Kubernetes
Monthly
Authentication bypass in qSnapper's privileged D-Bus service (versions before 1.3.3) allows any local unprivileged user to invoke administrative snapshot operations by piggy-backing on another client's prior Polkit authentication. The service stored authentication state in a single shared m_authenticated flag rather than per-client, so once any caller passed an admin check, all subsequent D-Bus callers inherited that state. No public exploit identified at time of analysis; the issue was found during a SUSE-coordinated security review and patched in v1.3.3.
VM escape in Kata Containers allows any Kubernetes user with pod-creation rights to break out of the VM sandbox and gain full read/write access to the host filesystem. All Kata Containers installations prior to commit ffa59ce3aa78 are affected when using the default configuration.toml, which enables the `virtio_fs_extra_args` and `kernel_params` pod annotations out of the box. An attacker crafts a pod with two annotations: one to redirect virtiofsd to serve the host root filesystem (`/`) into the guest VM, and a second to enable the agent debug console - after which the entire host filesystem is accessible from inside the supposedly isolated VM. A fully working proof-of-concept with confirmed output against Kata Containers 3.28.0 on Ubuntu 24.04 has been publicly disclosed; no public exploit confirmed as actively exploited (CISA KEV) at time of analysis.
Unvalidated jarURI handling in Apache Flink Kubernetes Operator exposes authenticated low-privilege users to server-side request forgery and arbitrary file read primitives against the operator pod. Any Kubernetes principal holding Custom Resource create permissions can submit a malicious FlinkSessionJob CR with a crafted jarURI - pointing to local filesystem paths, internal cloud metadata endpoints, link-local addresses, or any backing store reachable through Flink's pluggable filesystem layer - and retrieve that content via the submitted job. No public exploit has been identified at time of analysis, and EPSS sits at 0.01% (3rd percentile), but the confidentiality impact is rated High by NVD given the breadth of accessible internal resources.
runAsNonRoot bypass in containerd allows crafted container images to execute as UID 0 despite Kubernetes security policies designed to prevent root execution. The flaw stems from containerd treating numeric USER directives that overflow a 32-bit integer as usernames, and if the image's /etc/passwd maps that string to root, the container runs as root. No public exploit identified at time of analysis, but the issue was responsibly disclosed by Lei Wang (@ssst0n3) and fixed in multiple containerd release branches.
Authorization bypass in mcp-server-kubernetes versions prior to 3.6.0 allows authenticated clients to invoke any Kubernetes tool - including destructive operations like kubectl_delete, exec_in_pod, and node_management - regardless of ALLOW_ONLY_READONLY_TOOLS, ALLOW_ONLY_NON_DESTRUCTIVE_TOOLS, or ALLOWED_TOOLS restrictions. The controls were enforced only at the tools/list discovery layer, leaving tools/call unguarded, which effectively reduces operator-configured least-privilege policies to cosmetic filters. Publicly available exploit code exists (a single curl invocation reproduces it), and in cluster-admin deployments the flaw is equivalent to full cluster compromise for any client reaching the endpoint.
Privilege escalation in Fission (Kubernetes serverless framework) versions <= 1.22.0 allows any user able to deploy or update a Function CRD to read every Secret and ConfigMap in the runtime namespace, bypassing the Function.spec.secrets allowlist. The flaw stems from runtime pods running under the namespace-privileged fission-fetcher ServiceAccount whose automounted token is reachable by the user-code container alongside the fetcher sidecar. No public exploit identified at time of analysis, but the vendor-issued advisory (GHSA-85g2-pmrx-r49q) describes a trivially reachable abuse path once a function can be deployed.
Unauthenticated archive CRUD in Fission's storagesvc (≤ v1.22.0) lets any in-cluster workload list, download, replace, or delete function deployment archives across all tenants by hitting the ClusterIP-exposed /v1/archive and /v1/archives endpoints. Because uploaded archives are later fetched and executed by function specialization, the flaw escalates from a tenant data-exposure issue to in-cluster code execution. No public exploit identified at time of analysis, but the trivial HTTP pattern and lack of auth middleware make weaponization straightforward for any attacker with a foothold pod.
Arbitrary code execution in InternLM lmdeploy <= 0.12.3 occurs because trust_remote_code=True is hardcoded across HuggingFace model-loading call sites in lmdeploy/archs.py and lmdeploy/utils.py. An attacker who can influence the model_path passed to an lmdeploy serving process can point it at a malicious HuggingFace repository, causing Transformers to download and execute attacker-controlled Python code with the privileges of the serving daemon. Publicly available exploit code exists in the GHSA advisory, and an upstream fix has been merged via PR #4511 (fixed in 0.13.0).
Cross-namespace privilege escalation in Apache Camel K (2.0.0-2.8.0, 2.9.0-2.9.1, 2.10.0) lets an authenticated Kubernetes user who can create Build resources dictate the target namespace for the resulting Pod, including the privileged operator namespace. By steering Pod generation into the operator's namespace, a tenant confined to their own namespace can break out of RBAC boundaries and interact with higher-privilege resources. CVSS 8.1; no public exploit identified at time of analysis and EPSS is negligible (0.02%), but SSVC rates the technical impact as total.
Privilege escalation in the Apache Airflow CNCF Kubernetes provider (versions before 10.17.0) exposes worker JWT tokens to any principal holding read-only access to Kubernetes Pods. When Airflow runs with the KubernetesExecutor, the scheduler embeds a 'workload'-scoped Execution API JWT in the pod's workload payload, so a low-privileged user who can read pod objects can harvest that token and impersonate a running task via the Task SDK, including actions that mutate Airflow database state. There is no public exploit identified at time of analysis; EPSS is 0.01% (2nd percentile) and CISA SSVC rates exploitation as none, so this is an insider/lateral-movement risk rather than a mass-exploited flaw.
Stored cross-site scripting in Argo CD lets a low-privileged developer escalate to admin by planting a malicious `link.argocd.argoproj.io/*` annotation on any Application; the value is rendered as an unvalidated `<a href>` in the Summary tab's URLs section, so a `javascript:` payload hidden behind a benign 'GitHub Repo' label executes in the admin's authenticated same-origin session when clicked. A detailed public exploit exists (PoC in the vendor advisory demonstrating session exfiltration via Burp Collaborator), though EPSS is very low (0.04%) and there is no public exploit identified as actively used; this is a CVSS 8.7 authenticated, user-interaction-dependent issue affecting Argo CD v2.x, v3.2.x, v3.3.x and v3.4.x. An attacker with `applications` write access can capture the admin's session data and drive Argo CD APIs as admin.
Information disclosure in Argo CD v3.x exposes plaintext Kubernetes Secret values to authenticated users who can view application diffs via the ServerSideDiff feature. This is an incomplete fix for a prior vulnerability (GHSA-3v3m-wc6v-x4x3): the original patch masked top-level Secret data in ServerSideDiff responses but failed to sanitize Secret content embedded in the `kubectl.kubernetes.io/last-applied-configuration` annotation on `predictedLive` objects, leaving raw `data`, `stringData`, and sensitive annotation values readable in UI and CLI diff output. A publicly available proof-of-concept exists; no KEV listing is present at time of analysis, but the Changed Scope (S:C) in the CVSS vector indicates that exposed secrets may belong to workloads beyond the Argo CD application boundary, amplifying real-world impact in multi-tenant environments.
Local file disclosure in NiceGUI versions <= 3.11.1 allows remote unauthenticated attackers to read arbitrary files accessible to the server process when applications pass user-controlled content to ui.restructured_text(). The flaw stems from Docutils being invoked without disabling file-insertion directives (include, csv-table :file:, raw :file:), enabling exfiltration of secrets, credentials, and source code. No public exploit identified at time of analysis, but the vendor advisory provides full directive-level proof patterns.
Open WebUI versions ≤0.9.4 allow authenticated users to bypass SSRF protections and read internal network resources via HTTP redirect-following. The vulnerability exists in five distinct code paths: web-fetch retrieval, image-loading endpoints, chat-completion image inlining, and OAuth/tool-server execution flows. Any authenticated user can submit a public URL that 302-redirects to internal addresses (127.0.0.1, 169.254.169.254 cloud metadata, RFC1918 private networks) and receive the response body. The vendor confirms active exploitation in the wild was NOT observed, but publicly available exploit code exists (PoC in advisory) and EPSS score of 0.00043 (4.3%) suggests low automated scanning targeting at time of analysis. Fixed in version 0.9.5 released March 2025.
Kuma Control Plane with default configuration leaks admin bootstrap tokens and signing keys to any website an operator visits if the control plane is reachable from their browser. The vulnerability combines default CORS settings allowing all origins (CorsAllowedDomains: [".*"]) with LocalhostIsAdmin: true, which grants admin privileges to any request from 127.0.0.1 without validating whether it originates from a trusted same-origin context. An attacker's JavaScript on a visited webpage can cross-origin fetch the admin token and cryptographic material via browser requests to localhost:5681. This is not actively exploited in the wild but represents a realistic threat in developer and testing environments where control planes run on workstations with web browsers.
Token leakage in Portainer's authentication middleware allows JWT bearer tokens passed via the `?token=<JWT>` URL query parameter to be harvested from reverse-proxy access logs, browser history, and HTTP Referer headers, enabling account takeover for the validity window of the token (default 8 hours). The flaw affected any user with container exec/attach rights - not just administrators - and a leaked admin token grants full control of Portainer and every managed Docker/Kubernetes environment. No public exploit identified at time of analysis, though the underlying behavior was present since JWT auth was introduced and the GitHub Security Advisory provides sufficient detail to weaponize.
Authorization bypass in Portainer 2.33.0 through 2.33.7 allows authenticated users to access Kubernetes cluster resources beyond their assigned permissions due to missing return statement in middleware error handling. Any user with a valid Portainer session can exploit this to read or modify Kubernetes secrets, pods, deployments, and other resources on endpoints they should not access. The flaw affects both Community Edition and Enterprise Edition. Fixed in version 2.33.8 and inherently absent from 2.39.0+. No public exploit code identified at time of analysis, though the single-line code fix and detailed GitHub advisory make reproduction straightforward for authenticated attackers.
Arbitrary file read in Portainer Community Edition allows authenticated low-privileged users to exfiltrate any file readable by the Portainer process — commonly root — by submitting a Git-backed stack whose docker-compose.yml is a symlink to a target like /etc/shadow or a Kubernetes service account token. The flaw stems from go-git v5 materializing Git symlink blobs as real OS symlinks during checkout combined with GetFileContent reading the entry point through os.ReadFile without resolving links, and is amplified by Portainer's scheduled stack auto-update, which lets a previously benign repository turn malicious after initial review. A fully validated proof-of-concept exists from the reporter (b-hermes), but there is no public exploit identified at time of analysis and no CISA KEV listing.
Stored Cross-Site Scripting (XSS) in vCluster Platform allows authenticated attackers with namespace creation privileges to inject malicious scripts via the templateRef name field, potentially escalating to Global-Admin access. The vulnerability affects versions prior to 4.4.3, 4.5.5, 4.6.2, 4.7.1, and 4.8.0. With a CVSS score of 9.0 and changed scope (S:C), this represents a critical privilege escalation path in Kubernetes multi-tenancy environments. No active exploitation confirmed by CISA KEV at time of analysis, but the low attack complexity (AC:L) and clear attack path make this an immediate patching priority for organizations running vCluster Platform.
Unauthorized write access to Traefik's live dynamic configuration is achievable by a low-privileged Kubernetes tenant in shared Gateway deployments, because the Gateway API provider's `isInternalService()` check accepts any `TraefikService` name ending in `@internal` - not exclusively the intended `api@internal`. This allows a tenant with `HTTPRoute` creation rights to route external traffic to `rest@internal`, fully bypassing the `providers.rest.insecure=false` safeguard and gaining `PUT /api/providers/rest` write access. No active exploitation is confirmed (not in CISA KEV, EPSS at 0.01%), but a detailed proof-of-concept is published in the GHSA advisory and the SSVC framework marks the attack as automatable for tenants meeting the prerequisites.
Sandbox escape in OpenClaude (npm package openclaude) versions before 0.5.1 allows a prompt-injected LLM to disable host sandboxing by setting the model-controlled `dangerouslyDisableSandbox: true` flag in any Bash tool_use call, yielding full unsandboxed command execution on the host. CVSS 4.0 scores this 9.3 Critical (AV:N/AC:L/PR:N/UI:N, VC/VI/VA:H); no public exploit identified at time of analysis beyond the reporter's PoC, but the upstream fix has been merged. The flaw is especially severe because it is reachable under default settings (`allowUnsandboxedCommands` defaults to true).
Credential-harvesting malware compromised 84 versions of 42 TanStack npm packages on 2026-05-11 via chained GitHub Actions exploitation. Attackers combined pull_request_target misconfiguration, Actions cache poisoning, and OIDC token memory extraction to publish malicious code under the legitimate TanStack identity. Installing any affected version executes a 2.3 MB obfuscated payload that exfiltrates AWS/GCP/Kubernetes credentials, npm tokens, GitHub secrets, SSH keys, and HashiCorp Vault tokens over encrypted Session/Oxen messenger infrastructure. The payload propagates by republishing victim-maintained packages with identical injection. Socket.dev and the TanStack team confirmed the incident via GHSA-g7cv-rxg3-hmpx. No EPSS or CISA KEV data available for this recent supply-chain attack. CVSS 9.6 reflects the cross-scope credential theft impact (S:C/C:H/I:H), though exploitation requires user-initiated package installation (UI:R).
Template injection in Rancher Local Path Provisioner allows Kubernetes cluster operators with ConfigMap edit permissions to escalate privileges to node-level root access. Attackers with write access to the local-path-config ConfigMap can inject malicious Pod templates that bypass security controls, creating privileged containers with full host filesystem access. This enables theft of ServiceAccount tokens from co-located pods, access to other tenants' persistent volume data, and arbitrary modification of host node files. Vendor-released patch: v0.0.36. CVSS 8.7 (High) reflects the high-privilege prerequisite (PR:H) but scope change to container escape (S:C). No public exploit identified at time of analysis, though exploitation is straightforward for authenticated cluster operators.
External Secrets Operator versions 0.1.0 through 2.4.0 allow authenticated users with ExternalSecret creation permissions to escalate privileges by crafting Service Account token templates that cause the operator to generate long-lived tokens for any service account in the namespace. An attacker can impersonate service accounts without requiring direct TokenRequest or Secret creation permissions, effectively bypassing RBAC controls. The attack requires the attacker already has ExternalSecret creation permissions and the cluster must have service account token generation enabled, limiting the practical scope to already-privileged users seeking lateral privilege expansion within a namespace.
Server-side request forgery in MCP Registry's HTTP namespace verification endpoint allows unauthenticated attackers to reach internal IPv4 addresses via specially-crafted IPv6 addresses that encode or tunnel to RFC1918 and cloud-metadata services. The vulnerability exists in the private-address blocklist used by `safeDialContext`, which fails to block IPv6 6to4 (2002::/16), NAT64 well-known (64:ff9b::/96), NAT64 local-use (64:ff9b:1::/48), and deprecated site-local (fec0::/10) prefixes. On dual-stack and IPv6-only cloud deployments (GKE IPv6, AWS IPv6-only EC2, Azure NAT64), this enables direct connections to metadata services and internal Kubernetes API servers. No public exploit code identified at time of analysis, but proof-of-concept has been demonstrated against the production registry.
Server-Side Request Forgery (SSRF) in nuxt-og-image 6.2.5 through 6.4.8 allows remote attackers to bypass the incomplete IPv6 denylist and redirect validation, reaching internal IP addresses and services through incomplete IPv6 prefix filtering and unauthenticated HTTP redirect following. The vulnerability affects the OG image rendering component used by Nuxt applications, enabling attackers to leak internal service responses by injecting crafted IPv6-mapped addresses or chaining external redirects to internal targets.
Denial-of-service in vm2 Node.js sandbox allows unauthenticated remote attackers to crash host processes via unbounded Buffer.alloc() calls. The vm2 library's timeout mechanism cannot interrupt synchronous C++ native calls, enabling attackers to bypass configured timeout limits and exhaust host heap memory with a single HTTP request. Version 3.11.0 patches this flaw by introducing bufferAllocLimit controls. Publicly available exploit code exists (GHSA-6785-pvv7-mvg7 includes working POC), and while EPSS data is unavailable and the vulnerability is not listed in CISA KEV, the vendor-confirmed POC demonstrates reliable exploitation against default configurations.
Remote unauthenticated attackers can crash Node.js processes running vm2 <= 3.10.5 by triggering an unhandled Promise rejection that terminates the host application. The vulnerability exploits an incomplete fix for CVE-2026-22709 - while previous patches sanitized `.then()` and `.catch()` callback chains, they failed to intercept unhandled rejections originating from Promise constructor executors. Publicly available exploit code exists (GitHub advisory GHSA-hw58-p9xv-2mjh). The attack requires minimal resources (150-byte HTTP request) but achieves high impact by crashing entire server processes serving all concurrent users, with demonstrated persistent DoS despite container orchestration restart policies.
etcd RBAC authorization bypass allows authenticated users to read unauthorized data or attach leases via PrevKv or lease attachment features in transaction Put requests, circumventing role-based access control checks. Affects etcd 3.4.x through 3.6.x before patched versions 3.4.44, 3.5.30, and 3.6.11. While Kubernetes deployments are typically not affected because the API server handles its own authorization, etcd deployments with reliance on etcd's built-in RBAC-particularly those managed directly or used outside Kubernetes-face exposure to privilege escalation and unauthorized data access by already-authenticated users.
Kubetail Dashboard prior to version 0.14.0 fails to validate the Origin header on WebSocket connection upgrades, enabling Cross-Site WebSocket Hijacking (CSWSH) attacks. An authenticated user visiting a malicious web page can be exploited to stream their Kubernetes container logs-including credentials, tokens, and PII often present in logs-to an attacker-controlled server. The vulnerability affects both desktop deployments at localhost:7500 and cluster deployments behind HTTP basic auth, with browser ambient credentials automatically attached to the WebSocket handshake.
Kubernetes Secret extraction in Argo CD v3.2.0-3.2.10 and v3.3.0-3.3.8 allows authenticated users with read-only application permissions to retrieve plaintext credential data including service account tokens, TLS certificates, database passwords, and API keys via the ServerSideDiff endpoint. The vulnerability exists due to missing data masking in the gRPC/REST ServerSideDiff function, which returns raw Kubernetes Server-Side Apply dry-run responses containing unredacted Secret values from etcd when applications are annotated with 'IncludeMutationWebhook=true'. A functional proof-of-concept exploit exists demonstrating automated extraction of all accessible secrets. Vendor-released patches (3.2.11, 3.3.9) are available. CVSS 9.6 reflects network-exploitable, low-complexity attack requiring only low-privilege authenticated access with cross-scope high confidentiality/integrity impact.
ServiceAccount impersonation bypass in Rancher Fleet allows tenants with git push access to multi-tenant clusters to read secrets from any namespace across all downstream clusters. Two distinct code paths failed to properly apply RBAC constraints: Helm's lookup function executed with cluster-admin credentials instead of the impersonated ServiceAccount, and valuesFrom secret references in fleet.yaml bypassed namespace isolation. Confirmed active exploitation status unknown (not in CISA KEV). CVSS 9.9 with scope-change modifier reflects potential credential leakage to external services. Fleet versions 0.12.0 through 0.15.0 affected across multiple Rancher release branches. Patches available for all supported versions with detailed version matrix provided by SUSE.
Unauthenticated remote attackers crash Gotenberg 8.x (≤ 8.31.0) by triggering a race condition between webhook goroutine context reuse and Echo framework connection pooling. When webhook middleware spawns an async goroutine holding an `echo.Context` reference, the synchronous handler returns immediately, recycling the context to Echo's `sync.Pool`. Concurrent requests reset the pooled context, causing unchecked type assertions in the still-running webhook goroutine to panic outside any `recover()` scope, terminating the process with exit code 2. Twenty-four webhook requests plus sixty concurrent GET requests demonstrate reliable two-second crash windows. No patch was available at initial disclosure; upstream commit fixes the panic in version 8.32.0. CVSS 7.5 (AV:N/AC:L/PR:N/UI:N) reflects trivial unauthenticated network exploitation producing complete service disruption.
Stored XSS in Kyverno policy-reporter-ui 2.5.1 and earlier allows an attacker with Kubernetes write access to PolicyReport resources to inject arbitrary HTML into the Vue 3 frontend, which executes in the browser of any authenticated UI user who views the affected namespace. The sink is the v-html directive in PropertyCard.vue, which is reached whenever a property value is not an http/https URL and exceeds 75 characters - a bypass condition trivially satisfied by padding any HTML payload. A public proof-of-concept demonstrating session cookie exfiltration via a crafted PolicyReport object is included in the GitHub advisory GHSA-q98m-7w8c-w388; no public exploit identified as confirmed actively exploited (not in CISA KEV), though EPSS of 0.03% reflects low opportunistic risk given the Kubernetes-specific prerequisite.
Kubewarden versions before 1.35.0 permit RBAC reconnaissance attacks when users with AdmissionPolicy or AdmissionPolicyGroup creation privileges craft policies using the unchecked `can_i` host capability. The vulnerability allows enumeration of any user or service account permissions across the cluster via SubjectAccessReview requests executed with policy-server privileges, despite the absence of context-aware resource grants. This information disclosure enables attackers to discover sensitive permission configurations without requiring cluster-wide policy creation rights, a capability not available by default but exploitable when granted.
Path traversal in MinIO's ReadMultiple internode storage-REST endpoint allows authenticated cluster peers or root-credential holders to read arbitrary files from the host filesystem outside configured drive roots. Distributed-erasure (multi-node) deployments are affected; single-node standalone deployments are not. The vulnerability exists in all releases from RELEASE.2022-07-24T01-54-52Z through RELEASE.2025-09-07T16-13-09Z and has been fixed as of MinIO AIStor RELEASE.2024-10-23T19-38-07Z (with security patch RELEASE.2026-04-14T21-32-45Z recommended). No public exploit code or active exploitation has been identified at time of analysis.
Argo Workflows executor logs artifact repository credentials in plaintext to pod logs during artifact operations, exposing S3 access/secret keys, GCS service account keys, Azure storage keys, and Git passwords. Users with Kubernetes RBAC permissions to read pod logs in the workflow namespace can extract these credentials directly from workflow execution logs. This vulnerability affects Argo Workflows v4.0.0 through v4.0.4 and represents an incomplete fix of CVE-2025-62157. Vendor-released patch (v4.0.5) is available with GitHub commit bdd40908 removing credential-bearing struct logging. No public exploit identified at time of analysis, though exploitation is trivial given the included working proof-of-concept YAML.
Argo Workflows v3 (< 3.7.14) and v4 (< 4.0.5) allow users to bypass templateReferencing Strict/Secure mode restrictions by setting WorkflowSpec fields like hostNetwork, serviceAccountName, securityContext, tolerations, and volumes. The incomplete fix for CVE-2026-31892 only blocked podSpecPatch but left other security-sensitive fields unvalidated. Authenticated users with create Workflow permission can inject host network access, switch service accounts, modify pod security contexts, or schedule on control-plane nodes despite referencing hardened WorkflowTemplates. Vendor-released patch: v3.7.14 and v4.0.5 (commit 2727f3f). No public exploit identified at time of analysis, but exploitation is straightforward given detailed reproduction steps in the advisory.
Memory exhaustion crashes Argo Workflows server via unauthenticated multi-gigabyte webhook requests to the publicly accessible `/api/v1/events/` endpoint. The Webhook Interceptor's `io.ReadAll` call allocates unbounded memory before signature verification, enabling remote attackers to trigger out-of-memory (OOM) conditions without authentication or credentials. Vendor-released patches enforce a 2MB body size limit via `io.LimitReader`. Publicly available exploit code exists (conceptual proof-of-concept published in GitHub security advisory GHSA-jcc8-g2q4-9fxq). No active exploitation confirmed by CISA KEV at time of analysis, but the low attack complexity (CVSS AV:N/AC:L/PR:N) and public PoC create immediate risk for internet-exposed Argo Workflows deployments.
Missing authorization checks in Argo Workflows v4.0.0-4.0.4 allow any authenticated user-even those with fake Bearer tokens-to create, read, update, and delete Kubernetes ConfigMaps containing workflow synchronization limits. The ConfigMap-backed sync provider (server/sync/sync_cm.go) completely omits auth.CanI permission validation on all four CRUD endpoints. Publicly available exploit code exists (detailed PoC in advisory). CVSS 8.5 reflects network-accessible authentication bypass enabling high integrity/availability impact through denial-of-service and arbitrary ConfigMap manipulation. Patch released in version 4.0.5 adding checkConfigMapPermission() calls to validate Kubernetes RBAC before operations.
Authenticated users with low privileges can read cleartext Kubernetes Secret data through Argo CD's ServerSideDiff feature in versions 3.2.0-3.2.10 and 3.3.0-3.3.8. This scope-changing vulnerability (CVSS:3.1 S:C) allows attackers to access sensitive credential data managed by Kubernetes, including database passwords, API tokens, and certificates, by exploiting the server-side diff functionality. With a 7.7 CVSS score and low attack complexity (AC:L), this represents a significant confidentiality breach requiring only network access and basic authentication-no public exploit identified at time of analysis, but the technical barrier to exploitation is minimal.
Traefik versions prior to 2.11.43, 3.6.14, and 3.7.0-rc.2 fail to enforce cross-namespace isolation for middleware references nested inside Chain middlewares, allowing actors with permission to create CRDs in their own namespace to bypass the allowCrossNamespace=false restriction and apply middleware from arbitrary namespaces. This authorization bypass affects Kubernetes clusters relying on namespace isolation controls and can enable unauthorized reuse of security-sensitive middleware policies across namespace boundaries.
Cross-namespace privilege escalation in Kyverno 1.17.x allows authenticated namespace administrators to bypass RBAC controls and read ConfigMaps from any Kubernetes namespace. The vulnerability exploits unvalidated `configMap.namespace` field in Kyverno's ConfigMap context loader, enabling attackers to leverage Kyverno's privileged service account permissions. This is a regression following incomplete fix for CVE-2026-22039, which addressed the same issue in `apiCall` context but missed the ConfigMap loader. Patch available in version 1.17.2. CVSS 7.7 with Changed Scope indicates significant multi-tenant cluster risk; EPSS data not available but the regression nature and RBAC bypass impact warrant immediate patching in multi-tenant environments.
Lua code injection in Contour (Kubernetes ingress controller for Envoy) from v1.19.0 through the fixes in v1.31.6/v1.32.5/v1.33.4 lets a low-privileged tenant achieve arbitrary code execution inside the shared Envoy proxy. A user with RBAC rights to create or modify HTTPProxy resources injects Lua via the cookie-rewrite pathRewrite.value field; although the code only runs on the attacker's own route, it executes in shared infrastructure and can steal Envoy's xDS credentials (exposing other tenants' TLS keys) or cause denial of service for co-tenants. No public exploit identified at time of analysis, and EPSS is low (0.07%, 22nd percentile), consistent with the requirement for pre-existing cluster RBAC access.
Denial of service in the Argo Workflows controller (versions 3.6.5 through 4.0.4, plus the 3.7.0-rc1 through 3.7.12 pre-releases) lets any user able to submit a workflow crash-loop the entire controller by attaching a malformed workflows.argoproj.io/pod-gc-strategy pod annotation that contains no '/' separator, triggering an out-of-range index panic in an informer goroutine that sits outside the controller's recover() scope. Because the poisoned pod persists across restarts, the controller enters CrashLoopBackOff and halts all workflow processing cluster-wide until the pod is manually deleted. A working proof-of-concept manifest is published in the GHSA advisory (SSVC exploitation status: poc), but there is no evidence of active exploitation and EPSS is very low at 0.04%.
Cache poisoning in @astrojs/node versions 9.4.4 and earlier allows unauthenticated remote attackers to poison CDN caches by sending malformed if-match headers to static asset endpoints, causing the server to return 500 errors with immutable one-year cache directives instead of the correct 412 Precondition Failed response. This vulnerability affects all subsequent requests to poisoned assets until the cache expires, breaking application functionality for legitimate users. The vulnerability is not actively exploited in the wild, but proof-of-concept exploitation is straightforward and requires only a single crafted HTTP request.
VolumeMount path restriction bypass in Tekton Pipelines prior to v1.11.1 permits authenticated users holding Task or TaskRun creation permissions to mount volumes over internal /tekton/ paths using dot-dot traversal sequences, violating Tekton's internal execution isolation boundary. The flaw exists in both the v1 and v1beta1 API validation layers (pkg/apis/pipeline/v1/container_validation.go and pkg/apis/pipeline/v1beta1/task_validation.go), where strings.HasPrefix is applied to the raw, un-normalized mount path instead of the filepath.Clean-resolved form. No public exploit code has been identified and SSVC assessment confirms no active exploitation; however, in shared or multi-tenant Kubernetes CI/CD environments the integrity risk is meaningful - an attacker can inject falsified task results trusted by downstream pipelines, alter step scripts before execution, or disrupt entrypoint coordination state.
Unbounded memory allocation in Tekton Pipelines' HTTP resolver allows any authenticated tenant to crash the shared resolver pod for an entire Kubernetes cluster. The `FetchHttpResource` function reads HTTP response bodies without a byte-level size cap, so a low-privileged user who can create TaskRuns or PipelineRuns can direct the resolver at an attacker-controlled server delivering multi-gigabyte responses, exhausting the pod's 4 GiB memory limit and triggering an OOM-kill. Because all resolver types - Git, Hub, Bundle, Cluster, and HTTP - share the same `tekton-pipelines-resolvers` pod, repeated exploitation causes a sustained crash loop that denies CI/CD pipeline resolution to every tenant on the cluster. A PoC is included in the GHSA advisory and SSVC confirms exploit code exists; however, no CISA KEV listing indicates no confirmed widespread exploitation at time of analysis.
Argument injection in the Tekton Pipelines git resolver (versions 1.0.0 through 1.11.0) lets a multi-tenant CI/CD user achieve remote code execution on the resolver pod by supplying a crafted 'revision' parameter. Because the revision string is passed unvalidated as a positional argument to 'git fetch', an attacker can smuggle git flags such as '--upload-pack=<binary>'; combined with validateRepoURL permitting local filesystem paths (URLs beginning with '/'), this executes an arbitrary binary. Since the tekton-pipelines-resolvers ServiceAccount has cluster-wide read on all Secrets, this escalates to full cluster secret exfiltration; a proof of concept exists per CISA SSVC, though the EPSS probability is very low (0.07%).
Broken access control in Esri Portal for ArcGIS 11.4, 11.5, and 12.0 (Windows, Linux, and Kubernetes) fails to correctly validate the permissions bound to developer credentials, allowing a holder of such credentials to access resources or perform actions beyond their intended authorization. The vendor rates it CVSS 9.8 (network, no privileges, no interaction), though the description's reference to 'developer credentials' implies some prior credential is involved. There is no public exploit identified at time of analysis and EPSS is very low (0.04%, 13th percentile), but SSVC flags the technical impact as total and the flaw as automatable.
Credential leakage in Tekton Pipelines git resolver allows authenticated users to exfiltrate system-configured Git API tokens (GitHub PAT, GitLab tokens) by directing the resolver to attacker-controlled endpoints. Affects versions 1.0.0 through 1.10.0 when users omit the token parameter in TaskRun or PipelineRun configurations. CVSS 7.7 with scope change reflects cross-tenant credential theft potential in multi-tenant Kubernetes environments. No active exploitation confirmed (not in CISA KEV), but exploitation is straightforward for authenticated cluster users with TaskRun/PipelineRun creation privileges.
Tekton Pipelines 0.43.0 through 1.11.0 allows authenticated attackers to bypass trusted resource verification policies via unanchored regular expression patterns that match substrings rather than exact resource sources, enabling policy manipulation and unauthorized verification mode changes. The vulnerability stems from Go's regexp.MatchString function matching patterns anywhere within a string rather than requiring full anchoring, permitting attackers to craft source URIs containing trusted patterns as substrings to trigger unintended policy matches and potentially apply weaker verification keys or modes.
Unauthenticated remote denial-of-service in ChilliCream's Hot Chocolate GraphQL server for .NET (HotChocolate.Language package) lets any client crash the entire worker process by sending a small, deeply nested GraphQL document. Because the recursive-descent Utf8GraphQLParser has no depth limit, ~40 KB of nested selection sets, object/list values, or list types triggers an uncatchable StackOverflowException that terminates the process, dropping all in-flight HTTP requests, IHostedService background tasks, and WebSocket subscriptions on that worker. No public exploit identified at time of analysis and EPSS is low (0.09%), but the attack is trivial (AC:L, PR:N) and CISA SSVC flags it as automatable, so it is a practical availability threat to internet-facing GraphQL endpoints.
Race condition in pyLoad's Flask session cookie handler allows unauthenticated attackers to manipulate the SESSION_COOKIE_SECURE flag globally across all concurrent requests by spoofing the X-Forwarded-Proto header. On deployments behind a TLS-terminating proxy, this enables session cookie downgrade attacks resulting in plaintext cookie transmission; on default plain HTTP deployments, it causes session denial of service by forcing the Secure flag and breaking all concurrent user sessions. The vulnerability requires no authentication and exploits a multi-threaded race window in the Cheroot WSGI server (request_queue_size=512) combined with missing proxy origin validation (acknowledged TODO in code).
Command injection in mcp-server-kubernetes port_forward function allows authenticated network attackers to expose internal Kubernetes services to external networks or bypass namespace restrictions. The vulnerability (CVSS 8.3) stems from unsafe string concatenation and space-splitting of kubectl arguments, enabling arbitrary flag injection via fields like resourceName or namespace. Attackers can inject '--address=0.0.0.0' to bind port-forwards on all network interfaces, exposing databases and internal APIs beyond localhost. Affects mcp-server-kubernetes <= 3.4.0. No public exploit identified at time of analysis, though exploitation requires only low complexity (AC:L) with authenticated access (PR:L).
Kyverno's apiCall service helper automatically injects the controller's ServiceAccount token into HTTP requests when ClusterPolicy or GlobalContextEntry authors omit an Authorization header, enabling token exfiltration to attacker-controlled endpoints via confused deputy vulnerability. Affects deployments where policy YAML repositories are compromised (GitOps threat model) or ClusterPolicy creation is possible. CVSS 8.1 (High) reflects network attack vector with low complexity and low privileges required. No CISA KEV listing or public exploit identified at time of analysis, but GitHub advisory includes working proof-of-concept demonstrating token injection and exfiltration.
Kubernetes Service Account token disclosure in the odh-dashboard component of Red Hat OpenShift AI (RHOAI) lets an authenticated low-privileged user retrieve SA tokens via an exposed NodeJS endpoint, then reuse them to reach Kubernetes resources beyond the dashboard's intended scope. Rated CVSS 9.9 with a changed scope, the flaw effectively converts limited dashboard access into broad cluster access. There is no public exploit identified at time of analysis and EPSS is very low (0.06%), but a vendor patch is already available via Red Hat errata.
Flux notification-controller prior to version 1.8.3 fails to validate the email claim in Google OIDC tokens used for Pub/Sub push authentication, allowing any valid Google-issued token to trigger unauthorized reconciliations via the gcr Receiver webhook endpoint. An attacker must know or discover the webhook URL (generated from a random token stored in a Kubernetes Secret) to exploit this vulnerability; however, practical impact is severely limited because Flux reconciliations are idempotent and deduplicated, meaning unauthorized requests result in no operational changes to cluster state unless the underlying Git/OCI/Helm sources have been modified.
Helm versions 3.20.1 and earlier, and 4.1.3 and earlier, allow local attackers with user interaction to write Chart contents to arbitrary directories via path traversal in the helm pull --untar command. A specially crafted Chart will bypass the expected subdirectory naming convention and extract files to the current working directory or a user-specified destination, potentially overwriting existing files. Vendor-released patches are available in versions 3.20.2 and 4.1.4.
Apache Tomcat's cloud membership clustering component logs Kubernetes bearer tokens in plaintext, enabling unauthenticated remote attackers to extract authentication credentials from log files. Affects Tomcat 9.0.13-9.0.116, 10.1.0-M1-10.1.53, and 11.0.0-M1-11.0.20. CVSS 7.5 (High) reflects confidentiality impact; no public exploit identified at time of analysis. Exploitation requires network access to log files or log aggregation systems, potentially enabling privilege escalation within Kubernetes clusters.
Aiven Operator versions 0.31.0 through 0.36.x allow developers with ClickhouseUser CRD creation permissions in their own namespace to exfiltrate secrets from arbitrary namespaces by exploiting a confused deputy vulnerability in the operator's ClusterRole. An attacker can craft a malicious ClickhouseUser resource that causes the operator to read privileged credentials (database passwords, API keys, service tokens) from production namespaces and write them into the attacker's namespace with a single kubectl apply command. The vulnerability is fixed in version 0.37.0.
Helm 4.0.0 through 4.1.3 silently installs Kubernetes plugins without cryptographic provenance verification even when signature verification is explicitly required, allowing local attackers to deliver malicious plugins that execute with Helm's privileges during installation. The flaw (CWE-636: Not Failing Securely) enables supply chain attacks where unsigned or tampered plugins bypass security controls intended to validate plugin integrity. Fixed in Helm 4.1.4. EPSS score is 2nd percentile (0.01% exploitation probability), no active exploitation confirmed, SSVC assessment indicates total technical impact but non-automatable exploitation requiring user interaction.
Path traversal in Helm 4.0.0-4.1.3 allows local attackers to write arbitrary files during plugin installation or update by embedding '/../' sequences in the plugin.yaml version field, achieving high integrity impact across system and vulnerable component scopes. EPSS score is 2nd percentile (0.01%) with no active exploitation or public POC identified, suggesting low immediate risk despite 8.4 CVSS score. Vendor patch released in version 4.1.4.
Container privilege escalation in Red Hat Multicluster Engine for Kubernetes allows authenticated local attackers to escalate from non-root container execution to full root privileges by exploiting group-writable permissions on the /etc/passwd file created during container image build time, enabling arbitrary UID assignment including UID 0.
Denial-of-service and information disclosure in Podman Desktop prior to 1.26.2 stem from an unauthenticated HTTP server that any network attacker can reach without credentials or user interaction. By abusing missing connection limits and timeouts, an attacker exhausts file descriptors and kernel memory to crash the application or freeze the entire host, while verbose error responses leak internal filesystem paths and system details (including Windows usernames). SSVC marks exploitation as proof-of-concept and automatable; publicly available exploit code exists, but EPSS probability is low (0.06%, 19th percentile).
Improper certificate validation in Red Hat's Open Cluster Management (OCM) and Multicluster Engine for Kubernetes allows managed cluster administrators with high-level local access to forge client certificates, achieving cross-cluster privilege escalation to other managed clusters including the hub cluster. The CVSS 8.2 rating reflects high impact across confidentiality, integrity, and availability with scope change, though exploitation requires existing high-privilege local access (PR:H) and local attack vector (AV:L). No public exploit code or CISA KEV listing identified at time of analysis, though technical details are publicly documented in researcher blog post.
Command injection in BentoML's cloud deployment path allows remote code execution on BentoCloud build infrastructure via malicious bentofile.yaml configurations. While commit ce53491 fixed command injection in local Dockerfile generation by adding shlex.quote protection, the cloud deployment code path (deployment.py:1648) remained vulnerable, directly interpolating system_packages into shell commands without sanitization. Attackers can inject shell metacharacters through bentofile.yaml to execut
Microsoft Azure Kubernetes Service (AKS) contains an improper authorization vulnerability enabling unauthenticated remote attackers to elevate privileges over a network with critical impact across confidentiality, integrity, and availability. The CVSS 10.0 critical rating reflects network-accessible exploitation requiring no authentication, low complexity, and scope change allowing compromise beyond the vulnerable component. No public exploit identified at time of analysis, though the authentication bypass nature and maximum severity warrant immediate priority.
Unauthenticated remote attackers can trigger complete database overwrites, server-side file reads, and SSRF attacks against Dgraph graph database servers (v24.x, v25.x prior to v25.3.1) via the admin API's restoreTenant mutation. The mutation bypasses all authentication middleware due to missing authorization configuration, allowing attackers to provide arbitrary backup source URLs (including file:// schemes for local filesystem access), S3/MinIO credentials, Vault configuration paths, and encry
Command injection in KubeAI Ollama model controller allows Kubernetes users with Model CRD write permissions to execute arbitrary shell commands inside model server pods. The vulnerability stems from unsanitized URL components (model ref and query parameters) being interpolated into bash startup probe scripts. With CVSS 8.7 (AV:N/AC:L/PR:H/UI:N/S:C), this represents a significant privilege escalation risk in multi-tenant clusters where Model creation is delegated to non-admin users. No public exploit identified at time of analysis, though detailed proof-of-concept payloads are documented in the GitHub advisory.
Server-Side Request Forgery in PraisonAI's passthrough API allows authenticated remote attackers to access internal cloud metadata services and private network resources. The vulnerability affects the praisonai Python package where the passthrough() and apassthrough() functions accept unvalidated caller-controlled api_base parameters that are directly concatenated and passed to httpx requests. With default AUTH_ENABLED=False configuration, this is remotely exploitable to retrieve EC2 IAM credent
A path traversal vulnerability (CVSS 7.5). High severity vulnerability requiring prompt remediation. Vendor patch is available.
Command injection in Flannel's experimental Extension backend allows authenticated Kubernetes users with node annotation privileges to execute arbitrary commands as root on all flannel nodes in the cluster. This affects Flannel versions prior to 0.28.2 using the Extension backend; other backends (vxlan, wireguard) are unaffected. No public exploit identified at time of analysis, but CVSS 7.5 reflects high impact once node annotation access is achieved. EPSS data not available for this recent CVE (2026 designation appears to be error; actual 2025 advisory).
Cilium Network Policy enforcement is bypassed for traffic from pods to L7 Services with local backends on the same node when Per-Endpoint Routing is enabled and BPF Host Routing is disabled, allowing authenticated local attackers to circumvent ingress network policies and access restricted services. This affects Cilium v1.19.0-v1.19.1, v1.18.0-v1.18.7, and all versions prior to v1.17.13, with the most common vulnerable deployment being Amazon EKS with Cilium ENI mode. Vendor-released patches are available (v1.19.2, v1.18.8, v1.17.14), and no public exploit code has been identified at the time of analysis.
LiquidJS versions 10.24.x and earlier contain a memory limit bypass vulnerability that allows unauthenticated attackers to crash Node.js processes through a single malicious template. By exploiting reverse range expressions to drive the memory counter negative, attackers can allocate unlimited memory and trigger a V8 Fatal error that terminates the entire process, causing complete denial of service. A detailed proof-of-concept exploit is publicly available demonstrating the full attack chain from bypass to process crash.
A SSRF vulnerability (CVSS 6.3) that allows an attacker. Remediation should follow standard vulnerability management procedures.
The Kubernetes NFS CSI Driver fails to properly validate the subDir parameter in volume identifiers, allowing privileged users to inject path traversal sequences that bypass intended directory restrictions. Attackers with PersistentVolume creation privileges can craft malicious volume identifiers to access and modify arbitrary directories on the NFS server during cleanup operations. No patch is currently available for this medium-severity vulnerability affecting Kubernetes environments.
This is an authentication and authorization bypass vulnerability in etcd's gRPC API layer that allows unauthorized users to execute privileged operations when etcd auth is enabled. Affected are etcd versions prior to 3.4.42, 3.5.28, and 3.6.9 (specifically the Go packages go.etcd.io/etcd/v3 and go.etcd.io/etcd). Attackers can enumerate cluster topology via MemberList, trigger denial of service through Alarm APIs, manipulate Lease operations affecting TTL-based keys, and force compaction to permanently delete historical data. Standard Kubernetes deployments are not affected as they do not rely on etcd's built-in authentication. No EPSS score or KEV listing is currently available, and the vulnerability was responsibly disclosed by multiple security researchers.
An authenticated user with restricted RBAC permissions on specific key ranges in etcd can use nested transactions to completely bypass key-level authorization controls and access the entire etcd data store. This affects etcd versions 3.4.x before 3.4.42, 3.5.x before 3.5.28, and 3.6.x before 3.6.9. While Kubernetes deployments are typically protected because Kubernetes handles authentication and authorization at the API server layer rather than relying on etcd's built-in controls, direct etcd deployments with RBAC restrictions are at significant risk.
A configuration injection vulnerability in Kubernetes ingress-nginx controller allows authenticated attackers to inject arbitrary nginx configuration through specially crafted Ingress annotations, leading to remote code execution with controller privileges and exposure of all cluster Secrets. The vulnerability has a high CVSS score of 8.8 and affects the ingress-nginx controller's annotation parsing mechanism. No active exploitation (not in KEV) or public POC has been reported, though the attack requires only low privileges and network access.
Path traversal in Apple and Kubernetes DAG management APIs allows authenticated attackers to access arbitrary files outside the intended directory by injecting URL-encoded forward slashes into file name parameters on GET, DELETE, RENAME, and EXECUTE endpoints. The vulnerability affects systems where a previous patch (CVE-2026-27598) only secured the CREATE endpoint while leaving other API functions unprotected. An attacker with valid credentials can read, modify, or execute unintended DAG files on the affected system.
Budibase, a low-code platform distributed as a Docker/Kubernetes application, contains a Server-Side Request Forgery (SSRF) vulnerability in its REST datasource query preview endpoint. Authenticated admin users can force the server to make HTTP requests to arbitrary URLs including cloud metadata services, internal networks, and Kubernetes APIs. A detailed proof-of-concept exists demonstrating theft of GCP OAuth2 tokens with cloud-platform scope, CouchDB credential extraction, and internal service enumeration. The CVSS score of 8.7 reflects high confidentiality and integrity impact with changed scope, requiring high privileges but low attack complexity.
The Tekton Pipelines git resolver contains a path traversal vulnerability allowing authenticated tenants to read arbitrary files from the resolver pod's filesystem via the pathInRepo parameter. Affected products include github.com/tektoncd/pipeline versions 1.0.0 through 1.10.0 across multiple release branches. The vulnerability enables credential exfiltration and privilege escalation from namespace-scoped access to cluster-wide secret reading capabilities. A proof-of-concept was provided by the vulnerability reporter Oleh Konko.
A denial of service vulnerability (CVSS 6.5). Remediation should follow standard vulnerability management procedures.
The kube-router proxy module fails to validate Service externalIPs and LoadBalancer IPs against configured IP ranges, allowing namespace-scoped users to bind arbitrary VIPs on all cluster nodes and hijack traffic to critical services like kube-dns. This affects all kube-router v2.x versions including v2.7.1, primarily impacting multi-tenant clusters where untrusted users have Service creation permissions. A detailed proof-of-concept demonstrates single-command cluster DNS takedown and arbitrary VIP binding with traffic redirection to attacker-controlled pods, though EPSS scoring is not available for this recently disclosed vulnerability.
CVE-2026-32769 is a security vulnerability. Remediation should follow standard vulnerability management procedures. Vendor patch is available.
CVE-2026-32768 is a security vulnerability. Remediation should follow standard vulnerability management procedures. Vendor patch is available.
A misconfigured NetworkPolicy in Kubernetes deployments allows attackers to perform unauthorized lateral movement between namespaces, breaking namespace isolation security boundaries. This vulnerability affects Kubernetes environments with improperly configured inter-namespace NetworkPolicies, specifically those with 'inter-ns' prefixed policies in monitoring namespaces. An attacker who compromises any component can pivot to access resources in other namespaces, potentially accessing sensitive data or systems they shouldn't have access to.
OneUptime versions prior to 10.0.24 contain a sensitive information exposure vulnerability where the password reset flow logs complete password reset URLs containing plaintext reset tokens at the INFO log level, which is enabled by default in production environments. Any actor with access to application logs-including log aggregation systems, Docker logs, or Kubernetes pod logs-can extract these tokens and perform account takeover on any user. The vulnerability is fixed in version 10.0.24.
Silent event loss in Inspektor Gadget prior to 0.50.1 allows local attackers to cause denial of service by filling the 256KB ring-buffer, which triggers undetected data drops without alerting users or administrators. When the buffer becomes full, gadgets silently discard events and fail to report the loss count, potentially hiding critical system events from Kubernetes cluster and Linux host monitoring. A local attacker with limited privileges can exploit this to obscure malicious activity or system anomalies by saturating the instrumentation buffer.
Authentication bypass in qSnapper's privileged D-Bus service (versions before 1.3.3) allows any local unprivileged user to invoke administrative snapshot operations by piggy-backing on another client's prior Polkit authentication. The service stored authentication state in a single shared m_authenticated flag rather than per-client, so once any caller passed an admin check, all subsequent D-Bus callers inherited that state. No public exploit identified at time of analysis; the issue was found during a SUSE-coordinated security review and patched in v1.3.3.
VM escape in Kata Containers allows any Kubernetes user with pod-creation rights to break out of the VM sandbox and gain full read/write access to the host filesystem. All Kata Containers installations prior to commit ffa59ce3aa78 are affected when using the default configuration.toml, which enables the `virtio_fs_extra_args` and `kernel_params` pod annotations out of the box. An attacker crafts a pod with two annotations: one to redirect virtiofsd to serve the host root filesystem (`/`) into the guest VM, and a second to enable the agent debug console - after which the entire host filesystem is accessible from inside the supposedly isolated VM. A fully working proof-of-concept with confirmed output against Kata Containers 3.28.0 on Ubuntu 24.04 has been publicly disclosed; no public exploit confirmed as actively exploited (CISA KEV) at time of analysis.
Unvalidated jarURI handling in Apache Flink Kubernetes Operator exposes authenticated low-privilege users to server-side request forgery and arbitrary file read primitives against the operator pod. Any Kubernetes principal holding Custom Resource create permissions can submit a malicious FlinkSessionJob CR with a crafted jarURI - pointing to local filesystem paths, internal cloud metadata endpoints, link-local addresses, or any backing store reachable through Flink's pluggable filesystem layer - and retrieve that content via the submitted job. No public exploit has been identified at time of analysis, and EPSS sits at 0.01% (3rd percentile), but the confidentiality impact is rated High by NVD given the breadth of accessible internal resources.
runAsNonRoot bypass in containerd allows crafted container images to execute as UID 0 despite Kubernetes security policies designed to prevent root execution. The flaw stems from containerd treating numeric USER directives that overflow a 32-bit integer as usernames, and if the image's /etc/passwd maps that string to root, the container runs as root. No public exploit identified at time of analysis, but the issue was responsibly disclosed by Lei Wang (@ssst0n3) and fixed in multiple containerd release branches.
Authorization bypass in mcp-server-kubernetes versions prior to 3.6.0 allows authenticated clients to invoke any Kubernetes tool - including destructive operations like kubectl_delete, exec_in_pod, and node_management - regardless of ALLOW_ONLY_READONLY_TOOLS, ALLOW_ONLY_NON_DESTRUCTIVE_TOOLS, or ALLOWED_TOOLS restrictions. The controls were enforced only at the tools/list discovery layer, leaving tools/call unguarded, which effectively reduces operator-configured least-privilege policies to cosmetic filters. Publicly available exploit code exists (a single curl invocation reproduces it), and in cluster-admin deployments the flaw is equivalent to full cluster compromise for any client reaching the endpoint.
Privilege escalation in Fission (Kubernetes serverless framework) versions <= 1.22.0 allows any user able to deploy or update a Function CRD to read every Secret and ConfigMap in the runtime namespace, bypassing the Function.spec.secrets allowlist. The flaw stems from runtime pods running under the namespace-privileged fission-fetcher ServiceAccount whose automounted token is reachable by the user-code container alongside the fetcher sidecar. No public exploit identified at time of analysis, but the vendor-issued advisory (GHSA-85g2-pmrx-r49q) describes a trivially reachable abuse path once a function can be deployed.
Unauthenticated archive CRUD in Fission's storagesvc (≤ v1.22.0) lets any in-cluster workload list, download, replace, or delete function deployment archives across all tenants by hitting the ClusterIP-exposed /v1/archive and /v1/archives endpoints. Because uploaded archives are later fetched and executed by function specialization, the flaw escalates from a tenant data-exposure issue to in-cluster code execution. No public exploit identified at time of analysis, but the trivial HTTP pattern and lack of auth middleware make weaponization straightforward for any attacker with a foothold pod.
Arbitrary code execution in InternLM lmdeploy <= 0.12.3 occurs because trust_remote_code=True is hardcoded across HuggingFace model-loading call sites in lmdeploy/archs.py and lmdeploy/utils.py. An attacker who can influence the model_path passed to an lmdeploy serving process can point it at a malicious HuggingFace repository, causing Transformers to download and execute attacker-controlled Python code with the privileges of the serving daemon. Publicly available exploit code exists in the GHSA advisory, and an upstream fix has been merged via PR #4511 (fixed in 0.13.0).
Cross-namespace privilege escalation in Apache Camel K (2.0.0-2.8.0, 2.9.0-2.9.1, 2.10.0) lets an authenticated Kubernetes user who can create Build resources dictate the target namespace for the resulting Pod, including the privileged operator namespace. By steering Pod generation into the operator's namespace, a tenant confined to their own namespace can break out of RBAC boundaries and interact with higher-privilege resources. CVSS 8.1; no public exploit identified at time of analysis and EPSS is negligible (0.02%), but SSVC rates the technical impact as total.
Privilege escalation in the Apache Airflow CNCF Kubernetes provider (versions before 10.17.0) exposes worker JWT tokens to any principal holding read-only access to Kubernetes Pods. When Airflow runs with the KubernetesExecutor, the scheduler embeds a 'workload'-scoped Execution API JWT in the pod's workload payload, so a low-privileged user who can read pod objects can harvest that token and impersonate a running task via the Task SDK, including actions that mutate Airflow database state. There is no public exploit identified at time of analysis; EPSS is 0.01% (2nd percentile) and CISA SSVC rates exploitation as none, so this is an insider/lateral-movement risk rather than a mass-exploited flaw.
Stored cross-site scripting in Argo CD lets a low-privileged developer escalate to admin by planting a malicious `link.argocd.argoproj.io/*` annotation on any Application; the value is rendered as an unvalidated `<a href>` in the Summary tab's URLs section, so a `javascript:` payload hidden behind a benign 'GitHub Repo' label executes in the admin's authenticated same-origin session when clicked. A detailed public exploit exists (PoC in the vendor advisory demonstrating session exfiltration via Burp Collaborator), though EPSS is very low (0.04%) and there is no public exploit identified as actively used; this is a CVSS 8.7 authenticated, user-interaction-dependent issue affecting Argo CD v2.x, v3.2.x, v3.3.x and v3.4.x. An attacker with `applications` write access can capture the admin's session data and drive Argo CD APIs as admin.
Information disclosure in Argo CD v3.x exposes plaintext Kubernetes Secret values to authenticated users who can view application diffs via the ServerSideDiff feature. This is an incomplete fix for a prior vulnerability (GHSA-3v3m-wc6v-x4x3): the original patch masked top-level Secret data in ServerSideDiff responses but failed to sanitize Secret content embedded in the `kubectl.kubernetes.io/last-applied-configuration` annotation on `predictedLive` objects, leaving raw `data`, `stringData`, and sensitive annotation values readable in UI and CLI diff output. A publicly available proof-of-concept exists; no KEV listing is present at time of analysis, but the Changed Scope (S:C) in the CVSS vector indicates that exposed secrets may belong to workloads beyond the Argo CD application boundary, amplifying real-world impact in multi-tenant environments.
Local file disclosure in NiceGUI versions <= 3.11.1 allows remote unauthenticated attackers to read arbitrary files accessible to the server process when applications pass user-controlled content to ui.restructured_text(). The flaw stems from Docutils being invoked without disabling file-insertion directives (include, csv-table :file:, raw :file:), enabling exfiltration of secrets, credentials, and source code. No public exploit identified at time of analysis, but the vendor advisory provides full directive-level proof patterns.
Open WebUI versions ≤0.9.4 allow authenticated users to bypass SSRF protections and read internal network resources via HTTP redirect-following. The vulnerability exists in five distinct code paths: web-fetch retrieval, image-loading endpoints, chat-completion image inlining, and OAuth/tool-server execution flows. Any authenticated user can submit a public URL that 302-redirects to internal addresses (127.0.0.1, 169.254.169.254 cloud metadata, RFC1918 private networks) and receive the response body. The vendor confirms active exploitation in the wild was NOT observed, but publicly available exploit code exists (PoC in advisory) and EPSS score of 0.00043 (4.3%) suggests low automated scanning targeting at time of analysis. Fixed in version 0.9.5 released March 2025.
Kuma Control Plane with default configuration leaks admin bootstrap tokens and signing keys to any website an operator visits if the control plane is reachable from their browser. The vulnerability combines default CORS settings allowing all origins (CorsAllowedDomains: [".*"]) with LocalhostIsAdmin: true, which grants admin privileges to any request from 127.0.0.1 without validating whether it originates from a trusted same-origin context. An attacker's JavaScript on a visited webpage can cross-origin fetch the admin token and cryptographic material via browser requests to localhost:5681. This is not actively exploited in the wild but represents a realistic threat in developer and testing environments where control planes run on workstations with web browsers.
Token leakage in Portainer's authentication middleware allows JWT bearer tokens passed via the `?token=<JWT>` URL query parameter to be harvested from reverse-proxy access logs, browser history, and HTTP Referer headers, enabling account takeover for the validity window of the token (default 8 hours). The flaw affected any user with container exec/attach rights - not just administrators - and a leaked admin token grants full control of Portainer and every managed Docker/Kubernetes environment. No public exploit identified at time of analysis, though the underlying behavior was present since JWT auth was introduced and the GitHub Security Advisory provides sufficient detail to weaponize.
Authorization bypass in Portainer 2.33.0 through 2.33.7 allows authenticated users to access Kubernetes cluster resources beyond their assigned permissions due to missing return statement in middleware error handling. Any user with a valid Portainer session can exploit this to read or modify Kubernetes secrets, pods, deployments, and other resources on endpoints they should not access. The flaw affects both Community Edition and Enterprise Edition. Fixed in version 2.33.8 and inherently absent from 2.39.0+. No public exploit code identified at time of analysis, though the single-line code fix and detailed GitHub advisory make reproduction straightforward for authenticated attackers.
Arbitrary file read in Portainer Community Edition allows authenticated low-privileged users to exfiltrate any file readable by the Portainer process — commonly root — by submitting a Git-backed stack whose docker-compose.yml is a symlink to a target like /etc/shadow or a Kubernetes service account token. The flaw stems from go-git v5 materializing Git symlink blobs as real OS symlinks during checkout combined with GetFileContent reading the entry point through os.ReadFile without resolving links, and is amplified by Portainer's scheduled stack auto-update, which lets a previously benign repository turn malicious after initial review. A fully validated proof-of-concept exists from the reporter (b-hermes), but there is no public exploit identified at time of analysis and no CISA KEV listing.
Stored Cross-Site Scripting (XSS) in vCluster Platform allows authenticated attackers with namespace creation privileges to inject malicious scripts via the templateRef name field, potentially escalating to Global-Admin access. The vulnerability affects versions prior to 4.4.3, 4.5.5, 4.6.2, 4.7.1, and 4.8.0. With a CVSS score of 9.0 and changed scope (S:C), this represents a critical privilege escalation path in Kubernetes multi-tenancy environments. No active exploitation confirmed by CISA KEV at time of analysis, but the low attack complexity (AC:L) and clear attack path make this an immediate patching priority for organizations running vCluster Platform.
Unauthorized write access to Traefik's live dynamic configuration is achievable by a low-privileged Kubernetes tenant in shared Gateway deployments, because the Gateway API provider's `isInternalService()` check accepts any `TraefikService` name ending in `@internal` - not exclusively the intended `api@internal`. This allows a tenant with `HTTPRoute` creation rights to route external traffic to `rest@internal`, fully bypassing the `providers.rest.insecure=false` safeguard and gaining `PUT /api/providers/rest` write access. No active exploitation is confirmed (not in CISA KEV, EPSS at 0.01%), but a detailed proof-of-concept is published in the GHSA advisory and the SSVC framework marks the attack as automatable for tenants meeting the prerequisites.
Sandbox escape in OpenClaude (npm package openclaude) versions before 0.5.1 allows a prompt-injected LLM to disable host sandboxing by setting the model-controlled `dangerouslyDisableSandbox: true` flag in any Bash tool_use call, yielding full unsandboxed command execution on the host. CVSS 4.0 scores this 9.3 Critical (AV:N/AC:L/PR:N/UI:N, VC/VI/VA:H); no public exploit identified at time of analysis beyond the reporter's PoC, but the upstream fix has been merged. The flaw is especially severe because it is reachable under default settings (`allowUnsandboxedCommands` defaults to true).
Credential-harvesting malware compromised 84 versions of 42 TanStack npm packages on 2026-05-11 via chained GitHub Actions exploitation. Attackers combined pull_request_target misconfiguration, Actions cache poisoning, and OIDC token memory extraction to publish malicious code under the legitimate TanStack identity. Installing any affected version executes a 2.3 MB obfuscated payload that exfiltrates AWS/GCP/Kubernetes credentials, npm tokens, GitHub secrets, SSH keys, and HashiCorp Vault tokens over encrypted Session/Oxen messenger infrastructure. The payload propagates by republishing victim-maintained packages with identical injection. Socket.dev and the TanStack team confirmed the incident via GHSA-g7cv-rxg3-hmpx. No EPSS or CISA KEV data available for this recent supply-chain attack. CVSS 9.6 reflects the cross-scope credential theft impact (S:C/C:H/I:H), though exploitation requires user-initiated package installation (UI:R).
Template injection in Rancher Local Path Provisioner allows Kubernetes cluster operators with ConfigMap edit permissions to escalate privileges to node-level root access. Attackers with write access to the local-path-config ConfigMap can inject malicious Pod templates that bypass security controls, creating privileged containers with full host filesystem access. This enables theft of ServiceAccount tokens from co-located pods, access to other tenants' persistent volume data, and arbitrary modification of host node files. Vendor-released patch: v0.0.36. CVSS 8.7 (High) reflects the high-privilege prerequisite (PR:H) but scope change to container escape (S:C). No public exploit identified at time of analysis, though exploitation is straightforward for authenticated cluster operators.
External Secrets Operator versions 0.1.0 through 2.4.0 allow authenticated users with ExternalSecret creation permissions to escalate privileges by crafting Service Account token templates that cause the operator to generate long-lived tokens for any service account in the namespace. An attacker can impersonate service accounts without requiring direct TokenRequest or Secret creation permissions, effectively bypassing RBAC controls. The attack requires the attacker already has ExternalSecret creation permissions and the cluster must have service account token generation enabled, limiting the practical scope to already-privileged users seeking lateral privilege expansion within a namespace.
Server-side request forgery in MCP Registry's HTTP namespace verification endpoint allows unauthenticated attackers to reach internal IPv4 addresses via specially-crafted IPv6 addresses that encode or tunnel to RFC1918 and cloud-metadata services. The vulnerability exists in the private-address blocklist used by `safeDialContext`, which fails to block IPv6 6to4 (2002::/16), NAT64 well-known (64:ff9b::/96), NAT64 local-use (64:ff9b:1::/48), and deprecated site-local (fec0::/10) prefixes. On dual-stack and IPv6-only cloud deployments (GKE IPv6, AWS IPv6-only EC2, Azure NAT64), this enables direct connections to metadata services and internal Kubernetes API servers. No public exploit code identified at time of analysis, but proof-of-concept has been demonstrated against the production registry.
Server-Side Request Forgery (SSRF) in nuxt-og-image 6.2.5 through 6.4.8 allows remote attackers to bypass the incomplete IPv6 denylist and redirect validation, reaching internal IP addresses and services through incomplete IPv6 prefix filtering and unauthenticated HTTP redirect following. The vulnerability affects the OG image rendering component used by Nuxt applications, enabling attackers to leak internal service responses by injecting crafted IPv6-mapped addresses or chaining external redirects to internal targets.
Denial-of-service in vm2 Node.js sandbox allows unauthenticated remote attackers to crash host processes via unbounded Buffer.alloc() calls. The vm2 library's timeout mechanism cannot interrupt synchronous C++ native calls, enabling attackers to bypass configured timeout limits and exhaust host heap memory with a single HTTP request. Version 3.11.0 patches this flaw by introducing bufferAllocLimit controls. Publicly available exploit code exists (GHSA-6785-pvv7-mvg7 includes working POC), and while EPSS data is unavailable and the vulnerability is not listed in CISA KEV, the vendor-confirmed POC demonstrates reliable exploitation against default configurations.
Remote unauthenticated attackers can crash Node.js processes running vm2 <= 3.10.5 by triggering an unhandled Promise rejection that terminates the host application. The vulnerability exploits an incomplete fix for CVE-2026-22709 - while previous patches sanitized `.then()` and `.catch()` callback chains, they failed to intercept unhandled rejections originating from Promise constructor executors. Publicly available exploit code exists (GitHub advisory GHSA-hw58-p9xv-2mjh). The attack requires minimal resources (150-byte HTTP request) but achieves high impact by crashing entire server processes serving all concurrent users, with demonstrated persistent DoS despite container orchestration restart policies.
etcd RBAC authorization bypass allows authenticated users to read unauthorized data or attach leases via PrevKv or lease attachment features in transaction Put requests, circumventing role-based access control checks. Affects etcd 3.4.x through 3.6.x before patched versions 3.4.44, 3.5.30, and 3.6.11. While Kubernetes deployments are typically not affected because the API server handles its own authorization, etcd deployments with reliance on etcd's built-in RBAC-particularly those managed directly or used outside Kubernetes-face exposure to privilege escalation and unauthorized data access by already-authenticated users.
Kubetail Dashboard prior to version 0.14.0 fails to validate the Origin header on WebSocket connection upgrades, enabling Cross-Site WebSocket Hijacking (CSWSH) attacks. An authenticated user visiting a malicious web page can be exploited to stream their Kubernetes container logs-including credentials, tokens, and PII often present in logs-to an attacker-controlled server. The vulnerability affects both desktop deployments at localhost:7500 and cluster deployments behind HTTP basic auth, with browser ambient credentials automatically attached to the WebSocket handshake.
Kubernetes Secret extraction in Argo CD v3.2.0-3.2.10 and v3.3.0-3.3.8 allows authenticated users with read-only application permissions to retrieve plaintext credential data including service account tokens, TLS certificates, database passwords, and API keys via the ServerSideDiff endpoint. The vulnerability exists due to missing data masking in the gRPC/REST ServerSideDiff function, which returns raw Kubernetes Server-Side Apply dry-run responses containing unredacted Secret values from etcd when applications are annotated with 'IncludeMutationWebhook=true'. A functional proof-of-concept exploit exists demonstrating automated extraction of all accessible secrets. Vendor-released patches (3.2.11, 3.3.9) are available. CVSS 9.6 reflects network-exploitable, low-complexity attack requiring only low-privilege authenticated access with cross-scope high confidentiality/integrity impact.
ServiceAccount impersonation bypass in Rancher Fleet allows tenants with git push access to multi-tenant clusters to read secrets from any namespace across all downstream clusters. Two distinct code paths failed to properly apply RBAC constraints: Helm's lookup function executed with cluster-admin credentials instead of the impersonated ServiceAccount, and valuesFrom secret references in fleet.yaml bypassed namespace isolation. Confirmed active exploitation status unknown (not in CISA KEV). CVSS 9.9 with scope-change modifier reflects potential credential leakage to external services. Fleet versions 0.12.0 through 0.15.0 affected across multiple Rancher release branches. Patches available for all supported versions with detailed version matrix provided by SUSE.
Unauthenticated remote attackers crash Gotenberg 8.x (≤ 8.31.0) by triggering a race condition between webhook goroutine context reuse and Echo framework connection pooling. When webhook middleware spawns an async goroutine holding an `echo.Context` reference, the synchronous handler returns immediately, recycling the context to Echo's `sync.Pool`. Concurrent requests reset the pooled context, causing unchecked type assertions in the still-running webhook goroutine to panic outside any `recover()` scope, terminating the process with exit code 2. Twenty-four webhook requests plus sixty concurrent GET requests demonstrate reliable two-second crash windows. No patch was available at initial disclosure; upstream commit fixes the panic in version 8.32.0. CVSS 7.5 (AV:N/AC:L/PR:N/UI:N) reflects trivial unauthenticated network exploitation producing complete service disruption.
Stored XSS in Kyverno policy-reporter-ui 2.5.1 and earlier allows an attacker with Kubernetes write access to PolicyReport resources to inject arbitrary HTML into the Vue 3 frontend, which executes in the browser of any authenticated UI user who views the affected namespace. The sink is the v-html directive in PropertyCard.vue, which is reached whenever a property value is not an http/https URL and exceeds 75 characters - a bypass condition trivially satisfied by padding any HTML payload. A public proof-of-concept demonstrating session cookie exfiltration via a crafted PolicyReport object is included in the GitHub advisory GHSA-q98m-7w8c-w388; no public exploit identified as confirmed actively exploited (not in CISA KEV), though EPSS of 0.03% reflects low opportunistic risk given the Kubernetes-specific prerequisite.
Kubewarden versions before 1.35.0 permit RBAC reconnaissance attacks when users with AdmissionPolicy or AdmissionPolicyGroup creation privileges craft policies using the unchecked `can_i` host capability. The vulnerability allows enumeration of any user or service account permissions across the cluster via SubjectAccessReview requests executed with policy-server privileges, despite the absence of context-aware resource grants. This information disclosure enables attackers to discover sensitive permission configurations without requiring cluster-wide policy creation rights, a capability not available by default but exploitable when granted.
Path traversal in MinIO's ReadMultiple internode storage-REST endpoint allows authenticated cluster peers or root-credential holders to read arbitrary files from the host filesystem outside configured drive roots. Distributed-erasure (multi-node) deployments are affected; single-node standalone deployments are not. The vulnerability exists in all releases from RELEASE.2022-07-24T01-54-52Z through RELEASE.2025-09-07T16-13-09Z and has been fixed as of MinIO AIStor RELEASE.2024-10-23T19-38-07Z (with security patch RELEASE.2026-04-14T21-32-45Z recommended). No public exploit code or active exploitation has been identified at time of analysis.
Argo Workflows executor logs artifact repository credentials in plaintext to pod logs during artifact operations, exposing S3 access/secret keys, GCS service account keys, Azure storage keys, and Git passwords. Users with Kubernetes RBAC permissions to read pod logs in the workflow namespace can extract these credentials directly from workflow execution logs. This vulnerability affects Argo Workflows v4.0.0 through v4.0.4 and represents an incomplete fix of CVE-2025-62157. Vendor-released patch (v4.0.5) is available with GitHub commit bdd40908 removing credential-bearing struct logging. No public exploit identified at time of analysis, though exploitation is trivial given the included working proof-of-concept YAML.
Argo Workflows v3 (< 3.7.14) and v4 (< 4.0.5) allow users to bypass templateReferencing Strict/Secure mode restrictions by setting WorkflowSpec fields like hostNetwork, serviceAccountName, securityContext, tolerations, and volumes. The incomplete fix for CVE-2026-31892 only blocked podSpecPatch but left other security-sensitive fields unvalidated. Authenticated users with create Workflow permission can inject host network access, switch service accounts, modify pod security contexts, or schedule on control-plane nodes despite referencing hardened WorkflowTemplates. Vendor-released patch: v3.7.14 and v4.0.5 (commit 2727f3f). No public exploit identified at time of analysis, but exploitation is straightforward given detailed reproduction steps in the advisory.
Memory exhaustion crashes Argo Workflows server via unauthenticated multi-gigabyte webhook requests to the publicly accessible `/api/v1/events/` endpoint. The Webhook Interceptor's `io.ReadAll` call allocates unbounded memory before signature verification, enabling remote attackers to trigger out-of-memory (OOM) conditions without authentication or credentials. Vendor-released patches enforce a 2MB body size limit via `io.LimitReader`. Publicly available exploit code exists (conceptual proof-of-concept published in GitHub security advisory GHSA-jcc8-g2q4-9fxq). No active exploitation confirmed by CISA KEV at time of analysis, but the low attack complexity (CVSS AV:N/AC:L/PR:N) and public PoC create immediate risk for internet-exposed Argo Workflows deployments.
Missing authorization checks in Argo Workflows v4.0.0-4.0.4 allow any authenticated user-even those with fake Bearer tokens-to create, read, update, and delete Kubernetes ConfigMaps containing workflow synchronization limits. The ConfigMap-backed sync provider (server/sync/sync_cm.go) completely omits auth.CanI permission validation on all four CRUD endpoints. Publicly available exploit code exists (detailed PoC in advisory). CVSS 8.5 reflects network-accessible authentication bypass enabling high integrity/availability impact through denial-of-service and arbitrary ConfigMap manipulation. Patch released in version 4.0.5 adding checkConfigMapPermission() calls to validate Kubernetes RBAC before operations.
Authenticated users with low privileges can read cleartext Kubernetes Secret data through Argo CD's ServerSideDiff feature in versions 3.2.0-3.2.10 and 3.3.0-3.3.8. This scope-changing vulnerability (CVSS:3.1 S:C) allows attackers to access sensitive credential data managed by Kubernetes, including database passwords, API tokens, and certificates, by exploiting the server-side diff functionality. With a 7.7 CVSS score and low attack complexity (AC:L), this represents a significant confidentiality breach requiring only network access and basic authentication-no public exploit identified at time of analysis, but the technical barrier to exploitation is minimal.
Traefik versions prior to 2.11.43, 3.6.14, and 3.7.0-rc.2 fail to enforce cross-namespace isolation for middleware references nested inside Chain middlewares, allowing actors with permission to create CRDs in their own namespace to bypass the allowCrossNamespace=false restriction and apply middleware from arbitrary namespaces. This authorization bypass affects Kubernetes clusters relying on namespace isolation controls and can enable unauthorized reuse of security-sensitive middleware policies across namespace boundaries.
Cross-namespace privilege escalation in Kyverno 1.17.x allows authenticated namespace administrators to bypass RBAC controls and read ConfigMaps from any Kubernetes namespace. The vulnerability exploits unvalidated `configMap.namespace` field in Kyverno's ConfigMap context loader, enabling attackers to leverage Kyverno's privileged service account permissions. This is a regression following incomplete fix for CVE-2026-22039, which addressed the same issue in `apiCall` context but missed the ConfigMap loader. Patch available in version 1.17.2. CVSS 7.7 with Changed Scope indicates significant multi-tenant cluster risk; EPSS data not available but the regression nature and RBAC bypass impact warrant immediate patching in multi-tenant environments.
Lua code injection in Contour (Kubernetes ingress controller for Envoy) from v1.19.0 through the fixes in v1.31.6/v1.32.5/v1.33.4 lets a low-privileged tenant achieve arbitrary code execution inside the shared Envoy proxy. A user with RBAC rights to create or modify HTTPProxy resources injects Lua via the cookie-rewrite pathRewrite.value field; although the code only runs on the attacker's own route, it executes in shared infrastructure and can steal Envoy's xDS credentials (exposing other tenants' TLS keys) or cause denial of service for co-tenants. No public exploit identified at time of analysis, and EPSS is low (0.07%, 22nd percentile), consistent with the requirement for pre-existing cluster RBAC access.
Denial of service in the Argo Workflows controller (versions 3.6.5 through 4.0.4, plus the 3.7.0-rc1 through 3.7.12 pre-releases) lets any user able to submit a workflow crash-loop the entire controller by attaching a malformed workflows.argoproj.io/pod-gc-strategy pod annotation that contains no '/' separator, triggering an out-of-range index panic in an informer goroutine that sits outside the controller's recover() scope. Because the poisoned pod persists across restarts, the controller enters CrashLoopBackOff and halts all workflow processing cluster-wide until the pod is manually deleted. A working proof-of-concept manifest is published in the GHSA advisory (SSVC exploitation status: poc), but there is no evidence of active exploitation and EPSS is very low at 0.04%.
Cache poisoning in @astrojs/node versions 9.4.4 and earlier allows unauthenticated remote attackers to poison CDN caches by sending malformed if-match headers to static asset endpoints, causing the server to return 500 errors with immutable one-year cache directives instead of the correct 412 Precondition Failed response. This vulnerability affects all subsequent requests to poisoned assets until the cache expires, breaking application functionality for legitimate users. The vulnerability is not actively exploited in the wild, but proof-of-concept exploitation is straightforward and requires only a single crafted HTTP request.
VolumeMount path restriction bypass in Tekton Pipelines prior to v1.11.1 permits authenticated users holding Task or TaskRun creation permissions to mount volumes over internal /tekton/ paths using dot-dot traversal sequences, violating Tekton's internal execution isolation boundary. The flaw exists in both the v1 and v1beta1 API validation layers (pkg/apis/pipeline/v1/container_validation.go and pkg/apis/pipeline/v1beta1/task_validation.go), where strings.HasPrefix is applied to the raw, un-normalized mount path instead of the filepath.Clean-resolved form. No public exploit code has been identified and SSVC assessment confirms no active exploitation; however, in shared or multi-tenant Kubernetes CI/CD environments the integrity risk is meaningful - an attacker can inject falsified task results trusted by downstream pipelines, alter step scripts before execution, or disrupt entrypoint coordination state.
Unbounded memory allocation in Tekton Pipelines' HTTP resolver allows any authenticated tenant to crash the shared resolver pod for an entire Kubernetes cluster. The `FetchHttpResource` function reads HTTP response bodies without a byte-level size cap, so a low-privileged user who can create TaskRuns or PipelineRuns can direct the resolver at an attacker-controlled server delivering multi-gigabyte responses, exhausting the pod's 4 GiB memory limit and triggering an OOM-kill. Because all resolver types - Git, Hub, Bundle, Cluster, and HTTP - share the same `tekton-pipelines-resolvers` pod, repeated exploitation causes a sustained crash loop that denies CI/CD pipeline resolution to every tenant on the cluster. A PoC is included in the GHSA advisory and SSVC confirms exploit code exists; however, no CISA KEV listing indicates no confirmed widespread exploitation at time of analysis.
Argument injection in the Tekton Pipelines git resolver (versions 1.0.0 through 1.11.0) lets a multi-tenant CI/CD user achieve remote code execution on the resolver pod by supplying a crafted 'revision' parameter. Because the revision string is passed unvalidated as a positional argument to 'git fetch', an attacker can smuggle git flags such as '--upload-pack=<binary>'; combined with validateRepoURL permitting local filesystem paths (URLs beginning with '/'), this executes an arbitrary binary. Since the tekton-pipelines-resolvers ServiceAccount has cluster-wide read on all Secrets, this escalates to full cluster secret exfiltration; a proof of concept exists per CISA SSVC, though the EPSS probability is very low (0.07%).
Broken access control in Esri Portal for ArcGIS 11.4, 11.5, and 12.0 (Windows, Linux, and Kubernetes) fails to correctly validate the permissions bound to developer credentials, allowing a holder of such credentials to access resources or perform actions beyond their intended authorization. The vendor rates it CVSS 9.8 (network, no privileges, no interaction), though the description's reference to 'developer credentials' implies some prior credential is involved. There is no public exploit identified at time of analysis and EPSS is very low (0.04%, 13th percentile), but SSVC flags the technical impact as total and the flaw as automatable.
Credential leakage in Tekton Pipelines git resolver allows authenticated users to exfiltrate system-configured Git API tokens (GitHub PAT, GitLab tokens) by directing the resolver to attacker-controlled endpoints. Affects versions 1.0.0 through 1.10.0 when users omit the token parameter in TaskRun or PipelineRun configurations. CVSS 7.7 with scope change reflects cross-tenant credential theft potential in multi-tenant Kubernetes environments. No active exploitation confirmed (not in CISA KEV), but exploitation is straightforward for authenticated cluster users with TaskRun/PipelineRun creation privileges.
Tekton Pipelines 0.43.0 through 1.11.0 allows authenticated attackers to bypass trusted resource verification policies via unanchored regular expression patterns that match substrings rather than exact resource sources, enabling policy manipulation and unauthorized verification mode changes. The vulnerability stems from Go's regexp.MatchString function matching patterns anywhere within a string rather than requiring full anchoring, permitting attackers to craft source URIs containing trusted patterns as substrings to trigger unintended policy matches and potentially apply weaker verification keys or modes.
Unauthenticated remote denial-of-service in ChilliCream's Hot Chocolate GraphQL server for .NET (HotChocolate.Language package) lets any client crash the entire worker process by sending a small, deeply nested GraphQL document. Because the recursive-descent Utf8GraphQLParser has no depth limit, ~40 KB of nested selection sets, object/list values, or list types triggers an uncatchable StackOverflowException that terminates the process, dropping all in-flight HTTP requests, IHostedService background tasks, and WebSocket subscriptions on that worker. No public exploit identified at time of analysis and EPSS is low (0.09%), but the attack is trivial (AC:L, PR:N) and CISA SSVC flags it as automatable, so it is a practical availability threat to internet-facing GraphQL endpoints.
Race condition in pyLoad's Flask session cookie handler allows unauthenticated attackers to manipulate the SESSION_COOKIE_SECURE flag globally across all concurrent requests by spoofing the X-Forwarded-Proto header. On deployments behind a TLS-terminating proxy, this enables session cookie downgrade attacks resulting in plaintext cookie transmission; on default plain HTTP deployments, it causes session denial of service by forcing the Secure flag and breaking all concurrent user sessions. The vulnerability requires no authentication and exploits a multi-threaded race window in the Cheroot WSGI server (request_queue_size=512) combined with missing proxy origin validation (acknowledged TODO in code).
Command injection in mcp-server-kubernetes port_forward function allows authenticated network attackers to expose internal Kubernetes services to external networks or bypass namespace restrictions. The vulnerability (CVSS 8.3) stems from unsafe string concatenation and space-splitting of kubectl arguments, enabling arbitrary flag injection via fields like resourceName or namespace. Attackers can inject '--address=0.0.0.0' to bind port-forwards on all network interfaces, exposing databases and internal APIs beyond localhost. Affects mcp-server-kubernetes <= 3.4.0. No public exploit identified at time of analysis, though exploitation requires only low complexity (AC:L) with authenticated access (PR:L).
Kyverno's apiCall service helper automatically injects the controller's ServiceAccount token into HTTP requests when ClusterPolicy or GlobalContextEntry authors omit an Authorization header, enabling token exfiltration to attacker-controlled endpoints via confused deputy vulnerability. Affects deployments where policy YAML repositories are compromised (GitOps threat model) or ClusterPolicy creation is possible. CVSS 8.1 (High) reflects network attack vector with low complexity and low privileges required. No CISA KEV listing or public exploit identified at time of analysis, but GitHub advisory includes working proof-of-concept demonstrating token injection and exfiltration.
Kubernetes Service Account token disclosure in the odh-dashboard component of Red Hat OpenShift AI (RHOAI) lets an authenticated low-privileged user retrieve SA tokens via an exposed NodeJS endpoint, then reuse them to reach Kubernetes resources beyond the dashboard's intended scope. Rated CVSS 9.9 with a changed scope, the flaw effectively converts limited dashboard access into broad cluster access. There is no public exploit identified at time of analysis and EPSS is very low (0.06%), but a vendor patch is already available via Red Hat errata.
Flux notification-controller prior to version 1.8.3 fails to validate the email claim in Google OIDC tokens used for Pub/Sub push authentication, allowing any valid Google-issued token to trigger unauthorized reconciliations via the gcr Receiver webhook endpoint. An attacker must know or discover the webhook URL (generated from a random token stored in a Kubernetes Secret) to exploit this vulnerability; however, practical impact is severely limited because Flux reconciliations are idempotent and deduplicated, meaning unauthorized requests result in no operational changes to cluster state unless the underlying Git/OCI/Helm sources have been modified.
Helm versions 3.20.1 and earlier, and 4.1.3 and earlier, allow local attackers with user interaction to write Chart contents to arbitrary directories via path traversal in the helm pull --untar command. A specially crafted Chart will bypass the expected subdirectory naming convention and extract files to the current working directory or a user-specified destination, potentially overwriting existing files. Vendor-released patches are available in versions 3.20.2 and 4.1.4.
Apache Tomcat's cloud membership clustering component logs Kubernetes bearer tokens in plaintext, enabling unauthenticated remote attackers to extract authentication credentials from log files. Affects Tomcat 9.0.13-9.0.116, 10.1.0-M1-10.1.53, and 11.0.0-M1-11.0.20. CVSS 7.5 (High) reflects confidentiality impact; no public exploit identified at time of analysis. Exploitation requires network access to log files or log aggregation systems, potentially enabling privilege escalation within Kubernetes clusters.
Aiven Operator versions 0.31.0 through 0.36.x allow developers with ClickhouseUser CRD creation permissions in their own namespace to exfiltrate secrets from arbitrary namespaces by exploiting a confused deputy vulnerability in the operator's ClusterRole. An attacker can craft a malicious ClickhouseUser resource that causes the operator to read privileged credentials (database passwords, API keys, service tokens) from production namespaces and write them into the attacker's namespace with a single kubectl apply command. The vulnerability is fixed in version 0.37.0.
Helm 4.0.0 through 4.1.3 silently installs Kubernetes plugins without cryptographic provenance verification even when signature verification is explicitly required, allowing local attackers to deliver malicious plugins that execute with Helm's privileges during installation. The flaw (CWE-636: Not Failing Securely) enables supply chain attacks where unsigned or tampered plugins bypass security controls intended to validate plugin integrity. Fixed in Helm 4.1.4. EPSS score is 2nd percentile (0.01% exploitation probability), no active exploitation confirmed, SSVC assessment indicates total technical impact but non-automatable exploitation requiring user interaction.
Path traversal in Helm 4.0.0-4.1.3 allows local attackers to write arbitrary files during plugin installation or update by embedding '/../' sequences in the plugin.yaml version field, achieving high integrity impact across system and vulnerable component scopes. EPSS score is 2nd percentile (0.01%) with no active exploitation or public POC identified, suggesting low immediate risk despite 8.4 CVSS score. Vendor patch released in version 4.1.4.
Container privilege escalation in Red Hat Multicluster Engine for Kubernetes allows authenticated local attackers to escalate from non-root container execution to full root privileges by exploiting group-writable permissions on the /etc/passwd file created during container image build time, enabling arbitrary UID assignment including UID 0.
Denial-of-service and information disclosure in Podman Desktop prior to 1.26.2 stem from an unauthenticated HTTP server that any network attacker can reach without credentials or user interaction. By abusing missing connection limits and timeouts, an attacker exhausts file descriptors and kernel memory to crash the application or freeze the entire host, while verbose error responses leak internal filesystem paths and system details (including Windows usernames). SSVC marks exploitation as proof-of-concept and automatable; publicly available exploit code exists, but EPSS probability is low (0.06%, 19th percentile).
Improper certificate validation in Red Hat's Open Cluster Management (OCM) and Multicluster Engine for Kubernetes allows managed cluster administrators with high-level local access to forge client certificates, achieving cross-cluster privilege escalation to other managed clusters including the hub cluster. The CVSS 8.2 rating reflects high impact across confidentiality, integrity, and availability with scope change, though exploitation requires existing high-privilege local access (PR:H) and local attack vector (AV:L). No public exploit code or CISA KEV listing identified at time of analysis, though technical details are publicly documented in researcher blog post.
Command injection in BentoML's cloud deployment path allows remote code execution on BentoCloud build infrastructure via malicious bentofile.yaml configurations. While commit ce53491 fixed command injection in local Dockerfile generation by adding shlex.quote protection, the cloud deployment code path (deployment.py:1648) remained vulnerable, directly interpolating system_packages into shell commands without sanitization. Attackers can inject shell metacharacters through bentofile.yaml to execut
Microsoft Azure Kubernetes Service (AKS) contains an improper authorization vulnerability enabling unauthenticated remote attackers to elevate privileges over a network with critical impact across confidentiality, integrity, and availability. The CVSS 10.0 critical rating reflects network-accessible exploitation requiring no authentication, low complexity, and scope change allowing compromise beyond the vulnerable component. No public exploit identified at time of analysis, though the authentication bypass nature and maximum severity warrant immediate priority.
Unauthenticated remote attackers can trigger complete database overwrites, server-side file reads, and SSRF attacks against Dgraph graph database servers (v24.x, v25.x prior to v25.3.1) via the admin API's restoreTenant mutation. The mutation bypasses all authentication middleware due to missing authorization configuration, allowing attackers to provide arbitrary backup source URLs (including file:// schemes for local filesystem access), S3/MinIO credentials, Vault configuration paths, and encry
Command injection in KubeAI Ollama model controller allows Kubernetes users with Model CRD write permissions to execute arbitrary shell commands inside model server pods. The vulnerability stems from unsanitized URL components (model ref and query parameters) being interpolated into bash startup probe scripts. With CVSS 8.7 (AV:N/AC:L/PR:H/UI:N/S:C), this represents a significant privilege escalation risk in multi-tenant clusters where Model creation is delegated to non-admin users. No public exploit identified at time of analysis, though detailed proof-of-concept payloads are documented in the GitHub advisory.
Server-Side Request Forgery in PraisonAI's passthrough API allows authenticated remote attackers to access internal cloud metadata services and private network resources. The vulnerability affects the praisonai Python package where the passthrough() and apassthrough() functions accept unvalidated caller-controlled api_base parameters that are directly concatenated and passed to httpx requests. With default AUTH_ENABLED=False configuration, this is remotely exploitable to retrieve EC2 IAM credent
A path traversal vulnerability (CVSS 7.5). High severity vulnerability requiring prompt remediation. Vendor patch is available.
Command injection in Flannel's experimental Extension backend allows authenticated Kubernetes users with node annotation privileges to execute arbitrary commands as root on all flannel nodes in the cluster. This affects Flannel versions prior to 0.28.2 using the Extension backend; other backends (vxlan, wireguard) are unaffected. No public exploit identified at time of analysis, but CVSS 7.5 reflects high impact once node annotation access is achieved. EPSS data not available for this recent CVE (2026 designation appears to be error; actual 2025 advisory).
Cilium Network Policy enforcement is bypassed for traffic from pods to L7 Services with local backends on the same node when Per-Endpoint Routing is enabled and BPF Host Routing is disabled, allowing authenticated local attackers to circumvent ingress network policies and access restricted services. This affects Cilium v1.19.0-v1.19.1, v1.18.0-v1.18.7, and all versions prior to v1.17.13, with the most common vulnerable deployment being Amazon EKS with Cilium ENI mode. Vendor-released patches are available (v1.19.2, v1.18.8, v1.17.14), and no public exploit code has been identified at the time of analysis.
LiquidJS versions 10.24.x and earlier contain a memory limit bypass vulnerability that allows unauthenticated attackers to crash Node.js processes through a single malicious template. By exploiting reverse range expressions to drive the memory counter negative, attackers can allocate unlimited memory and trigger a V8 Fatal error that terminates the entire process, causing complete denial of service. A detailed proof-of-concept exploit is publicly available demonstrating the full attack chain from bypass to process crash.
A SSRF vulnerability (CVSS 6.3) that allows an attacker. Remediation should follow standard vulnerability management procedures.
The Kubernetes NFS CSI Driver fails to properly validate the subDir parameter in volume identifiers, allowing privileged users to inject path traversal sequences that bypass intended directory restrictions. Attackers with PersistentVolume creation privileges can craft malicious volume identifiers to access and modify arbitrary directories on the NFS server during cleanup operations. No patch is currently available for this medium-severity vulnerability affecting Kubernetes environments.
This is an authentication and authorization bypass vulnerability in etcd's gRPC API layer that allows unauthorized users to execute privileged operations when etcd auth is enabled. Affected are etcd versions prior to 3.4.42, 3.5.28, and 3.6.9 (specifically the Go packages go.etcd.io/etcd/v3 and go.etcd.io/etcd). Attackers can enumerate cluster topology via MemberList, trigger denial of service through Alarm APIs, manipulate Lease operations affecting TTL-based keys, and force compaction to permanently delete historical data. Standard Kubernetes deployments are not affected as they do not rely on etcd's built-in authentication. No EPSS score or KEV listing is currently available, and the vulnerability was responsibly disclosed by multiple security researchers.
An authenticated user with restricted RBAC permissions on specific key ranges in etcd can use nested transactions to completely bypass key-level authorization controls and access the entire etcd data store. This affects etcd versions 3.4.x before 3.4.42, 3.5.x before 3.5.28, and 3.6.x before 3.6.9. While Kubernetes deployments are typically protected because Kubernetes handles authentication and authorization at the API server layer rather than relying on etcd's built-in controls, direct etcd deployments with RBAC restrictions are at significant risk.
A configuration injection vulnerability in Kubernetes ingress-nginx controller allows authenticated attackers to inject arbitrary nginx configuration through specially crafted Ingress annotations, leading to remote code execution with controller privileges and exposure of all cluster Secrets. The vulnerability has a high CVSS score of 8.8 and affects the ingress-nginx controller's annotation parsing mechanism. No active exploitation (not in KEV) or public POC has been reported, though the attack requires only low privileges and network access.
Path traversal in Apple and Kubernetes DAG management APIs allows authenticated attackers to access arbitrary files outside the intended directory by injecting URL-encoded forward slashes into file name parameters on GET, DELETE, RENAME, and EXECUTE endpoints. The vulnerability affects systems where a previous patch (CVE-2026-27598) only secured the CREATE endpoint while leaving other API functions unprotected. An attacker with valid credentials can read, modify, or execute unintended DAG files on the affected system.
Budibase, a low-code platform distributed as a Docker/Kubernetes application, contains a Server-Side Request Forgery (SSRF) vulnerability in its REST datasource query preview endpoint. Authenticated admin users can force the server to make HTTP requests to arbitrary URLs including cloud metadata services, internal networks, and Kubernetes APIs. A detailed proof-of-concept exists demonstrating theft of GCP OAuth2 tokens with cloud-platform scope, CouchDB credential extraction, and internal service enumeration. The CVSS score of 8.7 reflects high confidentiality and integrity impact with changed scope, requiring high privileges but low attack complexity.
The Tekton Pipelines git resolver contains a path traversal vulnerability allowing authenticated tenants to read arbitrary files from the resolver pod's filesystem via the pathInRepo parameter. Affected products include github.com/tektoncd/pipeline versions 1.0.0 through 1.10.0 across multiple release branches. The vulnerability enables credential exfiltration and privilege escalation from namespace-scoped access to cluster-wide secret reading capabilities. A proof-of-concept was provided by the vulnerability reporter Oleh Konko.
A denial of service vulnerability (CVSS 6.5). Remediation should follow standard vulnerability management procedures.
The kube-router proxy module fails to validate Service externalIPs and LoadBalancer IPs against configured IP ranges, allowing namespace-scoped users to bind arbitrary VIPs on all cluster nodes and hijack traffic to critical services like kube-dns. This affects all kube-router v2.x versions including v2.7.1, primarily impacting multi-tenant clusters where untrusted users have Service creation permissions. A detailed proof-of-concept demonstrates single-command cluster DNS takedown and arbitrary VIP binding with traffic redirection to attacker-controlled pods, though EPSS scoring is not available for this recently disclosed vulnerability.
CVE-2026-32769 is a security vulnerability. Remediation should follow standard vulnerability management procedures. Vendor patch is available.
CVE-2026-32768 is a security vulnerability. Remediation should follow standard vulnerability management procedures. Vendor patch is available.
A misconfigured NetworkPolicy in Kubernetes deployments allows attackers to perform unauthorized lateral movement between namespaces, breaking namespace isolation security boundaries. This vulnerability affects Kubernetes environments with improperly configured inter-namespace NetworkPolicies, specifically those with 'inter-ns' prefixed policies in monitoring namespaces. An attacker who compromises any component can pivot to access resources in other namespaces, potentially accessing sensitive data or systems they shouldn't have access to.
OneUptime versions prior to 10.0.24 contain a sensitive information exposure vulnerability where the password reset flow logs complete password reset URLs containing plaintext reset tokens at the INFO log level, which is enabled by default in production environments. Any actor with access to application logs-including log aggregation systems, Docker logs, or Kubernetes pod logs-can extract these tokens and perform account takeover on any user. The vulnerability is fixed in version 10.0.24.
Silent event loss in Inspektor Gadget prior to 0.50.1 allows local attackers to cause denial of service by filling the 256KB ring-buffer, which triggers undetected data drops without alerting users or administrators. When the buffer becomes full, gadgets silently discard events and fail to report the loss count, potentially hiding critical system events from Kubernetes cluster and Linux host monitoring. A local attacker with limited privileges can exploit this to obscure malicious activity or system anomalies by saturating the instrumentation buffer.