Daily Digest
The single most urgent item this period is CVE-2019-1068 in Microsoft SQL Server, a CISA KEV-confirmed actively exploited vulnerability with an expired remediation due date of August 29, 2026. While several CVSS-critical vulnerabilities appear in the database landscape—including an ArcadeDB authentication bypass (CVSS 9.8) and hardcoded credential exposures in PostgreSQL tooling (CVSS 9.8)—CISA's confirmed exploitation classification takes precedence over theoretical severity scores. Immediate action is required for the SQL Server entry; the remaining entries below are sorted by confirmed-exploitation status first, then CVSS descending.
Summary: MariaDB Connector/Node.js (versions prior to 3.3.3, 3.4.6, 3.5.3) transmits database credentials to the server before completing TLS certificate validation when SSL is enabled without a pinned CA or server certificate. An active man-in-the-middle attacker can present a self-signed certificate, capture the plaintext password, and authenticate directly—even though the connector subsequently rejects the server and closes the connection. This affects any application using the Node.js connector with SSL enabled but no certificate pinning.
Affected Features / Interoperability: TLS/SSL handshake and authentication handshake in the MariaDB Connector/Node.js library. Downstream impact: any application using this connector to connect to MariaDB or MySQL with `ssl` enabled but without pinned certificates or strict CA validation.
Source: https://github.com/mariadb-corporation/mariadb-connector-nodejs/releases/tag/3.3.3
Actions:
Summary: The Keep Backup Daily WordPress plugin (versions prior to 2.1.4) exposes a cron endpoint (`kbd_cron_process`) without authentication, allowing unauthenticated attackers to trigger full MySQL database dumps. Attackers can predict the partially randomized backup filename based on the database name, a limited random range, and the Unix timestamp, then download the backup from the publicly accessible uploads directory.
Affected Features / Interoperability: The plugin's backup creation and storage mechanism; any WordPress instance using this plugin with MySQL as its backend database.
Actions:
Summary: OpenEMR through version 8.2.0 contains an authenticated SQL injection vulnerability in the backup configuration import feature. Administrators with admin or super ACL privileges can upload a crafted SQL file at `form_step=202` in `backup.php`, and the application passes unfiltered input to `shell_exec` invoking the `mysql` CLI. Attackers can extract credential hashes, modify access controls, inject backdoor accounts, or write arbitrary files if MySQL FILE privileges and `secure_file_priv` settings permit.
Affected Features / Interoperability: OpenEMR's backup import functionality and MySQL CLI interaction; affects OpenEMR deployments using MySQL as the backend database.
Source: https://jivasecurity.com/writeups/openemr-backup-import-arbitrary-sql-cve-2026-39931
Actions:
Summary: MariaDB Connector/Node.js (versions prior to 3.2.4, 3.3.3, 3.4.6, 3.5.3) permits SQL injection when attacker-controlled Buffer parameters are escaped client-side under certain Asian character sets (big5, gbk, sjis, cp932, gb18030). The connector's byte-escaping logic bypasses charset-aware validation, allowing an attacker-controlled lead byte to consume a backslash as a multibyte trail byte and leave the following quote unescaped, terminating the string and enabling arbitrary SQL. The default `utf8mb4` charset and prepared statement parameters are not affected.
Affected Features / Interoperability: SQL parameter escaping under legacy Asian character sets in MariaDB Connector/Node.js; impacts applications that use these charsets and pass attacker-influenced data as parameters.
Source: https://github.com/mariadb-corporation/mariadb-connector-nodejs/releases/tag/3.2.4
Actions:
Summary: The sambitraj Student Management System (up to commit `56ba287f2e9031523ccb4244cb6e3fe530e4e5d5`) has an improper authorization flaw in the `mysqli_query` function within `student_dashboard.php`. A remote attacker can manipulate the `roll_no` argument to bypass authorization checks. The project has a rolling release model with no disclosed version numbers and has not responded to the reported issue.
Affected Features / Interoperability: Student Dashboard component using `mysqli_query`; impacts any deployment of this system that uses MySQL as its backend.
Source: https://github.com/sambitraj/STUDENT-MANAGEMENT-SYSTEM/issues/6
Actions:
Summary: MariaDB Connector/Node.js (versions prior to 3.2.4, 3.3.3, 3.4.6, 3.5.3) can disclose an account password when PAM dialog authentication is negotiated over an insecure transport. With default `sslMode=DISABLE` and `restrictedAuth=null` settings, a hostile or on-path server can request dialog authentication over plain TCP, causing the connector to transmit the password in cleartext. Proper TLS verification or local Unix sockets prevent this path, but fingerprint-only validation is insufficient.
Affected Features / Interoperability: PAM (Pluggable Authentication Modules) authentication in MariaDB Connector/Node.js; impacts deployments using PAM auth with the default SSL disabled.
Source: https://github.com/mariadb-corporation/mariadb-connector-nodejs/releases/tag/3.2.4
Actions:
Summary: MariaDB Connector/J (versions prior to 2.7.14, 3.3.5, 3.4.3, 3.5.9) can be susceptible to credential interception during the initial TLS handshake when connecting with `sslMode=verify-full` or `sslMode=verify-ca` and a password, but without configuring `serverSslCert` or a trustStore. The connector's initial-handshake path does not enforce certificate fingerprint validation, allowing an active man-in-the-middle to present a self-signed certificate, claim to be the database server, and receive the password via `mysql_clear_password` before the connection is rejected.
Affected Features / Interoperability: Initial TLS handshake and authentication in MariaDB Connector/J; impacts Java applications using these SSL modes without explicit certificate pinning.
Source: https://github.com/mariadb-corporation/mariadb-connector-j/security/advisories/GHSA-g9jj-cgmh-9f38
Actions:
Summary: MariaDB Connector/J (versions prior to 2.7.14, 3.3.5, 3.4.3, 3.5.9) allows PAM dialog authentication to transmit the account password over an insecure connection. The `mysql_clear_password` plugin is protected by secure-transport requirements, but the sibling PAM dialog handler does not inherit this protection, defaulting to no secure-transport requirement. With default `sslMode=DISABLE` and `restrictedAuth=null`, a hostile server can request dialog auth over plain TCP and receive the plaintext password.
Affected Features / Interoperability: PAM dialog authentication path in MariaDB Connector/J; impacts deployments using PAM auth without TLS enforcement.
Source: https://github.com/mariadb-corporation/mariadb-connector-j/security/advisories/GHSA-qxvw-fvwx-5cp7
Actions:
Summary: MariaDB Connector/J (versions prior to 2.7.14, 3.3.5, 3.4.3, 3.5.9) encodes and decodes protocol text assuming the connection character set is UTF-8. The server can announce a mid-session change to `character_set_client` (e.g., via `SET NAMES`, stored routines, triggers, or a hostile server), causing the driver to continue reading/writing UTF-8 while the server interprets the same bytes under a different encoding. This leads to silent data corruption and can defeat byte-wise quoting or escaping. The fix rejects any non-UTF-8 charset after initialization with a SQLException.
Affected Features / Interoperability: Character set handling in MariaDB Connector/J protocol; impacts applications where server-side configuration or attacker-controlled sessions can alter the character set mid-session.
Source: https://github.com/mariadb-corporation/mariadb-connector-j/releases/tag/2.7.14
Actions:
Summary: MariaDB Connector/R2DBC (versions prior to 1.4.1) has the same charset-confusion vulnerability as CVE-2026-55858, but in the R2DBC non-blocking driver for Java. A server can announce a mid-session character set change via OK-packet session-state tracking, causing silent data corruption when the driver continues using UTF-8 while the server interprets bytes under a different encoding. The fix rejects non-UTF-8 character sets after initialization with an exception.
Affected Features / Interoperability: Character set handling in the R2DBC driver for MariaDB and MySQL; impacts reactive/non-blocking Java applications using this driver.
Actions:
---
Summary: `openssl_encrypt` versions prior to 1.4.0 contain hardcoded database credentials in standalone server configuration files. Attackers on the same network can use these well-known default credentials to access PostgreSQL databases directly and retrieve sensitive data. This is a pre-configured credential exposure rather than a运行时 vulnerability.
Affected Features / Interoperability: Standalone server configuration files in `openssl_encrypt`; impacts any PostgreSQL database deployed alongside or managed by affected versions of this software.
Source: https://github.com/jahlives/openssl_encrypt/security/advisories/GHSA-v4vm-4xf2-fhqj
Actions:
Summary: baserCMS before version 5.3.0 has a SQL injection vulnerability in `BcDatabaseService.php` that allows authenticated administrators to inject attacker-controlled table names and configuration values directly into SQL statements. Attackers can chain this with a backup restore code injection flaw (PHP code outside class definitions in schema files executes unconditionally) to plant malicious table names and trigger error-based SQL injection against the PostgreSQL backend, retrieving database version, schema contents, and arbitrary data.
Affected Features / Interoperability: baserCMS database service layer; impacts any baserCMS deployment using PostgreSQL as its backend.
Source: https://github.com/baserproject/basercms/security/advisories/GHSA-cg65-f2m7-9fqj
Actions:
Summary: TimescaleDB through version 2.29.1 has an out-of-bounds read vulnerability in bulk text dictionary decompression. Authenticated attackers with direct DML access to a non-frozen physical compressed hypertable can supply a crafted Simple8b selector-11 value that bypasses index validation, causing an out-of-bounds read before the base of the live offsets array. This can result in incorrect aggregation output, backend `SIGSEGV`, or PostgreSQL crash recovery depending on build configuration.
Affected Features / Interoperability: TimescaleDB compressed hypertable operations; impacts TimescaleDB deployments using physical compression on hypertables with the Arrow dictionary-index type.
Source: https://github.com/timescale/timescaledb/pull/10360
Actions:
Summary: WAPT Server versions 2.6.1.17834 and earlier have a SQL injection vulnerability in the `columns` parameter of the `GET /api/v3/hosts` endpoint. A remote authenticated user with read-only privileges can inject arbitrary PostgreSQL expressions into the SQL query, bypass host scope restrictions, and read data from other rows or tables within the database.
Affected Features / Interoperability: WAPT Server API endpoint for host enumeration; impacts WAPT Server deployments using PostgreSQL as their backend.
Source: https://www.vulncheck.com/advisories/wapt-server-sql-injection-via-api-v3-hosts-endpoint
Actions:
Summary: phpMyFAQ before version 4.1.7, when using PostgreSQL via the native `pgsql` PHP extension, declares an incorrect LIKE ESCAPE character ('=') in the backend while the escape function uses a '|' prefix. This makes wildcard escaping a no-op, leaving `%` and `_` characters active in user-supplied search input. An unauthenticated attacker can submit these characters in the public FAQ search form to force maximally broad pattern matches and expensive sequential scans, resulting in denial of service. The PDO PostgreSQL backend is unaffected, and this does not enable SQL injection.
Affected Features / Interoperability: phpMyFAQ search functionality when using the native `pgsql` PHP extension; does not affect PDO-based PostgreSQL connections.
Source: https://github.com/thorsten/phpMyFAQ/security/advisories/GHSA-5hx6-c293-588h
Actions:
Summary: n8n before versions 1.123.67, 2.31.5, and 2.32.1 contains a SQL injection vulnerability in the PostgresTrigger node. User-supplied identifier parameters (channel, function, and trigger names) are interpolated into SQL statements without proper escaping. An authenticated user can inject arbitrary SQL executed against the connected PostgreSQL database with the configured credential's privileges, allowing full read and write access.
Affected Features / Interoperability: n8n's PostgresTrigger node; impacts n8n workflows that connect to PostgreSQL databases using this trigger.
Source: https://github.com/n8n-io/n8n/security/advisories/GHSA-jqwr-vx3p-r266
Actions:
Summary: The ash-project `ash_postgres` library (versions from 0.25.0 before 2.13.0) has an unchecked return value vulnerability in `AshPostgres.MultiTenancy.rename_tenant/3`. When a tenant is renamed to a name colliding with an existing tenant's schema, PostgreSQL rejects the rename, but the library discards the error result and unconditionally returns `:ok`. The calling action then commits the tenant row with the new name pointing to a different tenant's schema, allowing data access across tenant boundaries.
Affected Features / Interoperability: ash_postgres multi-tenancy schema management; impacts Elixir/Phoenix applications using ash_postgres for multi-tenant PostgreSQL deployments.
Source: https://github.com/ash-project/ash_postgres/security/advisories/GHSA-6fqq-j9c4-5766
Actions:
---
Summary: Budibase before version 3.40.0 contains a SQL injection vulnerability in the Oracle datasource connector's post-write row lookup that fails to escape table names in identifiers. Attackers with write permission on a table with a double-quote character in its name can inject SQL that executes as the datasource's database user, allowing read or modification of arbitrary data in the Oracle database.
Affected Features / Interoperability: Budibase Oracle datasource connector's row lookup after write operations; impacts Budibase deployments connecting to Oracle databases.
Source: https://github.com/Budibase/budibase/security/advisories/GHSA-xj29-x47g-9w2c
Actions:
---
Summary: ArcadeDB versions prior to 26.8.1 fail to enforce SASL authentication on data commands in the MongoDB wire-protocol plugin. Unauthenticated attackers can issue insert, find, update, delete, and create commands against any database by connecting to port 27017 without providing any credentials.
Affected Features / Interoperability: ArcadeDB's MongoDB wire-protocol plugin (port 27017); impacts any deployment of ArcadeDB using the MongoDB-compatible API without additional network controls.
Source: https://github.com/ArcadeData/arcadedb/security/advisories/GHSA-fq9c-x968-g278
Actions:
Summary: Users submitting SQL through applications using the MongoDB Connector for BI ODBC driver can supply a positioned-cursor statement whose cursor name exceeds the size of an internal fixed-length buffer. The name length is not bounded before the driver builds its diagnostic message, overwriting adjacent memory with user-supplied content. This can crash the hosting application and may allow unintended code execution within it.
Affected Features / Interoperability: MongoDB Connector for BI ODBC driver; impacts applications using BI tools (Tableau, Power BI, etc.) to query MongoDB via SQL.
Source: https://github.com/mongodb/mongo-bi-connector-odbc-driver/releases
Actions:
Summary: Budibase Server before version 3.40.0 has a NoSQL injection vulnerability in the MongoDB query execution endpoint where user-supplied parameters are interpolated into JSON query templates without sanitizing JSON metacharacters. Attackers with query write permission can inject JSON structural characters to alter MongoDB queries, bypassing filters to read, modify, or delete arbitrary documents.
Affected Features / Interoperability: Budibase MongoDB datasource query execution; impacts Budibase applications using MongoDB as a datasource.
Source: https://github.com/Budibase/budibase/security/advisories/GHSA-qw6m-8fw2-2v64
Actions:
Summary: In MongoDB Connector for BI, MongoDB object names (collection, field, index names) placed into quoted identifiers in DDL text returned by `SHOW CREATE` statements are not escaped. A user with write permission on a sampled MongoDB collection can choose a name that closes the quoted identifier early, injecting additional SQL text into the generated output. If an operator or automated tool replays that generated statement against a SQL server, the injected text executes with the session's privileges.
Affected Features / Interoperability: MongoDB Connector for BI schema sampling and `SHOW CREATE` output; impacts scenarios where BI-generated DDL is replayed on SQL systems.
Source: https://www.mongodb.com/docs/bi-connector/current/release-notes/
Actions:
Summary: Budibase before version 3.40.0 fails to redact datasource credentials stored in STRING typed fields, exposing MongoDB connection strings and Firebase private keys in plaintext. Users with table read permissions can retrieve datasource configurations through the read API to obtain live backend database credentials and service account keys.
Affected Features / Interoperability: Budibase datasource credential storage and read API; impacts any Budibase deployment with MongoDB or Firebase datasources where credentials are stored in STRING fields.
Source: https://github.com/Budibase/budibase/security/advisories/GHSA-6mpp-gfg5-x2vv
Actions:
Summary: A database user able to create a view in a namespace that MongoDB Connector for BI samples can define a view whose evaluation reliably fails, causing the schema-sampling routine to stop functioning. After retries are exhausted, the sampling logic proceeds without a valid result, leaving the `mongosqld` process running without a usable schema. SQL clients cannot obtain results until an operator removes the view or excludes its namespace from sampling.
Affected Features / Interoperability: MongoDB Connector for BI schema sampling; impacts BI deployments where untrusted users can create views in sampled namespaces.
Source: https://www.mongodb.com/docs/bi-connector/current/release-notes/
Actions:
Summary: An unauthenticated party able to reach a MongoDB Connector for BI (`mongosqld`) instance can generate enough routine connection log activity to exhaust the storage backing the configured log path. When a log write or rotation fails, the error is unhandled and the shared `mongosqld` process terminates, ending service for all connected SQL clients. The process continues to fail on startup until storage is restored.
Affected Features / Interoperability: MongoDB Connector for BI logging subsystem; impacts any mongosqld deployment where network access is not restricted.
Source: https://www.mongodb.com/docs/bi-connector/current/release-notes/
Actions:
Summary: When mongosqld is configured with a client certificate authority file, the listener requests a client certificate during TLS handshake but does not require it, so a client presenting no certificate is still accepted. In deployments relying on client certificates as the sole means of user identification, a remote party with network access can establish a session and read data exposed through the connector.
Affected Features / Interoperability: MongoDB Connector for BI mutual TLS (mTLS) configuration; impacts deployments using client certificates for authentication.
Source: https://www.mongodb.com/docs/bi-connector/current/release-notes/
Actions:
Summary: A network-reachable client that has not yet authenticated can hold a MongoDB Connector for BI authentication session open indefinitely by beginning a SASL login exchange and then declining to complete it. There is no overall time bound on the negotiation loop and no read deadline, so each incomplete session retains a worker, connection slot, and backend database connections until process restart. Repeated use can exhaust connection capacity and prevent legitimate users from establishing new sessions.
Affected Features / Interoperability: MongoDB Connector for BI SASL authentication handling; impacts any mongosqld deployment with network exposure.
Source: https://www.mongodb.com/docs/bi-connector/current/release-notes/
Actions:
Summary: Budibase before version 3.40.0 has a NoSQL injection vulnerability in the MongoDB datasource integration where user-supplied parameters are enriched
The single most urgent item in today's digest is CVE-2026-66384 affecting JFrog Artifactory's Docker cache handling, which CISA has added to the KEV catalog with a remediation deadline of 2026-09-10. Because this is a confirmed actively-exploited vulnerability with a BOD 26-04 due date, it outranks every other entry below regardless of CVSS. Beyond that, the Kubernetes ecosystem carries the highest theoretical severity this cycle — a CVSS 10.0 unauthenticated SSRF in Kubeflow Pipelines and a CVSS 9.8 unauthenticated arbitrary-file-write in the Dinky Flink dashboard (default Docker image, hardcoded token) — both of which warrant immediate inventory checks. GCP and AWS news surfaces one confirmed healthcare-sector breach tied to AWS infrastructure (Aesto, ~9.5M records), worth noting for threat-modeling context even though it is not a vulnerability in your stack.
Summary: Improper path-limitation ("path traversal" style) flaw in JFrog Artifactory that lets an authenticated user write data outside the intended Docker cache path when specific remote-repository conditions are met. CISA added it to the KEV on 2026-08-27; the required action is to apply vendor mitigations per CISA BOD 26-04 guidance for cloud services (or discontinue use). KEV remediation due date: 2026-09-10.
Affected Features / Interoperability: Specifically the Docker cache path under remote-repository configurations. Downstream impact includes container image integrity (any out-of-path write could corrupt or poison cached layers), build pipelines that pull cached images from affected repositories, and any CI/CD integration that trusts the Artifactory Docker feed.
Source: https://nvd.nist.gov/vuln/detail/CVE-2026-66384
Actions:
Summary: Unauthenticated Server-Side Request Forgery (SSRF) in the Kubeflow Pipelines frontend `/apis/v1beta1/_proxy/` (and v2beta1/pipeline variants) routes. The `routePathWithReferer()` helper forwards an attacker-controlled HTTP/HTTPS target to `createProxyMiddleware` with no allowlist and no filtering for loopback, link-local, RFC1918, or in-cluster addresses. The proxy can carry attacker-supplied methods, headers (including `Authorization`, `Cookie`, `X-Forwarded-For`), and POST bodies to internal services and return the upstream response to the unauthenticated client. No auth is required — only network reachability of the frontend.
Affected Features / Interoperability: Kubeflow Pipelines frontend proxy middleware (`server/proxy-middleware.ts`). Downstream exposure: cloud metadata endpoints (e.g., IMDS for credential theft), the Kubernetes API server, in-cluster services, and any reachable internal HTTP target. This affects ML workflow platforms that integrate with object storage, secrets, and model registries via the same frontend.
Source: https://github.com/kubeflow/pipelines/security/advisories/GHSA-gqww-5pj5-8fq7
Actions:
Summary: MCPHub's `POST /api/servers` and `PUT /api/servers/:name` endpoints accept a server configuration and immediately spawn it via `child_process.spawn`. Authentication is required, but no authorization check restricts these endpoints to admins, and there is no allowlist/sanitization on `command`/`args`. Any authenticated non-admin can specify `/bin/sh` (or any binary) with arbitrary arguments, resulting in OS-level code execution as the MCPHub process user (commonly `root` in the published Docker image). Patched in 0.12.15.
Affected Features / Interoperability: MCPHub server-registration API and the stdio MCP process orchestrator. The published Docker image, `npx`-hosted deployments, and systemd deployments are all impacted where the process user is privileged. Downstream impact: full container/host compromise given root execution.
Source: https://github.com/samanhappy/mcphub/security/advisories/GHSA-mx89-jjx9-gjr8
Actions:
Summary: The `POST /download/uploadFromRsByLocal` handler takes a caller-supplied `path` and writes to it with no validation. The route is `@SaIgnore`-marked and excluded from Sa-Token auth, so the only guard is a header equality check against a `dinkyToken` whose default value `efda1551-7958-4e0f-80a8-dfd107df3e38` is hardcoded in source and shipped to every deployment. Anyone who can reach Dinky's HTTP port (default 8888) with that token can write arbitrary files as the Dinky service account. The default Docker image has no proxy/auth on 8888 and chmod 777 on `/opt/dinky`, so demonstrated impact includes overwriting `/opt/dinky/config/static/index.html` (XSS to admin browsers) and writing `/opt/dinky/org/dinky/Dinky.class` (code execution as the service account at next JVM start via the classpath-shadow launched by `script/bin/auto.sh`). Writes run as uid 9999 (flink), not root, so `/etc`, `/root`, `/home`, `/usr` are refused. Affects Dinky v1.2.5 (current release) and the development branch — code is byte-identical. No patch is currently indicated in the advisory data.
Affected Features / Interoperability: Dinky Flink dashboard's `/download/**` HTTP handlers and the JVM classpath loaded by `auto.sh`. Downstream impact: admin-browser XSS, arbitrary code execution inside the Dinky container on next restart, and any Flink job artifacts living under `/opt/dinky`.
Source: https://github.com/DataLinkDC/dinky/security/advisories/GHSA-2p66-w3p3-5226
Actions:
Summary: `mapfish-print` accepts an attacker-controlled GML layer URL in requests to `/api/print3/print` and fetches XML parsed by `core/src/main/java/org/mapfish/print/map/geotools/GmlLayer.java` without disabling external entities and external DTDs. A remote XML/DTD can expand a local file entity, exposing its contents via the GML parsing and error path, allowing unauthenticated attackers to read files such as OS account data, Kubernetes service-account tokens, and certificates. Swapping the entity target to an internal HTTP endpoint also enables SSRF. Fixed in versions 3.28.30, 3.30.32, 3.31.24, 3.33.16, and 4.0.5.
Affected Features / Interoperability: MapFish Print `/api/print3/print` endpoint and the GmlLayer parser. Downstream: any cluster or service-account-token mounted into the MapFish pod is at risk of exfiltration; any reachable internal HTTP service is at risk of SSRF.
Source: https://github.com/mapfish/mapfish-print/security (referenced commits linked from advisory)
Actions:
Summary: Grafana Alloy's `prometheus.operator.servicemonitors` component lets a user who can create/modify `ServiceMonitor` resources in a watched namespace specify an arbitrary local file via `bearerTokenFile`. Alloy reads the file and sends its contents as a bearer token to an attacker-controlled scrape endpoint, disclosing files readable by the Alloy process — including its projected Kubernetes service-account token. Exploitation requires ServiceMonitor write access and lower privileges than Alloy's service account.
Affected Features / Interoperability: Grafana Alloy's ServiceMonitor controller. Downstream: Kubernetes service-account tokens and any other files readable by Alloy are exfiltratable to an attacker-controlled Prometheus endpoint; compromise of the SA enables further cluster action depending on RBAC.
Source: https://grafana.com/security/security-advisories/cve-2026-75889
Actions:
Summary: SSRF in Kyverno's Service Call functionality (before v1.13.4). An attacker with permission to create Kyverno `(Cluster)Policy` resources can specify an external URL in a policy's `apiCall`/`service` configuration; although Service Call is documented for in-cluster services, it also resolves external addresses. Policy context data — including Kubernetes resource contents such as Secrets — is sent in these requests, allowing exfiltration of sensitive cluster data.
Affected Features / Interoperability: Kyverno policy `apiCall`/`service` blocks; any Secret or ConfigMap content that flows through policy context evaluation is at risk of exfiltration. Downstream integrations that consume Kyverno policy evaluation results may also be impacted.
Source: https://github.com/kyverno/kyverno/security/advisories/GHSA-459x-q9hg-4gpq
Actions:
Summary: Improper input validation in the CData JDBC driver integration in Google Cloud BigQuery Data Transfer Service allows an authenticated attacker to achieve remote code execution in the connector container and escalate privileges in the tenant project using crafted JDBC connection-string parameters. Patched on 1 May 2026; no customer action is needed per Google.
Affected Features / Interoperability: BigQuery DTS CData JDBC connector container; downstream tenant-project privilege escalation if the connector is exploited before the patch window.
Source: https://docs.cloud.google.com/support/bulletins#gcp-2026-056
Actions:
Summary: Improper privilege management in Google SecOps (Chronicle SOAR) prior to 6.3.85 allows an authenticated attacker to escalate privileges to system-level administrative access using a crafted internal authentication header. Patched in 6.3.85; no customer action is needed per Google.
Affected Features / Interoperability: Chronicle SOAR internal authentication header handling and privilege escalation paths. Downstream: SOAR playbooks, integrations, and any connected security tooling that trust the SOAR's identity model.
Source: https://docs.cloud.google.com/chronicle/docs/soar/release-notes#May_23_2026
Actions:
Summary: SQL injection in a legacy dashboard widget API in Chronicle SOAR prior to 6.3.85 allows an authenticated attacker to execute blind SQL queries using a crafted request parameter. Patched in 6.3.85; no customer action is needed per Google.
Affected Features / Interoperability: Legacy dashboard widget API in SOAR; SQL backend and any data exposed through that backend.
Source: https://docs.cloud.google.com/chronicle/docs/soar/release-notes#May_23_2026
Actions:
Summary: Incorrect authorization in the GitHub Trigger Comment Control in Cloud Build prior to 2026-06-24 allows a remote attacker to execute unreviewed code in the build environment by suppressing the webhook review flow. Patched on 24 June 2026; no customer action is needed per Google.
Affected Features / Interoperability: Cloud Build GitHub trigger comment control and the build execution environment. Downstream: any artifact produced by suppressed builds, and any deployment pipeline downstream of Cloud Build that trusts the artifact.
Source: https://docs.cloud.google.com/build/docs/release-notes#August_24_2026
Actions:
Summary: Predictable resource name vulnerability in BigQuery Import Staging in Vertex AI Search for Commerce prior to 2026-04-27. An attacker who knows the victim's project number can obtain read/write access to staged data and error logs by guessing predictable bucket names. Patched; no customer action is needed per Google.
Affected Features / Interoperability: BigQuery import staging buckets for Vertex AI Search for Commerce; error logs and staged data.
Source: https://unit42.paloaltonetworks.com/hijacking-vertex-ai-model/
Actions:
Unverified chatter — not authoritative for patch decisions.