for WooCommerce plugin in versions 2.0.11 and earlier, exploitable by users with contributor-level privileges or higher. CVE-2026-81277, rated CVSS 3.1 8.5, could allow attackers to extract sensitive data from the WordPress database, including customer information, order details, and user credentials stored by WooCommerce. Site administrators running this plugin should update beyond version 2.0.11 immediately, audit database access logs for anomalous queries, and consider whether any contributor-level accounts may have been compromised or misused.

Vendor Advisory

🟠 [HIGH] dolibarr/dolibarr

1 CVE | CVSS 4.0: 8.8 | AAS 9.2

  • cpe:2.3:a:dolibarr:dolibarr:*:*:*:*:*:*:*:* (>= 9.0.0, < 23.0.5)

Dolibarr ERP/CRM — Path Traversal via Email Attachment Filename (1 CVE)

A high-severity path traversal vulnerability affects Dolibarr versions 9.0.0 through 23.0.4, where the email collector’s saveAttachment() function writes inbound email attachments using the MIME-supplied filename without sanitizing it to a safe basename. CVE-2026-81730, rated CVSS 4.0 8.8, allows a remote attacker to craft an email with a malicious attachment filename to write arbitrary files to attacker-controlled paths on the server, potentially achieving remote code execution by placing a webshell in a publicly accessible directory. Organizations running Dolibarr with the email collector feature enabled should update beyond version 23.0.4 immediately, audit their filesystems for unexpected files outside normal attachment directories, and review inbound email logs for suspicious attachment names containing path traversal sequences.

Vendor Advisory

🟠 [HIGH] nltk/nltk

1 CVE | CVSS 4.0: 8.3 | AAS 9.1

  • cpe:2.3:a:nltk:nltk:*:*:*:*:*:*:*:*

NLTK — Path Traversal Bypass in Model-Artifact APIs (1 CVE)

A high-severity path traversal vulnerability in NLTK through version 3.10.3 allows attackers to read or write arbitrary files outside the intended sandbox roots by exploiting raw file operations in several model-artifact APIs, including TransitionParser, AveragedPerceptron, PerceptronTagger, and maxent parameter handlers. CVE-2026-81726, rated CVSS 4.0 8.3, bypasses NLTK’s pathsec enforcement and is particularly concerning for applications that process untrusted model files or allow user-controlled paths in NLP pipelines. Teams using NLTK in production or research environments should monitor the project’s GitHub security advisory for a patched release, restrict model loading to trusted sources only, and audit any workflows where external input influences model file paths.

Vendor Advisory

🟠 [HIGH] flowintel/flowintel

1 CVE | CVSS 4.0: 9.1 | AAS 9.0

  • cpe:2.3:a:flowintel:flowintel:*:*:*:*:*:*:*:*

Flowintel — Session Persistence After Password Change (1 CVE)

A high-severity session management vulnerability in Flowintel allows previously authenticated sessions to remain valid after a user’s password is changed, meaning an attacker with a stolen or compromised session token retains full access until the session expires naturally. CVE-2026-81826, rated CVSS 4.0 9.1, is especially dangerous in incident response scenarios where a defender changes credentials believing it will revoke attacker access. Organizations running Flowintel should apply the upstream fix immediately, manually invalidate all active sessions after any password change or suspected compromise, and review session logs for any persistent access that continued beyond a credential reset.

Vendor Advisory