A vulnerability in the team folders (formerly group folders) app when used in combination with the workspace app allowed API/REST-only delegated administrators to bypass folder-level authorization controls. The workspace app enables organizations to delegate limited administrative privileges for team folder management via API/REST only, restricting access to folders for which the admin has advanced permissions.
Lara Dashboard versions through 1.3.1 have a server-side request forgery (SSRF) vulnerability in the POST /api/admin/builder/markdown/fetch endpoint. This flaw allows any authenticated user to fetch arbitrary URLs and read the response body without host validation or redirect restrictions. Exploitation can lead to reading internal HTTP services and cloud metadata, including IAM credentials.
Wallos is an open-source, self-hostable personal subscription tracker. Prior to version 4.9.6, Admin-configured OIDC token_url and user_info_url in includes/oidc/handle_oidc_callback.php:18-49 are used directly in curl_init() with zero SSRF filtering. Unlike logo/webhook URLs which have validate_webhook_url_for_ssrf(), OIDC URLs bypass all protections. Admin sets URL to http://169.254.169.254/latest/meta-data/ for cloud metadata access or internal network pivoting. This issue has been patched in version 4.9.6.
MCPHub is a unified hub for centrally managing and dynamically orchestrating multiple MCP servers/APIs into separate endpoints with flexible routing strategies. Prior to version 1.0.30, MCPHub scopes non-admin users to servers they own (list views and config edits enforce ownership), but the tool-execution API does not. Any authenticated non-admin user can invoke tools on MCP servers owned by other users — servers they cannot even see in GET /api/servers. Because connected MCP servers carry real capability (filesystem, HTTP fetch, cloud APIs with the owner's keys), this is cross-tenant compromise: demonstrated arbitrary host file read (/etc/passwd, another user's secrets) and SSRF. This issue has been patched in version 1.0.30.
Ech0 versions before 4.4.3 have a server-side request forgery (SSRF) vulnerability in the validateWebhookURL function. The function improperly validates webhook URLs by only checking literal IP addresses and failing to reject hostnames that resolve to private or internal IP addresses. An attacker with admin privileges can exploit this to create webhooks that cause the server to make unauthorized requests to internal services or cloud metadata endpoints. This vulnerability is fixed in version 4.4.3.
CVE-2026-59335 is a high-severity vulnerability in Cloud Foundry UAA affecting deployments backed by MySQL with default collation. It involves improper handling of case sensitivity in the identity zone authorization check, allowing a remote authenticated attacker with zones.write authority to bypass restrictions by using non-lowercase zone identifiers. This leads to unauthorized access to the privileged system identity zone, enabling the attacker to overwrite the system zone's JWT signing key and forge tokens with admin privileges, fully compromising UAA and dependent Cloud Foundry deployments. PostgreSQL and HSQLDB backends are not affected.
CVE-2026-55976 is a critical Server-Side Request Forgery (SSRF) vulnerability in Apache Hive before version 4.2.1. It allows an authenticated user with CREATE TABLE privileges to cause the Hive server to fetch attacker-controlled URLs during Avro schema resolution. This can lead to exposure of cloud instance metadata, internal network services, or local server files accessible by the Hive process identity. The vulnerability requires network access to HiveServer2 or Metastore and valid Hive authentication but does not require admin privileges. Users are advised to upgrade to Apache Hive 4.2.1, which addresses this issue.
CVE-2026-61400 is a command injection vulnerability in Apache CloudStack affecting the run and get diagnostics functionality for system VMs and virtual routers. Authenticated users with admin permissions on the getDiagnosticsData or runDiagnostics APIs can execute arbitrary commands as root or diagnostics-process user. This allows full compromise of affected instances and potential lateral movement within the CloudStack infrastructure. The vulnerability affects Apache CloudStack versions from 4.20.0.0 through 4.20.3.0 and 4.21.0.0 through 4.22.1.0. Upgrading to versions 4.20.3.1 or 4.22.1.1 or later mitigates the issue.
CVE-2026-66722 is an improper authorization vulnerability in Apache CloudStack affecting versions 4.15.0.0 through 4.20.3.0 and 4.21.0.0 through 4.22.1.0. It allows a Domain Admin to perform create, update, delete, and list operations on project roles and permissions across any domain, not limited to their own. This occurs because the authorization check only verifies the caller's Domain Admin status without confirming domain ownership of the target project. The issue is fixed in versions 4.20.3.1 and 4.22.1.1 or later.
CVE-2026-76251 is a vulnerability in Splunk Enterprise versions 10.0 up to but not including 10.0.9, 10.2 up to but not including 10.2.6, and 10.4 up to but not including 10.4.2. It allows users without admin or power roles to cause the Splunk App for Splunk Observability Cloud to forward requests with the stored access token, potentially exposing data and allowing limited changes in Splunk Observability Cloud. The issue arises because the app's REST API does not enforce proper authorization checks before forwarding requests. Versions 9.3 and 9.4 are not affected.
ContextForge is an AI gateway, registry, and proxy that provides centralized discovery, guardrails, and management for MCP, A2A, and REST or gRPC APIs. Prior to 1.0.3, the /admin/gateways/test call site in mcpgateway/admin.py calls validate_gateway_test_url() in mcpgateway/common/validators.py to resolve and reject private, loopback, link-local, and cloud-metadata addresses, but ResilientHttpClient later resolves the original hostname again without binding the validated address. When MCPGATEWAY_ADMIN_API_ENABLED is enabled, an attacker with a database-backed role containing explicit gateways.read permission can use DNS rebinding to return a public address during validation and a private or metadata address during connection, bypassing ssrf_blocked_networks and ssrf_dns_fail_closed because those controls apply only to the validation-time result. The endpoint's allow_admin_bypass=False setting means a bootstrap-only virtual platform-admin identity without a database role is not sufficient. Successful exploitation can reach internal services and cloud metadata, expose cloud credentials, access internal APIs, or probe internal network ports. This issue is fixed in version 1.0.3.
SeaweedFS versions prior to 4.24 contain an unauthenticated Server-Side Request Forgery (SSRF) vulnerability in the VolumeServer.FetchAndWriteNeedle RPC. This RPC fetches a caller-supplied remote endpoint and writes the response into a needle without authentication or target validation, allowing attackers with network access to the volume server's gRPC port to coerce requests to arbitrary internal or external hosts and read the responses. This includes sensitive cloud metadata endpoints that can disclose instance metadata and IAM credentials. The vulnerability is fixed in version 4.24 by requiring admin authorization and blocking requests to loopback, link-local, RFC 1918, and cloud metadata IPs. Workarounds include restricting access to the gRPC port and enabling mTLS.
Red Hat has released security updates for the Red Hat build of Keycloak 26.6.5, addressing multiple vulnerabilities including authorization bypasses, privilege escalation, denial of service, arbitrary code execution, and security bypasses in various components such as the admin UI, FGAPv2, DCR protocol, LDAP user search, Quarkus REST, SAML broker, and OIDC redirect URI. These vulnerabilities affect the OpenShift containerized image and operator for on-premise or private cloud deployments. The update aligns with the standalone Keycloak product release and provides new container images to mitigate these issues.
Odysseus before commit 87babb5 contains a server-side request forgery vulnerability that allows admin-privileged attackers to direct the server to probe internal network resources by supplying arbitrary URLs to the embedding endpoint configuration without scheme, host, IP range, or DNS rebind validation. Attackers can submit loopback addresses, RFC 1918 ranges, or link-local addresses through the embedding endpoint API to partially read responses from cloud instance metadata services, internal APIs, and other hosts reachable from the server.
The news coverage surrounding the breach of the AI platform Hugging Face by OpenAI’s autonomous agents reads like a scene straight out of a Terminator movie (apparently, a prequel). Yet for enterprise security teams — even those that don’t develop or deploy artificial intelligence — the detailed incident report published by Hugging Face provides a critical operational roadmap. The incident needs to be evaluated on the basis of a single, fundamental question: “Which organizational security flaws, missing controls, configurations, or technologies allowed this attack to succeed?” What happened? During cyber-capability testing of OpenAI’s advanced AI models against the ExploitGym benchmark, an agent found a way to break out of its confinement, reached the internet, and breached the infrastructure of several companies — one of them being Hugging Face. The model reasoned that it could find solutions to the benchmark challenges on these external systems. Dozens of articles have since covered these incidents in detail , so no need to go over that here. What truly matters is what transpired within Hugging Face’s systems between July 9 and 13 while the AI agent operated unchecked inside. The Hugging Face incident report includes an interactive attack timeline, so we’ll just list the key stages. A malicious configuration in a dataset uploaded by the agent allowed it to exfiltrate credentials for a worker pod , and execute code within it. The agent then exploited cloud environment metadata and escaped the privileged pod to obtain root-level access on the host. It retrieved a substantial number of secrets from storage and used a stolen VPN key and shared cluster administrator credentials to pivot deeper into the internal network and source code repository before the Hugging Face security team blocked its access. How significant is the fact that this was next-generation AI? Regardless of the specific area of activity of a company, tomorrow it may become a target of human attackers, fully autonomous systems, or AI agents manually controlled by human hackers. Many AI practitioners have rushed to dismiss this incident based on assumptions that OpenAI will never allow its tech to get into serious hacking. However, security researchers have argued for months that even currently available open-source models can be used to launch both offensive research and real-world attacks. Human oversight, while still a requirement, can be minimal. Even mid-tier models can easily iterate through a dozen vulnerabilities, inventory a compromised infrastructure, and quickly figure out how to expand their reach across a network. Defenders must analyze incidents like this in detail, map the key findings to their own environments, and adapt security controls to counter both stealthy human actors and rapid, noisy AI agents. Three distinguishing features of AI-powered attacks The techniques detailed in the Hugging Face report are not novel — the agent didn’t invent anything new. However, three factors regarding the attack’s speed and scale fundamentally shift the economics for both attackers and defenders: Trial and error becomes significantly cheaper. Where a human hacker might test five attack vectors, an agent can iterate through five hundred. Scenarios that security teams previously dismissed as “theoretically possible, but impractical and unlikely” and placed them on the backlog can now become active threats. Attack execution and incident response windows shrink. High-speed attacks are not unheard of, while some ransomware operators achieve compromise within hours even without AI. However, this accelerated pace will likely become the new baseline in cybersecurity, as demonstrated by the agent obtaining admin privileges at Hugging Face within 13 hours. Numerous alerts and log entries generated by the agent’s trial-and-error attempts can both help and hinder attackers as well as defenders. This noise can serve both as an indicator for threat detection, and as cam…
A VCO feature does not sufficiently validate caller-supplied input, allowing requests to be made on behalf of authenticated tenant accounts to internal services that are not otherwise accessible. This vulnerability requires a minimum role of Enterprise Standard Admin.
This issue was discovered internally by Arista and the company is not aware of any malicious uses of this issue in customer networks.