LiteLLM prior to 1.83.14 allows an authenticated internal_user to create API keys with access to routes that their role does not permit. When generating a key, the allowed_routes field is stored without verifying that the specified routes fall within the user's own permissions. A key created with access to admin-only routes can then be used to reach those routes successfully, bypassing the role-based access controls that would otherwise block the request, enabling full privilege escalation from internal_user to proxy_admin.

LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. Prior to 1.83.0, the /config/update endpoint does not enforce admin role authorization. A user who is already authenticated into the platform can then use this endpoint to modify proxy configuration and environment variables, register custom pass-through endpoint handlers pointing to attacker-controlled Python code, achieving remote code execution, read arbitrary server files by setting UI_LOGO_PATH and fetching via /get_image, and take over other privileged accounts by overwriting UI_USERNAME and UI_PASSWORD environment variables. Fixed in v1.83.0.

Trivy is a security scanner. On March 19, 2026, a threat actor used compromised credentials to publish a malicious Trivy v0.69.4 release, force-push 76 of 77 version tags in `aquasecurity/trivy-action` to credential-stealing malware, and replace all 7 tags in `aquasecurity/setup-trivy` with malicious commits. This incident is a continuation of the supply chain attack that began in late February 2026. Following the initial disclosure on March 1, credential rotation was performed but was not atomic (not all credentials were revoked simultaneously). The attacker could have use a valid token to exfiltrate newly rotated secrets during the rotation window (which lasted a few days). This could have allowed the attacker to retain access and execute the March 19 attack. Affected components include the `aquasecurity/trivy` Go / Container image version 0.69.4, the `aquasecurity/trivy-action` GitHub Action versions 0.0.1 – 0.34.2 (76/77), and the`aquasecurity/setup-trivy` GitHub Action versions 0.2.0 – 0.2.6, prior to the recreation of 0.2.6 with a safe commit. Known safe versions include versions 0.69.2 and 0.69.3 of the Trivy binary, version 0.35.0 of trivy-action, and version 0.2.6 of setup-trivy. Additionally, take other mitigations to ensure the safety of secrets. If there is any possibility that a compromised version ran in one's environment, all secrets accessible to affected pipelines must be treated as exposed and rotated immediately. Check whether one's organization pulled or executed Trivy v0.69.4 from any source. Remove any affected artifacts immediately. Review all workflows using `aquasecurity/trivy-action` or `aquasecurity/setup-trivy`. Those who referenced a version tag rather than a full commit SHA should check workflow run logs from March 19–20, 2026 for signs of compromise. Look for repositories named `tpcp-docs` in one's GitHub organization. The presence of such a repository may indicate that the fallback exfiltration mechanism was triggered and secrets were successfully stolen. Pin GitHub Actions to full, immutable commit SHA hashes, don't use mutable version tags.

LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.74.2 to before version 1.83.7, two endpoints used to preview an MCP server before saving it — POST /mcp-rest/test/connection and POST /mcp-rest/test/tools/list — accepted a full server configuration in the request body, including the command, args, and env fields used by the stdio transport. When called with a stdio configuration, the endpoints attempted to connect, which spawned the supplied command as a subprocess on the proxy host with the privileges of the proxy process. The endpoints were gated only by a valid proxy API key, with no role check. Any authenticated user — including holders of low-privilege internal-user keys — could therefore run arbitrary commands on the host. This issue has been patched in version 1.83.7.

LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.80.5 to before version 1.83.7, the POST /prompts/test endpoint accepted user-supplied prompt templates and rendered them without sandboxing. A crafted template could run arbitrary code inside the LiteLLM Proxy process. The endpoint only checks that the caller presents a valid proxy API key, so any authenticated user could reach it. Depending on how the proxy is deployed, this could expose secrets in the process environment (such as provider API keys or database credentials) and allow commands to be run on the host. This issue has been patched in version 1.83.7.

LiteLLM through 2026-04-08 allows remote attackers to execute arbitrary code via bytecode rewriting at the /guardrails/test_custom_code URI.

A vulnerability in Cisco Connected Mobile Experiences (CMX) could allow a remote, authenticated attacker without administrative privileges to alter the password of any user on an affected system. The vulnerability is due to incorrect handling of authorization checks for changing a password. An authenticated attacker without administrative privileges could exploit this vulnerability by sending a modified HTTP request to an affected device. A successful exploit could allow the attacker to alter the passwords of any user on the system, including an administrative user, and then impersonate that user.

An improper authorization vulnerability exists in Jenkins Jira Plugin 3.0.1 and earlier in JiraSite.java that allows attackers with Overall/Read access to have Jenkins connect to an attacker-specified URL using attacker-specified credentials IDs obtained through another method, capturing credentials stored in Jenkins.

Multiple vulnerabilities in the web-based management interface of Cisco SD-WAN vManage Software could allow an authenticated, remote attacker to bypass authorization and modify the configuration of an affected system, gain access to sensitive information, and view information that they are not authorized to access. For more information about these vulnerabilities, see the Details section of this advisory.

Vault Community, Vault Enterprise (“Vault”) Azure Auth method did not correctly validate the claims in the Azure-issued token, resulting in the potential bypass of the bound_locations parameter on login. Fixed in Vault Community Edition 1.19.1 and Vault Enterprise 1.19.1, 1.18.7, 1.17.14, 1.16.18.

The ACL (Access Control List) of SAP Message Server - versions KERNEL 7.22, KERNEL 7.53, KERNEL 7.54, KERNEL 7.77, RNL64UC 7.22, RNL64UC 7.22EXT, RNL64UC 7.53, KRNL64NUC 7.22, KRNL64NUC 7.22EXT, can be bypassed in certain conditions, which may enable an authenticated malicious user to enter the network of the SAP systems served by the attacked SAP Message server. This may lead to unauthorized read and write of data as well as rendering the system unavailable.

Incorrect user role checking in multiple REST API endpoints in ProLion CryptoSpike 3.0.15P2 allows a remote attacker with low privileges to execute privileged functions and achieve privilege escalation via REST API endpoint invocation.

gomatrixserverlib is a Go library for matrix protocol federation. Dendrite is a Matrix homeserver written in Go, an alternative to Synapse. The power level parsing within gomatrixserverlib was failing to parse the `"events_default"` key of the `m.room.power_levels` event, defaulting the event default power level to zero in all cases. Power levels are the matrix terminology for user access level. In rooms where the `"events_default"` power level had been changed, this could result in events either being incorrectly authorised or rejected by Dendrite servers. gomatrixserverlib contains a fix as of commit `723fd49` and Dendrite 0.9.3 has been updated accordingly. Matrix rooms where the `"events_default"` power level has not been changed from the default of zero are not vulnerable. Users are advised to upgrade. There are no known workarounds for this issue.

Hasura GraphQL Engine before 2.15.2 mishandles row-level authorization in the Update Many API for Postgres backends. The fixed versions are 2.10.2, 2.11.3, 2.12.1, 2.13.2, 2.14.1, and 2.15.2. (Versions before 2.10.0 are unaffected.)

AWS Cloud Development Kit (AWS CDK) is an open-source software development framework to define cloud infrastructure in code and provision it through AWS CloudFormation. In the packages `aws-cdk-lib` 2.0.0 until 2.80.0 and `@aws-cdk/aws-eks` 1.57.0 until 1.202.0, `eks.Cluster` and `eks.FargateCluster` constructs create two roles, `CreationRole` and `default MastersRole`, that have an overly permissive trust policy. The first, referred to as the `CreationRole`, is used by lambda handlers to create the cluster and deploy Kubernetes resources (e.g `KubernetesManifest`, `HelmChart`, ...) onto it. Users with CDK version higher or equal to 1.62.0 (including v2 users) may be affected. The second, referred to as the `default MastersRole`, is provisioned only if the `mastersRole` property isn't provided and has permissions to execute `kubectl` commands on the cluster. Users with CDK version higher or equal to 1.57.0 (including v2 users) may be affected. The issue has been fixed in `@aws-cdk/aws-eks` v1.202.0 and `aws-cdk-lib` v2.80.0. These versions no longer use the account root principal. Instead, they restrict the trust policy to the specific roles of lambda handlers that need it. There is no workaround available for CreationRole. To avoid creating the `default MastersRole`, use the `mastersRole` property to explicitly provide a role.

SGUDA U-Lock central lock control service’s user management function has incorrect authorization. A remote attacker with general user privilege can exploit this vulnerability to call privileged APIs to access, modify and delete user information.

In Mailu before version 1.7, an authenticated user can exploit a vulnerability in Mailu fetchmail script and gain full access to a Mailu instance. Mailu servers that have open registration or untrusted users are most impacted. The master and 1.7 branches are patched on our git repository. All Docker images published on docker.io/mailu for tags 1.5, 1.6, 1.7 and master are patched. For detailed instructions about patching and securing the server afterwards, see https://github.com/Mailu/Mailu/issues/1354

Insecure Permission vulnerability in Schlix Web Inc SCHLIX CMS 2.2.7-2 allows attacker to upload arbitrary files and execute arbitrary code via the tristao parameter. NOTE: this is disputed by the vendor because an admin is intentionally allowed to upload new executable PHP code, such as a theme that was obtained from a trusted source or was developed for their own website. Only an admin can upload such code, not someone else in an "attacker" role.

SENS v1.0 is vulnerable to Incorrect Access Control vulnerability.

Hitachi Vantara Pentaho Business Analytics Server versions before 9.4.0.1 and 9.3.0.2, including 8.3.x do not correctly perform an authorization check in the data source management service. 

IBM Storage Virtualize 8.4, 8.5, 8.6, and 8.7 could allow an authenticated user to escalate their privileges in an SSH session due to incorrect authorization checks to access resources.

IBM Data Risk Manager (iDNA) 2.0.6 could allow an authenticated user to escalate their privileges to administrator due to insufficient authorization checks. IBM X-Force ID: 184981.

VestaCP 0.9.8-26 contains a session token vulnerability in the LoginAs module that allows remote attackers to manipulate authentication tokens. Attackers can exploit insufficient token validation to access user accounts and perform unauthorized login requests without proper administrative permissions.

Paramiko version 2.4.1, 2.3.2, 2.2.3, 2.1.5, 2.0.8, 1.18.5, 1.17.6 contains a Incorrect Access Control vulnerability in SSH server that can result in RCE. This attack appear to be exploitable via network connectivity.

Adobe Commerce versions 2.4.3-p2 (and earlier), 2.3.7-p3 (and earlier) and 2.4.4 (and earlier) are affected by an Incorrect Authorization vulnerability. An authenticated attacker can exploit this vulnerability to achieve information exposure and privilege escalation.

M/Monit 3.7.4 contains a privilege escalation vulnerability that allows authenticated users to modify user permissions by manipulating the admin parameter. Attackers can send a POST request to the /api/1/admin/users/update endpoint with a crafted payload to grant administrative access to a standard user account.

The Administrator function of EasyTest has an Incorrect Authorization vulnerability. A remote attacker authenticated as a general user can exploit this vulnerability to bypass the intended access restrictions, to make API functions calls, manipulate system and terminate service.

iDS6 DSSPro Digital Signage System 6.2 contains an improper access control vulnerability that allows authenticated users to elevate privileges through console JavaScript functions. Attackers can create users, modify roles and permissions, and potentially achieve full application takeover by exploiting insecure direct object references.

Redaxo CMS Mediapool Addon 5.5.1 and older contains an arbitrary file upload vulnerability that allows authenticated users to bypass file extension blacklist restrictions. Attackers with editor accounts can upload executable files by using obfuscated extensions like php71 or php53 to evade the blacklist filter and execute arbitrary code.

An issue was discovered in the XCloner Backup and Restore plugin before 4.2.13 for WordPress. It gave authenticated attackers the ability to modify arbitrary files, including PHP files. Doing so would allow an attacker to achieve remote code execution. The xcloner_restore.php write_file_action could overwrite wp-config.php, for example. Alternatively, an attacker could create an exploit chain to obtain a database dump.

A vulnerability in the REST API endpoint of Cisco Data Center Network Manager (DCNM) could allow an authenticated, remote attacker with a low-privileged account to bypass authorization on the API of an affected device. The vulnerability is due to insufficient authorization of certain API functions. An attacker could exploit this vulnerability by sending a crafted request to the API using low-privileged credentials. A successful exploit could allow the attacker to perform arbitrary actions through the REST API with administrative privileges.

In Cloudera CDH before 5.7.1, Impala REVOKE ALL ON SERVER commands do not revoke all privileges.

A vulnerability in Role Based Access Control (RBAC) functionality of Cisco IOS XE Web Management Software could allow a Read-Only authenticated, remote attacker to execute commands or configuration changes as an Admin user. The vulnerability is due to incorrect handling of RBAC for the administration GUI. An attacker could exploit this vulnerability by sending a modified HTTP request to the affected device. An exploit could allow the attacker as a Read-Only user to execute CLI commands or configuration changes as if they were an Admin user.

A vulnerability in Veeam Backup & Replication allows a low-privileged user to start an agent remotely in server mode and obtain credentials, effectively escalating privileges to system-level access. This allows the attacker to upload files to the server with elevated privileges. The vulnerability exists because remote calls bypass permission checks, leading to full system compromise.

An issue exists within the SSH console of Akkadian Provisioning Manager 4.50.02 which allows a low-level privileged user to escape the web configuration file editor and escalate privileges.

A vulnerability has been identified in SINEMA Remote Connect Server (All versions < V3.0). Unpriviledged users can access services when guessing the url. An attacker could impact availability, integrity and gain information from logs and templates of the service.

Multiple flaws were found in the way samba AD DC implemented access and conformance checking of stored data. An attacker could use this flaw to cause total domain compromise.

An access control issue in ZKTeco ZKBioSecurity V5000 3.0.5_r allows attackers to arbitrarily create admin users via a crafted HTTP request.

Talos Linux is a Linux distribution built for Kubernetes deployments. Talos worker nodes use a join token to get accepted into the Talos cluster. Due to improper validation of the request while signing a worker node CSR (certificate signing request) Talos control plane node might issue Talos API certificate which allows full access to Talos API on a control plane node. Accessing Talos API with full level access on a control plane node might reveal sensitive information which allows full level access to the cluster (Kubernetes and Talos PKI, etc.). Talos API join token is stored in the machine configuration on the worker node. When configured correctly, Kubernetes workloads don't have access to the machine configuration, but due to a misconfiguration workload might access the machine configuration and reveal the join token. This problem has been fixed in Talos 1.2.2. Enabling the Pod Security Standards mitigates the vulnerability by denying hostPath mounts and host networking by default in the baseline policy. Clusters that don't run untrusted workloads are not affected. Clusters with correct Pod Security configurations which don't allow hostPath mounts, and secure access to cloud metadata server (or machine configuration is not supplied via cloud metadata server) are not affected.

ZITADEL combines the ease of Auth0 and the versatility of Keycloak.**Actions**, introduced in ZITADEL **1.42.0** on the API and **1.56.0** for Console, is a feature, where users with role.`ORG_OWNER` are able to create Javascript Code, which is invoked by the system at certain points during the login. **Actions**, for example, allow creating authorizations (user grants) on newly created users programmatically. Due to a missing authorization check, **Actions** were able to grant authorizations for projects that belong to other organizations inside the same Instance. Granting authorizations via API and Console is not affected by this vulnerability. There is currently no known workaround, users should update.

A vulnerability has been identified in SINEMA Remote Connect Server (All versions < V3.0). The webserver could allow unauthorized actions via special urls for unpriviledged users. The settings of the UMC authorization server could be changed to add a rogue server by an attacker authenticating with unprivilege user rights.

Vulnerability in the Oracle Database component of Oracle Database Server. Supported versions that are affected are 19.27 and 23.4-23.8. Easily exploitable vulnerability allows low privileged attacker having Create Session, Create Procedure privilege with network access via Oracle Net to compromise Oracle Database. Successful attacks of this vulnerability can result in takeover of Oracle Database. CVSS 3.1 Base Score 8.8 (Confidentiality, Integrity and Availability impacts). CVSS Vector: (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H).

Adobe Commerce versions 2.4.3-p2 (and earlier), 2.3.7-p3 (and earlier) and 2.4.4 (and earlier) are affected by an Improper Access Control vulnerability that could result in Privilege escalation. An attacker with a low privilege account could leverage this vulnerability to perform an account takeover for a victim. Exploitation of this issue does not require user interaction.

An improper access control vulnerability in GitLab CE/EE affecting all versions from 17.4 prior to 17.8.6, 17.9 prior to 17.9.3, and 17.10 prior to 17.10.1 allows a user who was an instance admin before but has since been downgraded to a regular user to continue to maintain elevated privileges to groups and projects.

Improper Authorization in Elastic Cloud Enterprise can lead to Privilege Escalation where the built-in readonly user can call APIs that should not be allowed. The list of APIs that are affected by this issue is: post:/platform/configuration/security/service-accounts delete:/platform/configuration/security/service-accounts/{user_id} patch:/platform/configuration/security/service-accounts/{user_id} post:/platform/configuration/security/service-accounts/{user_id}/keys delete:/platform/configuration/security/service-accounts/{user_id}/keys/{api_key_id} patch:/user post:/users post:/users/auth/keys delete:/users/auth/keys delete:/users/auth/keys/_all delete:/users/auth/keys/{api_key_id} delete:/users/{user_id}/auth/keys delete:/users/{user_id}/auth/keys/{api_key_id} delete:/users/{user_name} patch:/users/{user_name}

This vulnerability impacts all versions of IdentityIQ and allows an authenticated identity that is the requestor or assignee of a work item to edit the definition of a role without having an assigned capability that would allow role editing.

In S+ Operations and S+ Historian, not all client commands correctly check user permission as expected. Authenticated but Unauthorized remote users could execute a Denial-of-Service (DoS) attack, execute arbitrary code, or obtain more privilege than intended on the machines.

Capsule is a multi-tenancy and policy-based framework for Kubernetes. In Capsule v0.7.0 and earlier, the tenant-owner can patch any arbitrary namespace that has not been taken over by a tenant (i.e., namespaces without the ownerReference field), thereby gaining control of that namespace. Version 0.7.1 contains a patch.

Rescue Dispatch Management System 1.0 is vulnerable to Incorrect Access Control via http://localhost/rdms/admin/?page=system_info.

Jenkins Sounds Plugin 0.5 and earlier does not perform permission checks in URLs performing form validation, allowing attackers with Overall/Read access to execute arbitrary OS commands as the OS user account running Jenkins.