🚨 CVE-2026-53585
libgit2 is a portable C implementation of the Git core methods provided as a linkable library with a solid API, allowing to build Git functionality into your application. Prior to 1.8.6 and 1.9.5, git_delta_apply in src/libgit2/delta.c trusts the attacker-controlled res_sz value parsed by hdr_sz from a delta object header and passes that amount to git__malloc before validating delta instructions. Malicious pack data supplied through git_clone, git_fetch, git_remote_fetch, git_indexer_append, or a local attacker-supplied repository can use a very small multi-level OFS_DELTA chain to retain extremely large allocations and exhaust memory. This issue is fixed in versions 1.8.6 and 1.9.5.

🎖@cveNotify

🚨 CVE-2026-61663
django CMS is an easy-to-use and developer-friendly enterprise content management system powered by Django. Prior to 5.0.9, render_object_structure fails to authorize non-PageContent objects that use PlaceholderRelationField. An active staff user without cms.use_structure or model-level view or change permission can request the //admin/cms/placeholder/object/<content_type_id>/structure/<object_id>/ endpoint with guessed content_type_id and object_id values. The response discloses placeholder slot names, plugin trees, plugin identifiers, labels, and object existence for frontend-editable objects. The fix applies user_can_view_placeholder_source to the non-PageContent branch while keeping the structure board read-only for view-only users. This issue is fixed in versions 5.0.9.

🎖@cveNotify

🚨 CVE-2026-66001
Frappe is a full-stack web application framework. Prior to 15.114.0 and 16.26.0, the approve and authorize functions in frappe/integrations/oauth2.py allow the OAuth2 consent flow to proceed without restricting approve to POST, without a csrf_token in frappe/templates/includes/oauth_confirmation.html, and without scoping an active OAuth token check to the requesting client. An attacker can cause an authenticated user to approve an OAuth grant or reuse authorization state for the wrong client, exposing data and permitting actions within the granted scopes. This issue is fixed in versions 15.114.0 and 16.26.0.

🎖@cveNotify

🚨 CVE-2026-66785
A flaw was found in Submariner. This vulnerability allows a malicious cluster (spoke) to redirect network traffic from other connected clusters (peer clusters) by publishing a specially crafted network endpoint. The system fails to properly validate the network subnets provided by the malicious cluster, enabling it to declare arbitrary network ranges. Consequently, all network traffic intended for these arbitrary ranges from peer clusters will be rerouted through the attacker's tunnel, potentially leading to unauthorized information disclosure or network disruption.

🎖@cveNotify

🚨 CVE-2026-72852
hank-ai/darknet sizes a convolutional layer's weight and output heap buffers by multiplying configuration fields taken from a .cfg file in unchecked 32-bit int arithmetic. In src-lib/convolutional_layer.cpp, l.nweights is computed as (c / groups) * n * size * size and l.outputs as l.out_h * l.out_w * l.out_c, and both feed xcalloc directly. A .cfg whose true dimension product exceeds INT_MAX wraps to a small or zero value, so the allocation is undersized; for example width and height of 256 with filters of 65536 gives 2^32, which wraps to 0. forward_convolutional_layer then re-derives the GEMM dimensions with a different operand order, computing k as l.size*l.size*l.c / l.groups where the allocation divided before multiplying, and reads and writes through the undersized buffer. Loading the crafted .cfg for inference or training is sufficient and no valid .weights file is required. The reported proof of concept observed a heap buffer overflow read in gemm_nn_fast under AddressSanitizer and glibc allocator metadata corruption in a release build of the same input, indicating an out-of-bounds write.

🎖@cveNotify

🚨 CVE-2026-70651
libvips is a fast image processing library with low memory needs. Prior to version 8.18.3, libvips built without libtiff support but with ImageMagick support can overflow the combined frame height while loading a crafted multi-page TIFF through VipsForeignLoadMagick. The vulnerable calculations in libvips/foreign/magick6load.c and libvips/foreign/magick7load.c multiply the per-page Ysize by n_frames without a checked bound, which can cause a heap buffer over-read and process crash. Most package-manager builds include libtiff and do not use this affected fallback path. This issue is fixed in version 8.18.3.

🎖@cveNotify

🚨 CVE-2026-71485
Centrifugo is an open-source scalable real-time messaging server. Prior to 6.9.0, Centrifugo copies the client-controlled protocol.ConnectRequest.headers map through OnClientConnecting in internal/client/handler.go, ConnectEvent.Headers, and SetEmulatedHeadersToContext. The requestHeaders path in internal/proxy/http.go, the requestMetadata path in internal/proxy/grpc.go, and the Consume path in internal/unigrpc/grpc.go can forward an allowlisted value as a trusted backend header or metadata value. A remote client can spoof a header such as x-trusted-user for connect, refresh, subscribe, publish, RPC, and related proxy calls when the backend relies on that header for authentication or authorization. The unidirectional gRPC transport has no transport-level HTTP header that can override the emulated value. This issue is fixed in version 6.9.0.

🎖@cveNotify

🚨 CVE-2026-16964
IBM AIX 7.2, and 7.3 and IBM PowerVM VIOS 4.1 could allow a remote attacker to intercept messages and forge replies due to the exposure of sensitive information.

🎖@cveNotify

🚨 CVE-2026-17060
IBM AIX 7.2, and 7.3 and IBM PowerVM VIOS 4.1 could allow a remote attacker to obtain sensitive information and cause a denial of service due to a kernel heap over-read.

🎖@cveNotify

🚨 CVE-2026-17163
IBM AIX 7.2, and 7.3 and IBM PowerVM VIOS 4.1 could allow a remote attacker to cause a denial of service due to improper validation of an array size field.

🎖@cveNotify

🚨 CVE-2026-49217
Mailu is a mail server as a set of Docker images. Prior to version 2024.06.52, a missing authorization check in the Mailu admin REST API allows any unauthenticated attacker to remove any potential IP restriction or update the comment field from any existing user token provided the REST API is enabled. Upgrade to Mailu 2024.06.52 to receive a patch or, as a workaround, turn the REST API off.

🎖@cveNotify

🚨 CVE-2026-54389
Ghidra before 12.1.3 contains an uncontrolled resource consumption vulnerability in the PDB parser that allows attackers to terminate the Ghidra process by supplying a crafted PDB file with an oversized parameters section. The AbstractPdb deserialization routine reads all remaining parameters into an unbounded list, causing uncontrolled heap growth that triggers an OutOfMemoryError which bypasses exception handling and crashes the application.

🎖@cveNotify

🚨 CVE-2026-54505
TREK is a collaborative travel planner. Prior to 3.1.0, when the Journey add-on is enabled, TREK interpolates the unescaped activeSuggestion.title value into journey.frontpage.suggestionText through client/src/i18n/TranslationContext.tsx and renders the result with dangerouslySetInnerHTML in client/src/pages/JourneyPage.tsx. A trip owner can store HTML in a qualifying trip title, and GET /api/journeys/suggestions returns that title through getSuggestions(userId) to a collaborator who opens the authenticated Journey page. The markup is inserted as live DOM in the collaborator's session, enabling content spoofing and UI redress, although the default Content Security Policy blocks inline handlers and script execution. This issue is fixed in version 3.1.0.

🎖@cveNotify

🚨 CVE-2026-55765
CloudNativePG is a platform designed to manage PostgreSQL databases within Kubernetes environments. Prior to 1.28.4 and 1.29.2, CloudNativePG embedded cleartext role passwords in `ALTER ROLE` and `CREATE ROLE` statements generated by SetUserPassword in pkg/management/postgres/utils/roles.go and appendPasswordOption in internal/management/controller/roles/postgres.go. When pg_stat_statements was preloaded with track_utility enabled and an untrusted tenant held pg_monitor or pg_read_all_stats, the tenant could recover platform-managed superuser or application-owner passwords, reconnect through enabled superuser TCP access, and execute operating system commands in the database pod with `COPY ... FROM PROGRAM`. Clusters using SCRAM-SHA-256 verifiers in managed-role Secrets were not affected. This issue is fixed in versions 1.28.4, 1.29.2, and 1.30.0.

🎖@cveNotify

🚨 CVE-2026-65770
Improper neutralization of argument delimiters in a command ('argument injection') in Azure Managed Instance for Apache Cassandra allows an unauthorized attacker to execute code over a network.

🎖@cveNotify

🚨 CVE-2026-69519
Observable response discrepancy in Azure Stack HCI allows an unauthorized attacker to disclose information over a network.

🎖@cveNotify

🚨 CVE-2026-69558
Authorization bypass through user-controlled key in Microsoft Partner Center allows an unauthorized attacker to disclose information over a network.

🎖@cveNotify

🚨 CVE-2026-69836
Deserialization of untrusted data in Microsoft Entra ID allows an unauthorized attacker to execute code over a network.

🎖@cveNotify

🚨 CVE-2026-69851
Server-side request forgery (ssrf) in Azure Active Directory allows an authorized attacker to elevate privileges over a network.

🎖@cveNotify

🚨 CVE-2026-72860
The POST /api/provider-nodes/validate route in 9router takes a caller-supplied baseUrl and issues server-side HTTP requests to it, guarding the destination with assertPublicUrl from src/shared/utils/ssrfGuard.js. That guard compares hostname strings only: it resolves no DNS, does not revalidate after a redirect, and its IPv4-mapped IPv6 branch is unreachable. The branch matches ^::ffff:(\d+\.\d+\.\d+\.\d+)$, but the WHATWG URL parser canonicalizes such literals to hextets before the guard runs, so new URL("http://[::ffff:127.0.0.1]/").hostname yields [::ffff:7f00:1] and the pattern is tested against a string it is never handed. Every IPv4-mapped address therefore passes, and http://[::ffff:7f00:1] and http://[::ffff:a9fe:a9fe] reach loopback and link-local metadata addresses; a hostname whose A record points at an internal address passes as well because no resolution occurs. In the custom-embedding branch the upstream response body is truncated to 200 bytes and returned to the caller whenever the upstream status is neither 2xx nor 401 nor 403, which discloses the beginning of internal responses, and the other validation types remain usable for blind internal port scanning through status and timing differences. The caller-supplied apiKey is forwarded to the internal destination as an Authorization Bearer header. A dashboard session is required by default, and none is required when requireLogin is disabled.

🎖@cveNotify

🚨 CVE-2026-73267
A flaw was found in the clusterclaims-controller component of multicluster engine (MCE). A tenant with standard permissions to create and delete ClusterClaim resources can exploit this by manipulating the `spec.namespace` field. This allows the tenant to specify and delete any ManagedCluster, including the hub's local-cluster or other tenants' clusters, due to a missing ownership check. This vulnerability can lead to a denial of service by enabling unauthorized deletion of ManagedClusters.

🎖@cveNotify