The internet is broken right now.
45 critical CVEs without a fix in the last 48 hours. Developers are crying quietly into their IDEs. 61 more already have a fix that someone still has to apply.
-
CVE-2026-116333
Buffer over-read in PostgreSQL parallel B-tree scans allows an object creator to read attacker-chosen server-memory addresses, via a function incorrectly marked PARALLEL SAFE or STABLE to make participating workers build different scan-key arrays. Within major versions 17-18, minor versions before PostgreSQL 18.7 and 17.12 are affected. Versions before PostgreSQL 17 are unaffected.
-
CVE-2026-17391
IBM API Connect 10.0.8.0 through 10.0.8.12 and 12.1.0.0 through 12.1.1.2 could allow a remote authenticated attacker to obtain sensitive information due to broken object-level authorization.
-
CVE-2026-130007
k8s-mcp-server manages Kubernetes clusters through the Model Context Protocol. Prior to 1.1.0, SERVER_MODE defaults to SSE, with the listener on 0.0.0.0:8080, but main.go creates the SSE server without an authentication option or context function. The GET /sse handler creates a sessionId that is only a correlation value, and the POST /message handler accepts a loadable sessionId without an authorization or Origin check before dispatching tools/call; consequently, any client able to reach the port can enumerate and invoke registered Kubernetes tools without credentials. Read-only mode is disabled by default, leaving write-capable Kubernetes tools available to run with the server's configured cluster privileges. This issue is fixed in version 1.1.0.
-
CVE-2026-119122
Crawl4AI before 0.9.5 contains a blind server-side request forgery vulnerability in the Docker server that allows API clients to bypass the egress proxy by enabling check_cache_freshness. Attackers can cache a controlled URL, then trigger CacheValidator to follow 302 redirects or DNS rebinding to internal and cloud-metadata addresses for port and service discovery.
-
CVE-2026-109860
CAST Imaging is vulnerable to Path Traversal vulnerability in "/api/server-folders" and "/api/server-folders/download" endpoints. These file-management endpoints accept user-supplied filesystem paths and allow selected system commands to be executed at the specified location, but they do not enforce sufficient authorization boundaries or restrict access to a predefined set of resources. An authenticated attacker can enumerate server directories via the "ls" command and retrieve arbitrary files outside of the intended application workspace, such as "/etc/passwd", exposing sensitive data. The vulnerability has only been confirmed in version 3.6.2 but may also affect other versions. The vendor did not provide the affected version range, but assured that the vulnerability is fixed "in newer versions".
-
CVE-2026-109858
CAST Imaging has multiple path traversal vulnerabilities in various endpoints. The application does not properly validate the path parameters, allowing an authenticated attacker to use "../" sequences to escape the intended directory and read arbitrary files. Critically, by reading "/proc/self/environ", the attacker can obtain environment variables containing sensitive credentials such as Keycloak and PostgreSQL passwords, leading to full server compromise and administrative access. The vulnerability has only been confirmed in version 3.6.2 but may also affect other versions. The vendor did not provide the affected version range, but assured that the vulnerability is fixed "in newer versions".
-
CVE-2026-126197
pgEdge Postgres MCP Server enables SQL queries against PostgreSQL databases through Model Context Protocol compatible clients. Prior to 1.1.0, custom pl-do tools interpolated JSON arguments into a PostgreSQL DO block delimited by fixed dollar-quote tags. An argument containing $mcp_custom_tool$ could terminate the block and inject arbitrary SQL statements, while $mcp_args$ could inject procedural code into a plpgsql block. Because the statement used the simple query protocol, injected text could include additional statements and transaction control, bypassing read_only and accessing data or modifying resources available to the connecting role. The issue requires at least one pl-do custom tool; sql and pl-func tools pass arguments as bind parameters and are not affected by this injection mechanism. This issue is fixed in 1.1.0.
-
CVE-2026-120809
A use-after-free read issue exists in MongoDB’s SBE (Slot-Based Execution) engine when processing aggregation pipelines with $unwind on time-series collections. An internal error in state-preservation logic can cause a value still referenced by downstream stages to have its underlying memory freed prematurely. An authenticated user can craft an aggregation pipeline under specific conditions to trigger this condition, potentially reading recycled server heap memory in query results or crashing the server. This issue is only triggerable when a non-default server parameter related to the SBE engine is enabled.
-
CVE-2026-127077
OpsTree BuildPiper before 2.7.0 contains an authorization bypass vulnerability that allows authenticated users to create environments on unassigned Kubernetes clusters by tampering with cluster_id and namespaces values. Attackers can modify requests to POST /api/v1/project/{project_id}/env/ to target another project's cluster, then trigger builds and deployments into it, bypassing project isolation.
-
CVE-2026-87755
Stateful session bean identifiers are not generated from a cryptographically secure random source. A remote attacker with access to the EJB remote endpoint who holds one valid identifier can derive those of other clients' sessions and invoke methods on their beans, exposing or modifying other users' session state. Users are recommended to upgrade to version 10.3.0 or 11.0.0-M2, which fix the issue.
-
CVE-2026-123804
SuperTokens Core is an open source authentication server. From 7.0.0 until 11.0.9, 11.1.4, 11.2.4, 11.3.8, 11.4.8, 12.0.11, 12.1.2, 12.2.1, and 12.3.1, createNewSession can resolve a primary user before unlinkAccounts detaches a login method and then insert the session after unlink has enumerated and revoked the existing sessions. A user who controls the linked login method can exploit this race to retain a signed, database-backed, refreshable session as the former primary identity after that method is unlinked; the session persists until expiry or explicit revocation. This issue is fixed in versions 11.0.9, 11.1.4, 11.2.4, 11.3.8, 11.4.8, 12.0.11, 12.1.2, 12.2.1, and 12.3.1.
-
CVE-2026-76171
In Apache James server prior version 3.9.1 improper validation in JMAP webPush SSRF validation did lead to the possibility to reach internal addresses. Users are recommended to upgrade to version 3.9.1 which fixes this issue.
-
CVE-2026-118375
llama.cpp b8841 before b11865 contains an out-of-bounds write vulnerability in the ggml-rpc RDMA transport rdma_recv() function that allows unauthenticated peers to overflow the rpc-server stack. Attackers advertising RDMA capabilities in the HELLO message can send up to 256 KiB into a 1-byte stack variable in rpc_serve_client(), crashing the server or potentially executing code.
-
CVE-2026-16021
Mattermost versions 11.11.x <= 11.11.0, 11.7.x <= 11.7.11, 11.10.x <= 11.10.2 fail to bound the resources consumed while processing an uploaded custom emoji image which allows an authenticated user with default permissions to make the server unresponsive for all users via repeated custom emoji uploads.. Mattermost Advisory ID: MMSA-2026-00720
-
CVE-2026-126398
vLLM from 0.28.0 before 0.30.0 contains a reachable assertion vulnerability in GPUModelRunner._init_xdrope_positions that allows remote attackers to crash the engine by submitting embeddings-only requests. Attackers can send a /v1/completions request containing only prompt_embeds to XD-RoPE models with --enable-prompt-embeds, terminating the EngineCore and API server for all clients.
-
CVE-2026-77167
Abusing the federated share endpoint, an attacker can write arbitrary rows into the oc_share_external database table. These rows persist indefinitely — there is no automatic cleanup mechanism. Three distinct denial-of-service conditions follow from this write primitive, all triggered from the same unauthenticated source and all self-renewing on a 24-hour cycle for as long as the injected rows remain in the database.
-
CVE-2026-106076
A vulnerability allowing an authenticated organization-transfer peer to inject SQL into the configuration database of the source Veeam Backup for Microsoft 365 server, leading to cross-tenant read and write of configuration data and remote code execution as the PostgreSQL service account.
-
CVE-2026-116322
Improper locking in PostgreSQL startup processing allows an unauthenticated network client to trigger a server restart, via arranging for authentication_timeout to arrive while it holds a spinlock during cancel handling. Within major version 18, minor versions before PostgreSQL 18.7 are affected. Versions before PostgreSQL 18 are unaffected.
-
CVE-2026-120803
A use-after-free issue exists in MongoDB’s SBE (Slot-Based Execution) implementation of a window function accumulator used by certain $setWindowFields aggregation pipelines. Under specific conditions, the accumulator can retain a dangling pointer to memory that has since been freed. When this stale pointer is later dereferenced during serialization of the carried-forward value, it can result in a heap out-of-bounds read that may expose adjacent heap contents and crash the mongod server. An authenticated user with readWrite privileges on a database can trigger this issue by crafting a specific aggregation pipeline.
-
CVE-2026-117545
IBM webMethods Integration Server 12.1 could allow a remote authenticated attacker to bypass security restrictions due to improper authentication.
-
CVE-2026-121415
Parse Server before 8.6.111 and 9.0.0 before 9.10.6-alpha.3 on MongoDB skips field type validation for null-valued dot notation keys, allowing sub-documents to be written beneath Pointer, Number, String, Date, or Boolean fields. Any client with write access, including unauthenticated clients where public writes are permitted, can corrupt field values, make objects unreadable via internal date fields, or break atomic operations on number fields.
-
CVE-2026-71367
An incorrect authorization vulnerability was found in AWX's WebSocket consumer. The consumer_access() function only enforces access controls for object-level event groups (job_events, workflow_events, ad_hoc_command_events) but does not restrict subscriptions to global broadcast groups such as 'jobs-status_changed'. When an authenticated user subscribes to a global broadcast group, the server accepts the subscription without verifying that the user has permission to view the associated resources. In multi-organization deployments, this allows a low-privilege user in one organization to receive real-time job status metadata from all organizations, including job IDs, status transitions, timestamps, and infrastructure topology details such as instance group names, bypassing the organization isolation enforced by the REST API.
-
CVE-2026-120814
An improper privilege management issue in the sharding subsystem of MongoDB Server may allow a user holding certain shard-level administrative privileges to cause persisted internal coordination state to be reprocessed with elevated internal authority. Because that state is trusted rather than re-validated, operations may be carried out beyond the user's authorized scope, including changes to cluster-wide security configuration. Successful triggering this issue requires an authenticated session directly against a cluster member and can result in unauthorized access to and modification of data across the cluster.
-
CVE-2026-108500
Input validation vulnerability in Apache ActiveMQ configured to use the LDAPAuthorizationMapPlugin allows an authenticated attacker to create or address a destination with a crafted name containing LDAP DN special characters (such as comma, plus, equals, backslash) can manipulate the LDAP search base used for ACL lookups. This issue affects Apache ActiveMQ: from 6.3.0 before 6.3.3, from 5.19.0 before 5.19.12. Users are recommended to upgrade to version 6.4.0, 6.3.3 or 5.19.12, which fixes the issue.
-
CVE-2026-118624
MLflow Tracking Server Jobs Missing Authorization Information Disclosure Vulnerability. This vulnerability allows remote attackers to disclose sensitive information on affected installations of MLflow. Authentication is required to exploit this vulnerability. The specific flaw exists within the _find_validator function. The issue results from the lack of authorization prior to allowing access to functionality. An attacker can leverage this vulnerability to disclose information in the context of the service account. Was ZDI-CAN-31894.
-
CVE-2026-119641
** UNSUPPORTED WHEN ASSIGNED ** Improper input validation vulnerability in Apache Oozie. An authenticated Oozie user can reconfigure another tenant's workflow job and change its ownership. This issue affects Apache Oozie: all versions. As this project is retired, we do not plan to release a version that fixes this issue. Users are recommended to find an alternative or restrict access to the instance to trusted users. NOTE: This vulnerability only affects products that are no longer supported by the maintainer.
-
CVE-2026-63341
Aedes is an MQTT broker that can run on any stream server. Prior to 1.1.1, the MQTT broker does not bound topic depth or structural complexity before passing topics through `mqemitter` into `qlobber`. A `PUBLISH`, `SUBSCRIBE`, or registered `Will` message containing an excessively deep topic can reach `qlobber._split_words` in `lib/qlobber.js` and trigger an uncaught `too many words` exception. Calls from `handlers/subscribe.js` and the publishing logic are not protected against that exception, so it escapes to the `Node.js event loop` and terminates the entire broker. The `Will` path is independently reachable when a client disconnects ungracefully: it calls `broker.publish` directly and bypasses the `PUBLISH` handler. An attacker must be able to establish an `MQTT session`, authenticated or unauthenticated according to broker configuration. A successful attack disconnects all legitimate clients and prevents reconnection until the broker process restarts. This issue is fixed in version 1.1.1.
-
CVE-2026-110720
Consul-k8s versions 0.42.0 and later are vulnerable to a cross-namespace authorization bypass in component ACL binding rules that may allow an attacker to obtain elevated Consul privileges. This vulnerability (CVE-2026-110720) is fixed in consul-k8s 2.1.0 and 1.8.18.
-
CVE-2026-115370
Incus 6.0.5 before 6.0.7, 6.12.0 before 7.0.2, and 7.1.0 before 7.5.0 contains a restriction bypass vulnerability that allows restricted project users to use pull mode migration via clustered refresh requests. Attackers with instance creation rights can send a refresh migration request for an existing instance to make the server open outbound TLS websocket connections and probe reachable services.
-
CVE-2026-115454
An excessive resource allocation issue in the remote workflow editing capabilities of KNIME Business Hub prior to version 1.19.0 allows a authenticated attack to exhaust resources on the server side. This leads to intermittent unavailability of workflow execution functionality. A workaround is not possible.
-
CVE-2026-130681
SiYuan before 3.8.10 contains a stored cross-site scripting vulnerability, an incomplete fix for CVE-2026-44670, because renderAVAttribute() inserts an Attribute View name unescaped into the database row attribute panel header. Attackers who can persist the name, such as via the setAttrViewName action, can inject an onerror handler that runs OS commands through Node.js when a desktop user opens the panel.
-
CVE-2026-130374
AdminNeo before 5.9.1 contains an sql injection vulnerability that allows remote attackers to inject SQL by supplying crafted IN and NOT IN search values that process_length() passes unescaped. Attackers can send a logged-in user a cross-site GET link using a backslash-escaped apostrophe to run blind SQL with that user's database rights on PostgreSQL, SQLite, MS SQL, Oracle, or MySQL in NO_BACKSLASH_ESCAPES mode.
-
CVE-2026-124267
Missing Authorization vulnerability in Noor Alam Gmail SMTP gmail-smtp allows Exploiting Incorrectly Configured Access Control Security Levels.This issue affects Gmail SMTP: from n/a through 1.2.3.23.
-
CVE-2026-126163
A use of externally-controlled format string vulnerability in Fortinet FortiMail 8.0.0 through 8.0.1 may allow a remote unauthenticated attacker to execute unauthorized code or commands via a specially crafted SMTP command.
-
CVE-2026-119278
MariaDB Connector/J is used to connect applications developed in Java to MariaDB and MySQL databases. From 3.0.0 until 3.3.7 and 3.4.5, during caching_sha2_password authentication, the authentication-more-data response was treated as a length-encoded byte array, memory was allocated from the server-declared length, and authResult[0] was accessed without ensuring a byte existed. A malicious or compromised server, or a machine-in-the-middle attacker, could declare a very large length or send an empty response before any password was transmitted. The malformed packet could cause excessive allocation or an unchecked ArrayIndexOutOfBoundsException escaping from DriverManager.getConnection(). This issue is fixed in versions 3.3.7 and 3.4.5.
-
CVE-2026-128080
OpenTelemetry JavaScript is the OpenTelemetry JavaScript client. From 0.204.0 until 0.222.0, @opentelemetry/instrumentation-http writes the raw query string from incoming HTTP requests to the url.query server-span attribute without applying the redactedQueryParams protection used for outgoing spans. Services receiving signed URLs or other credentials in query parameters can therefore export still-valid secrets to tracing backends, where observability infrastructure and users with span-read access may read them. The redactedQueryParams option does not protect incoming spans in affected versions. This issue is fixed in version 0.222.0.
-
CVE-2026-105948
In affected versions of Octopus Server, incorrect validation of permissions assigned to a user that would allow them to edit their own login identities could result in that user being granted elevated permissions, up to and including system administrator access. This issue affects instances using LDAP authentication with self-service login editing enabled.
-
CVE-2026-82120
Heap-based Buffer Overflow Vulnerability in HiRDB. This issue affects HiRDB Server Version 10: from 10-00 before 10-12-01; HiRDB Server Version 9: from 09-00 through 09-66-22; HiRDB Server with Additional Function Version 9: from 09-00 through 09-66/V; HiRDB Server Version 9(32): from 09-02 through 09-66-22; HiRDB Server with Additional Function Version 9(32): from 09-00 through 09-66-/V.
-
CVE-2026-131800
A stored cross-site scripting vulnerability was identified in GitHub Enterprise Server that allowed an organization member with permission to manage custom-property definitions and edit property values to add an attacker-controlled SSH authentication key to the personal account of another member through HTML stored in a single-select custom-property option. Exploitation required another definition manager in the same organization to attempt to remove the poisoned option while it was in use and save the definition with a valid sudo session. Repository access through the added key depended on the permissions of the victim and independent authorization controls. This vulnerability affected supported GitHub Enterprise Server prior to 3.23 and was fixed in versions 3.19.16, 3.20.12, 3.21.10, and 3.22.5. This vulnerability was reported via the GitHub Bug Bounty program.
-
CVE-2026-17228
IBM API Connect 12.1.0.0 through 12.1.1.2 could allow a remote authenticated attacker to obtain sensitive information due to server-side request forgery. Destination restrictions on outbound requests can be bypassed, which could allow the server to retrieve content from unintended resources.
-
CVE-2026-120306
Beta9 turns each shell command in a tenant image definition into a Dockerfile RUN line and executes it with Buildah on the build worker. The worker image sets BUILDAH_ISOLATION=chroot, and no buildah invocation passes --isolation, so the step only changes its filesystem root. It keeps the worker pod's PID, network, IPC, and user namespaces, runs as uid 0, and holds CAP_DAC_OVERRIDE. PID 1 in that procfs is the worker entrypoint, so /proc/1/root is the worker container's filesystem and is writable. That filesystem contains the runc, buildah, and runsc binaries the worker uses to start tenant containers, the worker's Kubernetes service account token, and the host directories mounted into the pod. Worker pods are privileged. Clip v2 layered builds are the default, so a tenant with a workspace token reaches this by submitting one image build. The read-only proof confirmed the shared namespaces, CAP_DAC_OVERRIDE, and that the worker's runc binary is visible at /proc/1/root. Fixed in worker 0.1.756: Buildah RUN steps, including the WORKDIR mkdir, use OCI isolation with private PID, IPC, and network namespaces, and tenant ARG values are passed only into the build container.
-
CVE-2026-115372
Incus from 7.3.0 before 7.5.0 contains a project restriction bypass that allows restricted project users to set block.create_options through volume update, copy, and move operations. Attackers can set the option on a custom volume and trigger a copy or move onto a formatting pool to inject arguments into mkfs executed as root.
-
CVE-2026-126950
Zipline through 4.4.0 contains a path traversal vulnerability that allows authenticated attackers to delete arbitrary server files by injecting traversal sequences via the x-zipline-file-extension header. Attackers can upload a file with a poisoned stored name and short expiry so LocalDatasource deletion removes chosen files, causing denial of service or data loss.
-
CVE-2026-19833
A missing authorization vulnerability in the ACL API in Synology DiskStation Manager (DSM) before 7.3.2-86009-5 and 7.4.1-90080-1 allows remote authenticated users to read or write specific files.
-
CVE-2026-124496
The DRPC HTTP endpoint lets an authenticated client act as another user without applying the configured impersonation authorizer. An authenticated user can execute DRPC functions that are restricted to another user. Users are recommended to upgrade to version 3.2.0, which fixes the issue.
-
CVE-2026-111822
Improper invalidation of cached authorizations vulnerability in Apache ActiveMQ, Apache ActiveMQ Runtime Config. After the runtime configuration plugin removes a principal's write grant on a destination, an authenticated connection opened before the change can keep sending to that destination, including through a new anonymous producer, while fresh connections and new explicit producers are denied. This issue affects Apache ActiveMQ: from 6.0.0 before 6.3.3, before 5.19.12; Apache ActiveMQ Runtime Config: from 6.0.0 before 6.3.3, before 5.19.12. Users are recommended to upgrade to version 6.4.0, 6.3.3 or 5.19.12, which fixes the issue.
-
CVE-2026-63378
TAKO AI Agent for Okta is a local-first AI agent that helps Okta administrators query tenant data using natural language. Prior to 3.0.1-beta, any authenticated user with a valid `fctr_session` cookie can submit arbitrary Python source in the `script_code` field of `POST /api/react/execute-script`, without administrator privilege. In `src/api/routers/react_stream.py`, the source is stored with `skip_discovery=True`, bypassing AI-based content filtering, and is passed to `validate_generated_code` in `src/utils/security_config.py`. `BLOCKED_PATTERNS` recognizes direct `__import__` calls but misses the import name passed as a string, `getattr` is in `ALLOWED_BUILTINS`, and the AST checker handles `ast.Name` and `ast.Attribute` targets but fails to handle `ast.Call` targets. Dynamic `getattr` call chains through `__builtins__` therefore evade validation. The `_write_temp_script` helper writes the accepted code to disk, and `asyncio.create_subprocess_exec()` executes it as the application process user, which is root in the default container. Command output is returned as Server-Sent Events through `GET /api/react/stream-react-updates?process_id=<id>`. The resulting container compromise can disclose environment secrets including Okta API tokens and the JWT key, expose the SQLite database, and permit backend-service pivoting and persistence. This issue is fixed in version 3.0.1-beta.
-
CVE-2026-116170
SiYuan before v3.8.7 contains an sql injection vulnerability that allows admin-token holders to read history.db by injecting unescaped op and type=3 query values. Attackers can send crafted UNION SELECT payloads to searchHistory or getHistoryItems endpoints to exfiltrate deleted document content, titles, and paths.
-
CVE-2026-111239
The optional SSH administration service grants full access to the administration shell to every user that authenticates against its configured login domain, without checking for an administrative role. An attacker holding the credentials of any user in that domain, including a low-privileged application user, can execute arbitrary code as the server process. Users are recommended to disable the SSH administration service, or to block network access to it from untrusted clients and restrict its login domain to administrative users.
-
CVE-2026-103002
GPUStack is an open-source GPU cluster manager for AI model serving and GPU instance provisioning. From 2.2.0 until 2.2.2, cluster_apiserver_proxy forwarded an authenticated non-administrator's requests to a selected Kubernetes cluster without a resource-level authorization check. The proxy used a reachable worker's Kubernetes ServiceAccount credentials, so the caller could read sensitive resources, create or modify workloads, or delete resources according to that ServiceAccount's permissions. Exploitation required a Kubernetes-provider cluster with a reachable worker, and attacker-controlled workloads could result in code execution in the target cluster. This issue is fixed in version 2.2.2.