AutoGPT is a workflow automation platform for creating, deploying, and managing continuous artificial intelligence agents. Prior to 0.6.52, an authenticated user can bypass the SSRF / private-IP protections in SendWebRequestBlock and reach internal network services. _is_ip_blocked() in backend/backend/util/request.py does not normalize IPv4-mapped IPv6 addresses before checking resolved IPs against the blocked IPv4 ranges, and does not block special-use ranges such as 100.64.0.0/10 (CGNAT, RFC 6598). A hostname that resolves to an IPv4-mapped IPv6 address therefore passes validation and the request reaches the embedded internal IPv4 endpoint. This affects all AutoGPT Platform deployments. This vulnerability is fixed in 0.6.52.
HTMLy 3.1.1 contains a Server-Side Request Forgery (SSRF) vulnerability in the RSS feed import functionality. The function get_feed() in system/admin/admin.php passes user-supplied $feed_url directly to file_get_contents() without any validation. An authenticated attacker with administrative privileges can exploit this by entering a crafted URL (e.g., http://dnslog.example.com, file:///etc/passwd, or http://169.254.169.254 in cloud contexts) via Tools -> Import RSS. The server will then make a request to the attacker-controlled target.
The WSO2 API Manager's message flow component, when processing WS-Addressing headers, does not sufficiently validate or restrict user-controlled input within these headers. This omission allows an attacker to manipulate WS-Addressing headers to specify arbitrary destinations for server-initiated requests.
Successful exploitation allows an unauthenticated attacker to control the destination of server-initiated requests originating from the WSO2 API Manager. This direct control can enable unauthorized access to internal network resources or services that would typically be inaccessible from external networks.
Server-Side Request Forgery in the markdown_to_pdf action of Rapid7 InsightConnect Markdown Plugin on Linux in versions prior to 4.0.2 allows remote attackers to make arbitrary outbound HTTP requests via unsanitized resource-loading HTML elements (img/src, CSS url(), @import) embedded in Markdown input. The initial fix in 4.0.0 disabled JavaScript but did not neutralize resource-loading vectors. Resolved in 4.0.2 by sanitizing HTML with an allowlist of tags, attributes, and URL schemes.
A flaw was found in Apicurio Registry. The WSDLReaderAccessor creates a wsdl4j WSDLReader without disabling the javax.wsdl.importDocuments feature. When the VALIDITY rule is set to FULL, an attacker with Developer-role access can upload a WSDL document containing attacker-controlled import locations, causing the registry to issue HTTP requests to arbitrary internal URLs (server-side request forgery).
Two data sources (DICOMWebProxy and DICOMJSON) shipped in the default configuration fetch an arbitrary URL parameter without validation. A global authentication service in OHIF automatically injects the authenticated user's OIDC Bearer token into the resulting requests, sending it to the attacker-controlled server. DICOMweb data sources are not impacted.
Red Hat build of Keycloak is an integrated sign-on solution,
available as a Red Hat JBoss Middleware for OpenShift containerized
image. The Red Hat build of Keycloak for OpenShift image provides
an authentication server that you can use to log in centrally, log
out, and register. You can also manage user accounts for web
applications, mobile applications, and RESTful web services.
Red Hat build of Keycloak Operator for OpenShift simplifies
deployment and management of Keycloak 26.4.13 clusters.
This erratum releases new images for Red Hat build of Keycloak
26.4.13 for use within the OpenShift Container Platform cloud
computing Platform-as-a-Service (PaaS) for on-premise or private
cloud deployments, aligning with the standalone product release.
Security fixes:
* Information disclosure via CORS header injection due to unvalidated JWT azp claim (CVE-2026-37977)
* Server-Side Request Forgery via OIDC token endpoint manipulation (CVE-2026-4874)
* eclipse-vertx/vert.x:Denial of Service via TLS handshake with wildcard server name (CVE-2026-6860)
* Improper Access Control on Keycloak Server when the account Account API feature is disabled (CVE-2026-7500)
* Policy bypass during WebAuthn credential registration via client-side JavaScript manipulation (CVE-2026-8830)
* Security flaw in org.keycloak/keycloak-services (CVE-2026-8922)
* Information disclosure through arbitrary filesystem path probing (CVE-2026-9083)
* Cross-site scripting (XSS) via case-insensitive URI validation bypass (CVE-2026-9086)
* Cross-Session Email Verification Proof Not Bound to Upstream Identity in First-Broker-Login (CVE-2026-9087)
* Information disclosure due to user profile permission bypass (CVE-2026-9088)
* Group-Admin Escalation to Realm-Admin (CVE-2026-9099)
* Privilege escalation due to oversized subject_token JWT (CVE-2026-9704)
* Attacker can re-enable and take over disabled clients via Registration Access Token (CVE-2026-9705)
* Organization Data Leak After Feature Disabled in Keycloak (CVE-2026-9791)
* Security restriction bypass allows unauthorized ROPC token acquisition (CVE-2026-9792)
* Information disclosure via SAML ECP endpoint (CVE-2026-9794)
* Privilege escalation via improper scope mapping enforcement (CVE-2026-9795)
* Unauthorized access to resources via UMA permission ticket bypass (CVE-2026-9799)
* Authorization bypass via incorrect URI comparison (CVE-2026-9800)
* Denial of Service via malformed LDAP password policy response (CVE-2026-9801)
* Unauthorized account access via replayed refresh tokens after cluster restart (CVE-2026-9802)
* Denial of Service via malformed Authorization header (CVE-2026-9803)
LibreChat is an enhanced ChatGPT clone that supports multiple AI providers. Prior to 0.8.4-rc1, LibreChat allows users to configure custom OpenAI-compatible API endpoints by setting a baseURL. This URL is used to construct HTTP requests without any SSRF validation — no private IP check, no scheme restriction, no DNS pinning. An authenticated user can set baseURL to internal network addresses. This vulnerability is fixed in 0.8.4-rc1.
Appsmith is a platform to build admin panels, internal tools, and dashboards. Prior to 1.99, the POST /api/v1/admin/send-test-email endpoint accepts attacker-controlled smtpHost and smtpPort values and establishes a raw JavaMail TCP connection without any IP validation. This completely bypasses WebClientUtils.IP_CHECK_FILTER, which only applies to Spring WebClient HTTP requests. Additionally, the raw MailException.getMessage() is returned verbatim in the API error response, enabling error-based internal port scanning and service banner enumeration. This vulnerability is fixed in 1.99.
Appsmith is a platform to build admin panels, internal tools, and dashboards. Prior to 2.1, the outbound HTTP host filter applied by WebClientUtils (used by the REST API and GraphQL datasource plugins) validates hosts against an exact-match string denylist. The comprehensive address-class check (loopback, any-local, link-local, fc00::/7) exists only on a separate code path used by SMTP, not by the HTTP plugin path. As a result, an authenticated user can craft outbound requests that reach loopback-bound services inside the container. This vulnerability is fixed in 2.1.
Appsmith is a platform to build admin panels, internal tools, and dashboards. Prior to 2.1, Appsmith's bundled supervisord exposes an XML-RPC interface on port 9001, reachable from outside the container via a Caddy reverse-proxy route at /supervisor/* on the public ingress. Combined with the APPSMITH_SUPERVISOR_PASSWORD exposed via GET /api/v1/admin/env, any authenticated administrator can send arbitrary XML-RPC calls to supervisord and execute OS commands inside the Docker container via twiddler.addProgramToGroup. This vulnerability is fixed in 2.1.
Gogs, an open source self-hosted Git service, has a Server-Side Request Forgery (SSRF) vulnerability in its repository migration feature prior to version 0.14.3. The vulnerability arises because the application only validates the initial URL hostname, but the git clone --mirror command follows HTTP redirects. An authenticated user can exploit this by submitting a public URL that redirects to an internal blocked endpoint, such as 127.0.0.1, potentially importing internal repository contents into an attacker-controlled repository. This issue is fixed in version 0.14.3.
Ghost is a Node.js content management system. From 6.0.9 until 6.21.1, when making an external request, it is possible to bypass the IP filter that ensures the request isn't going to an internal service using an IPv6 literal which maps to a private IPv4 address. This vulnerability is fixed in 6.21.1.
Ghost is a Node.js content management system. From 6.0.9 until 6.21.1, Ghost’s private-IP check for outbound HTTP requests could be bypassed via DNS rebinding, allowing an attacker to coerce the Ghost server into reaching hosts on internal networks through features that issue external fetches. This vulnerability is fixed in 6.21.1.
Ghost is a Node.js content management system. From 6.19.4 until 6.21.1, when re-rendering posts, Ghost would refetch missing image dimensions by issuing an outbound HTTP request to the URL stored on an image card — without restricting that URL to trusted image hosts. An authenticated staff user able to create or edit posts could therefore point an image card at an attacker-chosen host and cause the Ghost server to request it on their behalf, including hosts on internal networks or cloud instance metadata endpoints that would not normally be reachable from the public internet. This vulnerability is fixed in 6.21.1.
Docling simplifies document processing by parsing diverse formats and providing integrations with the generative AI ecosystem. FIn versions >= 2.82.0, < 2.91.0, if the HTML backend was explicitly configured for rendering (rendering option by default deactivated), then the Playwright-based rendering feature could allow JavaScript execution and unrestricted network access when processing untrusted HTML documents. An attacker could craft malicious HTML that executes arbitrary JavaScript in the rendering context or makes unauthorized network requests to internal services, potentially leading to SSRF attacks, data exfiltration, or remote code execution in the rendering environment. This vulnerability is fixed in 2.91.0.
A critical vulnerability in Admin GUI in Payara Server Full 4.x, 5.x, 6.x, 7.x, 7.2026.x, 6.2025.x, 6.2024.x on All platforms that allows the attacker to leak the admin gfresttoken to an attacker-controlled host that can result in a full unauthenticated takeover of Payara admin domain.
A Server-Side Request Forgery (SSRF) vulnerability in the DownloadServlet of the Admin GUI in Payara Server allows a remote attacker to exfiltrate the administrator's REST session token (gfresttoken) to an attacker-controlled host via a crafted request URL. Combined with the absence of CSRF protection on DownloadServlet, an unauthenticated attacker can trick a logged-in administrator into triggering the token leak, then replay the stolen token to gain full administrative access to the Payara domain, leading to arbitrary code execution via WAR deployment. The vulnerability exists in the DownloadServlet and associated ContentSource implementations (LogViewerContentSource, LogFilesContentSource, LBConfigContentSource, ClientStubsContentSource) within the admingui:console-common module.
The Kargo Takip plugin for WordPress is vulnerable to Server-Side Request Forgery in all versions up to, and including, 1.2 via the 'api_url' parameter. This makes it possible for unauthenticated attackers to make web requests to arbitrary locations originating from the web application and can be used to query and modify information from internal services. The script echoes internal API response data (specifically the value of any 'auth' key in a JSON response body) verbatim back to the attacker's browser, enabling direct exfiltration of responses from internal services such as cloud instance metadata endpoints.