19 vulnerabilities across 15 products scored HIGH or above on August 05, 2026.

  • 🟠 HIGH: 19

Exploit Status Upgrades

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

  • [UPGRADED] CVE-2026-15988 (suspended_starter/ai_engine) — F1: exploitable → functional, AAS: 9.6 → 11.6 (HIGH → HIGH). Originally in 2026-08-01 bulletin.
  • [UPGRADED] CVE-2026-67320 (axios/axios) — F1: theoretical → poc, AAS: 9.4 → 11.9 (HIGH → HIGH). Originally in 2026-08-01 bulletin.
  • [UPGRADED] CVE-2026-48449 (adobe/campaign) — F1: exploitable → functional, AAS: 10.2 → 12.8 (HIGH → CRITICAL). Originally in 2026-07-30 bulletin.

🟠 [HIGH] mervinpraison/praisonai

1 CVE | CVSS 3.1: 10.0 | AAS 10.8

  • cpe:2.3:a:mervinpraison:praisonai:*:*:*:*:*:*:*:* (< 4.6.40)

PraisonAI, the multi-agent teams framework by mervinpraison, is affected by one critical vulnerability. CVE-2026-48168 carries a CVSS score of 10.0 and involves a command injection flaw in the bundled Claude GitHub Actions workflow, where an attacker-controlled pull request branch name is embedded into a shell execution block without sanitization. The workflow also lacks proper authorization checks, allowing any user to trigger the job via an @claude comment regardless of collaborator status. This means an outside contributor can open a pull request from a fork with a malicious branch name and achieve arbitrary command execution in the CI environment.

Teams using PraisonAI with the bundled GitHub Actions workflow should treat this as an immediate priority. Versions prior to 4.6.40 are affected. Upgrade to version 4.6.40 or later, which addresses the injection and authorization issues. Review the vendor advisory and linked commit for details, and audit CI logs for any signs of prior exploitation.

Vendor Advisory

🟠 [HIGH] nuxt/devtools

1 CVE | CVSS 3.1: 9.6 | AAS 10.4

  • cpe:2.3:a:nuxt:devtools:*:*:*:*:*:*:*:* (< 3.3.1)

Nuxt DevTools, the development companion for the Nuxt web framework, is affected by one critical vulnerability. CVE-2026-71319 carries a CVSS score of 9.6 and stems from the DevTools RPC channel exposed over the Vite HMR WebSocket operating with no authentication whatsoever. Any client that can reach the Vite development server’s WebSocket endpoint can invoke privileged RPC methods such as updateOptions and clearOptions without any token, handshake, or origin verification, potentially allowing an attacker to manipulate the development environment or pivot further.

While this vulnerability is limited to development mode, developers running Nuxt DevTools on network-accessible interfaces or in shared environments should act promptly. Versions prior to 3.3.1 are affected. Upgrade to Nuxt DevTools 3.3.1 or later, which introduces authentication for the RPC channel, and ensure development servers are not exposed to untrusted networks.

Vendor Advisory

🟠 [HIGH] nuxt/nuxt

4 CVEs | CVSS 3.1: 8.2 | AAS 10.2

  • cpe:2.3:a:nuxt:nuxt:*:*:*:*:*:*:*:* (>= 3.21.7, < 3.21.10)
  • cpe:2.3:a:nuxt:nuxt:*:*:*:*:*:*:*:* (>= 4.0.0, < 4.5.1)
  • cpe:2.3:a:nuxt:nuxt:*:*:*:*:*:*:*:* (>= 3.4.0, < 3.21.10)
  • cpe:2.3:a:nuxt:nuxt:*:*:*:*:*:*:*:* (>= 3.1.0, < 3.21.10)
  • cpe:2.3:a:nuxt:nuxt:*:*:*:*:*:*:*:* (>= 4.4.0, < 4.5.1)

The Nuxt web framework for Vue.js is affected by four vulnerabilities, with the highest carrying a CVSS score of 8.2. The lead issue, CVE-2026-71315, reveals that mixed-case routeRules keys can fail to match case-folded lookups when case-sensitive routing is disabled, allowing attackers to bypass appMiddleware authorization gates entirely. This is an incomplete fix for a previously patched vulnerability, and the remaining CVEs including CVE-2026-71320, CVE-2026-71314, and CVE-2026-71316 represent additional exploitable issues in the framework.

Teams running Nuxt 3.x versions from 3.21.7 through 3.21.9 or Nuxt 4.x versions prior to 4.5.1 should prioritize patching, particularly if applications rely on route rules or middleware for access control. Upgrade to Nuxt 3.21.10 or 4.5.1 immediately, and review application route configurations to confirm authorization middleware is being applied as expected.

Vendor Advisory

🟠 [HIGH] jenkins_project/jenkins

1 CVE | CVSS 3.1: 9.0 | AAS 10.2

  • cpe:2.3:a:jenkins:jenkins:*:*:*:*:*:*:*:* (< 2.576)

Jenkins is affected by one critical vulnerability, CVE-2026-70426, carrying a CVSS score of 9.0. The flaw exists in the Remoting component where the JEP-200 deserialization class filter is not enforced on a fallback class resolution path, allowing agent processes or attackers with Agent/Connect permission to bypass deserialization protections for classes on the Jenkins core classpath. This effectively undermines a key security boundary between Jenkins agents and the controller, potentially enabling remote code execution on the controller from a compromised or malicious agent.

Organizations running Jenkins 2.575 and earlier or LTS 2.568.1 and earlier should treat this as a high-priority update. Upgrade to the latest Jenkins release as outlined in the vendor advisory at jenkins.io, and review which users and processes hold Agent/Connect permissions to limit exposure in the interim.

Vendor Advisory

🟠 [HIGH] milvus-io/milvus

1 CVE | CVSS 4.0: 8.7 | AAS 10.1

  • cpe:2.3:a:milvus-io:milvus:*:*:*:*:*:*:*:* (>= 3.0.0)

Milvus, the open-source vector database, is affected by one high-severity vulnerability. CVE-2026-69111 carries a CVSS score of 8.7 and allows unauthenticated remote attackers to shut down critical service components by sending a crafted HTTP GET request to the management server on port 9091. The /management/stop endpoint bypasses REST API authentication middleware entirely, enabling an attacker to terminate proxy, datanode, or querynode components by simply specifying a role parameter, resulting in a complete denial of service.

Organizations running Milvus through version 2.6.22 or version 3.0.0 should act immediately, especially if the management port is accessible from untrusted networks. As a mitigation, restrict network access to port 9091 to trusted hosts only, and monitor the vendor issue tracker for a patched release. Review the advisory linked at the Milvus GitHub repository for further updates.

Vendor Advisory

🟠 [HIGH] huggingface/peft

1 CVE | CVSS 3.1: 8.8 | AAS 10.1

  • cpe:2.3:a:huggingface:peft:*:*:*:*:*:*:*:* (>= 0.13.0)

Hugging Face PEFT, the popular parameter-efficient fine-tuning library, is affected by one high-severity vulnerability. CVE-2026-71281 carries a CVSS score of 8.8 and stems from unsafe use of torch.load() without weights_only=True in the LoRA-GA and CorDA initialization modules, enabling full pickle deserialization of attacker-controlled cache and covariance files. This bypasses PEFT’s own safe-loading wrapper used elsewhere in the codebase, meaning loading a malicious shared or downloaded LoRA-GA or CorDA cache file can result in arbitrary code execution.

Teams using PEFT for model fine-tuning, particularly those loading externally sourced cache or covariance files, should treat this as a high priority. Avoid loading untrusted LoRA-GA or CorDA cache files until a patched version is available, and monitor the Hugging Face PEFT repository for an updated release that enforces safe deserialization across all code paths.

Vendor Advisory

🟠 [HIGH] suse/rancher

1 CVE | CVSS 3.1: 9.1 | AAS 10.0

  • cpe:2.3:a:suse:rancher:*:*:*:*:*:*:*:*

SUSE Rancher is affected by one critical privilege escalation vulnerability. CVE-2026-44945 carries a CVSS score of 9.1 and exists in Rancher’s impersonation middleware, where any authenticated user holding the default user global role can escalate to full administrative access over the Rancher control plane and transitively over all downstream clusters it manages. This represents a complete compromise path from a low-privileged authenticated user to full cluster administration.

Organizations running Rancher versions 2.11.0 through 2.11.15, 2.12.0 through 2.12.11, 2.13.0 through 2.13.7, or 2.14.0 through 2.14.1 should treat this as an urgent priority. Upgrade immediately to Rancher 2.11.16, 2.12.12, 2.13.8, or 2.14.2 respectively, and audit access logs for any signs of unauthorized privilege escalation or unexpected administrative actions by standard user accounts.

Vendor Advisory

🟠 [HIGH] boringproxy/boringproxy

1 CVE | CVSS 4.0: 8.5 | AAS 9.9

  • cpe:2.3:a:boringproxy:boringproxy:*:*:*:*:*:*:*:*

boringproxy, the lightweight reverse proxy and tunnel manager, is affected by one high-severity vulnerability. CVE-2026-70615 carries a CVSS score of 8.5 and allows authenticated low-privileged users with tunnel-creation permission to inject arbitrary lines into the server’s SSH authorized_keys file by supplying a percent-encoded newline character in the domain parameter of the tunnel creation endpoint. An attacker can insert an unrestricted public key to gain persistent shell access to the server, and subsequently read cleartext credentials from the database, escalating from limited tunnel permissions to full system compromise.

Organizations running boringproxy through version 0.10.0 should treat this as a high priority. No patched release is currently available, so administrators should restrict tunnel-creation permissions to trusted users only, audit the server’s authorized_keys file for unexpected entries, and monitor the project repository for an updated release. Review the vendor advisory for full technical details and indicators of compromise.

Vendor Advisory

🟠 [HIGH] louislam/uptime-kuma

1 CVE | CVSS 3.1: 8.1 | AAS 9.9

  • cpe:2.3:a:louislam:uptime-kuma:*:*:*:*:*:*:*:* (>= 2.0.0-beta.0, < 2.4.2)

Uptime Kuma, the popular open-source monitoring tool, is affected by one high-severity vulnerability. CVE-2026-71285 carries a CVSS score of 8.1 and exists in the Matomo analytics integration, where the admin-configurable siteId value is injected as an unquoted JavaScript expression into a script block rendered on every public status page. The escaping pipeline fails to neutralize characters such as brackets, parentheses, and semicolons, allowing an attacker with admin access to break out of the JavaScript context and execute arbitrary scripts in the browsers of all status page visitors.

Organizations running Uptime Kuma with Matomo analytics enabled on public status pages should take immediate action. Disable the Matomo analytics integration as an interim mitigation, audit the configured siteId value for any injected content, and monitor the Uptime Kuma repository for a patched release that properly sanitizes the siteId parameter before script injection.

Vendor Advisory

🟠 [HIGH] usememos/memos

1 CVE | CVSS 3.1: 8.5 | AAS 9.8

  • cpe:2.3:a:usememos:memos:*:*:*:*:*:*:*:* (< 0.24.0)

Memos, the open-source note-taking and knowledge management application, is affected by one high-severity vulnerability. CVE-2026-71272 carries a CVSS score of 8.5 and is a DNS rebinding flaw in the webhook dispatch function, where the hostname is validated against reserved IP ranges during an initial DNS lookup but then dialed using the original hostname rather than the validated IP address. Because the dialer performs its own independent DNS resolution, an attacker controlling DNS for a webhook target can use short TTL records to pass validation with a public IP and then rebind to an internal address, enabling server-side request forgery against internal services.

Organizations self-hosting Memos with webhook functionality enabled should treat this as a priority issue. As an interim mitigation, restrict outbound network access from the Memos server to internal networks using firewall rules, and limit webhook URLs to trusted destinations. Monitor the Memos GitHub repository for a patched release that addresses the TOCTOU gap in DNS validation.

Vendor Advisory

🟠 [HIGH] openssl/openssl

1 CVE | CVSS 3.1: 7.5 | AAS 9.8

  • cpe:2.3:a:openssl:openssl:*:*:*:*:*:*:*:* (>= 3.3.0, < 3.3.3)
  • cpe:2.3:a:openssl:openssl:*:*:*:*:*:*:*:* (>= 3.4.0, < 3.4.1)

OpenSSL is affected by one high-severity vulnerability. CVE-2026-54876 carries a CVSS score of 7.5 and is a memory leak triggered when a TLS client with OCSP response checking enabled connects to a malicious server that sends an OCSP response containing no single response entries. Each such handshake leaks an attacker-tunable amount of memory, and long-running client applications that repeatedly connect to a malicious server can have their memory exhausted, resulting in denial of service.

Any organization deploying applications that use OpenSSL as a TLS client with OCSP stapling validation enabled should prioritize this update, particularly for long-running services such as proxies, API gateways, and monitoring agents. Apply the patched OpenSSL release as soon as it is available from your vendor or upstream, and review the linked commit for details on affected versions.

Vendor Advisory

🟠 [HIGH] github/enterprise_server

1 CVE | CVSS 4.0: 8.8 | AAS 9.7

  • cpe:2.3:a:github:enterprise_server:*:*:*:*:*:*:*:* (>= 3.13.0, < 3.13.17)
  • cpe:2.3:a:github:enterprise_server:*:*:*:*:*:*:*:* (>= 3.14.0, < 3.14.14)
  • cpe:2.3:a:github:enterprise_server:*:*:*:*:*:*:*:* (>= 3.15.0, < 3.15.9)
  • cpe:2.3:a:github:enterprise_server:*:*:*:*:*:*:*:* (>= 3.16.0, < 3.16.5)
  • cpe:2.3:a:github:enterprise_server:*:*:*:*:*:*:*:* (>= 3.17.0, < 3.17.2)

GitHub Enterprise Server is affected by one critical vulnerability. CVE-2026-17556 carries a CVSS score of 8.8 and is a path traversal flaw where the X-GitHub-Request-Id header is used without sanitization as a filesystem path segment for the upload buffer directory. An unauthenticated attacker can supply a traversal value to point the buffer at an arbitrary path, causing the deferred cleanup process to delete arbitrary files and directories on the instance, including the entire user storage directory containing Git LFS objects, release assets, attachments, and avatars.

Organizations running GitHub Enterprise Server should treat this as an urgent priority given that exploitation requires no authentication and can result in catastrophic data loss. Upgrade to version 3.17.19 or later immediately, and review server logs for any suspicious X-GitHub-Request-Id header values that contain path traversal sequences as indicators of attempted exploitation.

Vendor Advisory

🟠 [HIGH] cacti/cacti

1 CVE | CVSS 3.1: 8.8 | AAS 9.6

  • cpe:2.3:a:cacti:cacti:*:*:*:*:*:*:*:* (< 1.2.30)

Cacti, the widely used network monitoring and graphing platform, is affected by one high-severity vulnerability. CVE-2026-71287 carries a CVSS score of 8.8 and is a SQL injection flaw in the sanitize_sql_column() function, where the regex allowlist intended to sanitize ORDER BY column names retains parentheses and other characters that permit SQL function calls such as SLEEP(5) to pass through unmodified. The sanitized value is then concatenated directly into raw SQL ORDER BY clauses, allowing authenticated attackers to inject arbitrary SQL statements through user-controlled sort parameters.

Organizations running Cacti should treat this as a high priority, as the flaw enables data exfiltration, database manipulation, and potentially further system compromise from any authenticated user session. Monitor the Cacti GitHub repository for a patched release, and as an interim measure consider restricting Cacti access to trusted administrators and deploying web application firewall rules to detect SQL injection patterns in sort-related request parameters.

Vendor Advisory

🟠 [HIGH] opensips/opensips

2 CVEs | CVSS 4.0: 8.7 | AAS 9.6

  • cpe:2.3:a:opensips:opensips:*:*:*:*:*:*:*:*

OpenSIPS, the open-source SIP server, is affected by two high-severity vulnerabilities. The most critical, CVE-2026-45809, carries a CVSS score of 8.7 and is a stack buffer overflow in the watcherinfo XML generation functionality, where an attacker can send a SUBSCRIBE request with an oversized From URI and then trigger watcherinfo generation to overflow a fixed-size buffer and crash the server. CVE-2026-46334 represents an additional exploitable denial of service issue in the same product.

Organizations running OpenSIPS versions prior to 3.6.6 or 4.0.0-rc1 in VoIP and telecommunications environments should prioritize patching, as these flaws can be triggered remotely to take down SIP infrastructure. Upgrade to OpenSIPS 3.6.6 or 4.0.0-rc1 immediately, and review SIP traffic logs for abnormally long From URI values in SUBSCRIBE requests as potential indicators of exploitation attempts.

Vendor Advisory

🟠 [HIGH] go-shiori/shiori

1 CVE | CVSS 3.1: 8.5 | AAS 9.3

  • cpe:2.3:a:go-shiori:shiori:*:*:*:*:*:*:*:*

Shiori, the open-source bookmark manager by go-shiori, is affected by one high-severity vulnerability. CVE-2026-71280 carries a CVSS score of 8.5 and is a server-side request forgery flaw in the DownloadBookmark() function, which fetches user-supplied bookmark URLs using a plain HTTP client with no destination IP validation. An authenticated user can supply loopback, private, or link-local addresses through bookmark creation or update endpoints, causing the server to make requests to internal services and potentially exposing sensitive data or enabling further attacks against the internal network.

Organizations self-hosting Shiori should treat this as a priority, particularly in environments where the server has access to internal services, cloud metadata endpoints, or other sensitive network resources. Restrict outbound network access from the Shiori server using firewall rules as an interim mitigation, limit account creation to trusted users, and monitor the go-shiori GitHub repository for a patched release that implements proper URL destination validation.

Vendor Advisory