CVE Database

SGLang contains an SSRF and local file read in the multimodal generation endpoint /v1/chat/completions due to unsanitized image_url, allowing access to internal metadata, secrets, and services.

Server-Side Request Forgery vulnerability in malach-it Boruta allows an unauthenticated remote attacker to cause the OAuth/OpenID authorization server to issue outbound HTTP requests to attacker-chosen URIs, including internal services and cloud metadata endpoints. Three code paths fetch remote URIs supplied by the requester without sufficient validation of the target. Boruta.Oauth.Request.Base.fetch_unsigned_request/1 in lib/boruta/oauth/request/base.ex dereferences the OAuth request_uri parameter from the authorization request via Finch.build(:get, request_uri) |> Finch.request(OpenIDHttpClient). Boruta.Openid.parse_registration_params/2 in lib/boruta/openid.ex dereferences the jwks_uri supplied in an OpenID Connect dynamic client registration request. Boruta.Ecto.Clients.refresh_jwk_from_jwks_uri/1 in lib/boruta/adapters/ecto/clients.ex later refreshes the stored jwks_uri for an existing client. In all three paths the only validation is that the URI parses with a scheme (and one of the two request_uri clauses does not even restrict the scheme to http or https). The implementations do not require HTTPS, do not enforce a host or IP allowlist, do not reject loopback, private, link-local, or other non-public ranges after DNS resolution, do not cap response size, and do not constrain redirects. An attacker can therefore steer the server's HTTP client at arbitrary network targets reachable from the Boruta host. This issue affects boruta: from 2.3.2 before 2.3.7.

A flaw was found in koku-metrics-operator. The operator's CostManagementMetricsConfig custom resource allows a user able to edit the CR to specify an arbitrary OAuth token endpoint. When authentication.type is set to service-account, the operator sends the tenant's Red Hat SSO client_id and client_secret to this user-controlled URL, allowing the attacker to obtain the credentials.

A flaw was found in koku-metrics-operator. The operator's CostManagementMetricsConfig custom resource allows user able to edit the CR to specify an arbitrary upload URL. When authentication.type is set to token (the default), the cluster-global Red Hat Cloud pull-secret bearer token is attached to HTTP requests sent to this user-controlled URL, allowing the attacker to obtain the token.

A flaw was found in Dogtag PKI's ACME responder where the HTTP-01 challenge validator accepts IP address literals as dns identifiers and follows HTTP redirects without validating that the target is a public address. An unauthenticated ACME account holder can exploit this to perform server-side request forgery (SSRF), making the Dogtag server send HTTP GET requests to internal network services. With the InMemory database backend, the response body of internal targets is disclosed to the attacker through the ACME challenge error.

Pydantic AI is a Python agent framework for building Generative AI applications. In versions 1.65.0 through 1.105.0, and 2.0.0b1 through 2.0.0b5, a client that submits message history to a Pydantic AI UI adapter (such as the Vercel AI adapter) can reference arbitrary files in the application's model-provider or cloud-storage account. While file URL parts are validated against a scheme allowlist, UploadedFile references — which point to a file by provider file ID or cloud-storage URI (e.g. s3://…, gs://…) — were forwarded without validation. Because the provider resolves an UploadedFile using the server-side identity (IAM role, service account, or provider API key) rather than the client's, an attacker can craft message history to make the server read objects from its own account or other tenants, given a referenceable identifier. Exploitation requires a valid file identifier, which is not always unguessable depending on how the application names objects. This issue has been fixed in versions 1.106.0 and 2.0.0b6.

Pydantic AI is a Python agent framework for building Generative AI applications. In versions 1.56.0 through 1.98.0, when an application opts a URL into force_download='allow-local' (disabling the default block on private/internal IPs), the cloud-metadata blocklist could be bypassed by encoding the metadata IP in an IPv6 transition form (IPv4-mapped IPv6, 6to4, or NAT64), exposing cloud IAM short-term credentials on dual-stack or translated networks. This is an incomplete fix of GHSA-2jrp-274c-jhv3 / CVE-2026-25580, whose remediation did not hold for IPv6-encoded forms of the metadata IPs. An application is affected only if it explicitly opts a FileUrl (ImageUrl, AudioUrl, VideoUrl, DocumentUrl) into force_download='allow-local' on a URL influenced by untrusted input; it is not affected when using bundled integrations to ingest user input (Agent.to_web / clai web, VercelAIAdapter, AGUIAdapter / Agent.to_ag_ui), since they do not propagate force_download from external data, nor when downloading only from developer-controlled URLs. This issue has been fixed in version 1.99.0.

The WP CTA plugin for WordPress is vulnerable to Server-Side Request Forgery via the 'sticky_s_media' parameter in imported JSON files in all versions up to, and including, 2.1.2. This is due to the import_sidebars() function passing user-supplied URLs from imported JSON data to file_get_contents() with only FILTER_VALIDATE_URL validation (which allows internal IPs). This makes it possible for authenticated attackers, with Administrator-level access and above, to make web requests to arbitrary locations originating from the web application, which can be used to query and modify information from internal services. The response content is saved as a WordPress media attachment, making this a full-read SSRF.

The Media Cleaner: Clean your WordPress! plugin for WordPress is vulnerable to Server-Side Request Forgery in all versions up to, and including, 7.0.3. This is due to the `get_urls_from_html()` function using `DOMDocument::loadHTMLFile()` to fetch iframe source URLs with an insufficient hostname validation check that relies on a substring match against the site's server name. This makes it possible for authenticated attackers, with Administrator-level access and above, to make web requests to arbitrary locations originating from the web application, which can be used to query and interact with internal services.

Pivotick did not validate the URL scheme of node imagePath values derived from graph data before assigning them to SVG image resources. An attacker able to supply crafted graph data could set an image path to a malicious URI. When a victim rendered the affected graph, the browser could resolve the attacker-controlled URI and initiate an unintended request or invoke scheme-specific handling in the victim’s context. Depending on the URI, browser behaviour, and installed protocol handlers, exploitation could disclose limited client or network metadata, facilitate rendering-based tracking, or attempt to access local or internal resources. Exploitation requires a victim to load or render graph data containing the malicious imagePath. The patch normalizes ASCII whitespace and control characters in URI schemes and restricts image paths to relative URLs or the http, https, data, and blob schemes.