Team deletion in Midday's tRPC API grants destructive Owner-equivalent authority to any authenticated Member due to a missing role check in `apps/api/src/trpc/routers/team.ts`. The `team-access` authorization helper confirms team membership but never validates whether the caller holds the `owner` role, despite that check being correctly implemented in adjacent member-removal and member-update procedures within the same router. Any invited team member - receiving the `Member` role by default - can delete the entire team and all scoped records, and additionally trigger a background cleanup job that exercises stored bank-connection OAuth tokens against connected financial providers. No public exploit identified at time of analysis; no CISA KEV listing.
Missing authorization on DROP MATERIALIZED VIEW in StarRocks allows any authenticated account to permanently delete legacy synchronous materialized views across all databases, regardless of grants. The flaw exists because `AuthorizerStmtVisitor.visitDropMaterializedViewStatement` exits early without invoking the Authorizer; the authorization branch executes only when the resolved object is a modern `MaterializedView`, not when it is a legacy synchronous view stored as a rollup index on an `OlapTable`. Versions through 4.0.13 are affected per the VulnCheck advisory, and no public exploit has been identified at time of analysis.
Frappe's REST API before v15.115.0 contains an access control bypass (CWE-863) that allows any low-privileged authenticated user to read fields from Linked DocType records outside their permitted scope. When a document is fetched via the REST API, the framework resolves Link field references to other DocTypes but inconsistently re-applies the linked DocType's own permission rules, enabling horizontal data leakage across role boundaries. No confirmed active exploitation or public exploit code has been identified, but the low attack complexity and network-accessible vector create meaningful data exfiltration risk in multi-role or multi-tenant Frappe/ERPNext deployments.
OpenStack Keystone before 29.0.3 allows any authenticated user holding role:reader on any project to enumerate all role assignments across any domain in the cloud by exploiting a type-confusion bug in the GET /v3/role_assignments endpoint. The attack supplies a domain ID as scope.project.id with include_subtree, triggering a null domain_id policy check that passes unconditionally for all callers because domains are internally modeled as projects with domain_id=null. With include_names, the response discloses the names and home-domain IDs of every user, group, project, and role across the targeted domain - and the well-known 'default' domain constant works against every deployment bootstrapped with keystone-manage bootstrap, requiring no prior reconnaissance. No public exploit code has been identified at time of analysis, and the flaw is not listed in CISA KEV.
Incorrect authorization in SeaweedFS S3 API's PutObjectAcl handler allows an authenticated low-privilege S3 principal to overwrite objects outside their permission scope. Versions prior to 4.40 compute the write-back directory from the bucket root rather than the actual parent directory of the nested key, meaning a principal scoped to `allowed/protected.txt` silently overwrites `protected.txt` at the bucket root - clobbering its content, metadata, owner, and ACL. No public exploit code identified at time of analysis, and this CVE is not listed in CISA KEV.
Man-in-the-middle interception in CTFd v3.7.6, a widely-used Capture The Flag competition platform, exposes high-confidentiality data to adjacent-network attackers with no authentication required. Rooted in CWE-300 (Channel Accessible by Non-Endpoint), the flaw allows an attacker sharing the same network segment to intercept communications, capturing sensitive data such as session tokens or credentials, with limited integrity modification also possible. No active exploitation is confirmed - EPSS sits at 0.15% (4th percentile) and no CISA KEV listing exists - but the attack surface is acutely relevant in live CTF event environments where competitors and organizers routinely share the same LAN or Wi-Fi.
Data corruption in the Linux kernel's XFS exchange-range subsystem allows a low-privileged local attacker to trigger incorrect reflink inode flag clearing, causing subsequent writes to bypass Copy-on-Write protection on shared extents. Affected are Linux kernel versions prior to 6.12.105, 7.1.10, and 7.2 on systems using XFS volumes with reflink support enabled. No public exploit or CISA KEV listing exists at time of analysis, but the low-complexity local vector and high integrity/confidentiality impact make this a meaningful priority for XFS-heavy enterprise deployments such as RHEL-based systems where XFS is the default filesystem.
Path traversal in Mage AI through version 0.9.79 allows any authenticated Viewer-role user to read arbitrary files accessible to the server process by supplying absolute paths to the browser-items API endpoint. The BrowserItemResource class in mage_ai/api/resources/BrowserItemResource.py omits the containment check that sibling classes FileContentResource and FileResource both apply, and the role-based access control enforcement that would otherwise restrict this route is disabled by default. No public exploit or CISA KEV listing has been identified at time of analysis, but the low privilege bar and default-off RBAC enforcement make this a meaningful risk for any multi-user Mage AI deployment.
Memory leak in the Linux kernel s390/vfio_ccw subsystem enables a guest VM user to exhaust host kernel memory by submitting crafted channel programs that trigger cp_init() failure paths, leaving ccwchain segments unfreed. Affected deployments are IBM s390 (mainframe) KVM hosts that have vfio channel command word (CCW) device passthrough configured, making the attack surface highly specific. No public exploit has been identified and the vulnerability is not listed in the CISA KEV catalog; CVSS 7.1 with S:C reflects the guest-to-host cross-boundary availability impact.
Error propagation failure in the Linux kernel's XFS realtime filesystem subsystem corrupts filesystem state during xfs_growfs operations. The function xfs_rtginode_ensure() silently discards all non-ENOENT errors returned by xfs_rtginode_load(), leaving the realtime group inode pointer unset while growfs proceeds as though load succeeded - producing a null pointer dereference, kernel panic, or realtime subvolume metadata corruption. Patch commits targeting Linux 6.18.46, 7.1.10, and 7.2 are available via kernel.org stable branches; no public exploit or CISA KEV listing exists at time of analysis.
The SpacemiT K3 clock driver in the Linux kernel fails to mark the HDMA clock as CLK_IS_CRITICAL, allowing the kernel's Common Clock Framework to gate it during power management. On SpacemiT K3 SoCs running an X100 RISC-V core, this produces an irrecoverable system hang whenever TCM-dependent operations - such as page fault handling triggered by vector instructions - are attempted while the HDMA clock is inactive. No public exploit has been identified at time of analysis and exposure is strictly bounded to SpacemiT K3 hardware; upstream fix commits are confirmed available for affected stable branches.
Server-side request forgery in OneUptime's webhook delivery subsystem allows an authenticated project member to bypass the SSRF protection filter by supplying an IPv4-mapped IPv6 address literal such as [::ffff:127.0.0.1], directing the server to connect to loopback services, private network ranges, or cloud metadata endpoints. All OneUptime versions prior to 12.0.7 are affected; the underlying HTTP client treats the mapped form as the embedded IPv4 address, and the webhook response is returned to the attacker. No public exploit has been identified at time of analysis, but the bypass technique is well-documented in SSRF filter evasion literature and requires no special tooling.
LDAP injection in Apache APISIX's ldap-auth plugin allows a caller holding valid credentials for any LDAP directory entry to authenticate to the gateway as a different consumer - including entries outside the subtree the plugin was configured to restrict access to. Affected versions span 2.11.0 through 3.17.0, covering several years of releases. No public exploit code or active exploitation has been identified at time of analysis, but the primitive (credential-bearing lateral identity escalation at the API gateway layer) is high-value in multi-tenant or zero-trust deployments.
Response poisoning in Apache APISIX 2.12.0 through 3.17.0 enables remote attackers to manipulate the gateway's serverless plugin routes so that victims receive attacker-chosen content or other authenticated users' API responses. The root cause is inconsistent HTTP request boundary interpretation - the defining characteristic of HTTP Request/Response Smuggling - applied within APISIX's serverless plugin execution pipeline. No public exploit has been identified at time of analysis, and vendor-released patch version 3.18.0 is available.
Mailbox initialization in Trusted Firmware-M (TF-M) versions 2 through 2.3.0 on PSOC64 and RP2350 platforms fails to validate a pointer supplied by non-secure world code, allowing a low-privileged attacker with physical access to supply a crafted pointer that crosses the TrustZone-M hardware security boundary. Successful exploitation yields full read, write, and denial-of-service capability over the Secure Processing Environment (SPE), collapsing the hardware-enforced isolation TF-M is designed to guarantee - including potential exposure of cryptographic keys, attestation credentials, and secure boot state. No public exploit code exists and this vulnerability is absent from the CISA KEV catalog; the physical attack vector and high complexity substantially constrain opportunistic exploitation, but the full C/I/A scope change makes targeted supply-chain or insider-threat scenarios consequential.
Privilege escalation in FreeBSD's tty subsystem allows a local unprivileged user to gain elevated privileges by exploiting a race condition in the TIOCSCTTY ioctl handler. The handler fails to revalidate terminal state after a lock-drop sequence, enabling a concurrently destroyed terminal to be incorrectly linked to the attacker's session. No public exploit code has been identified at time of analysis, but vendor-confirmed patches exist for all affected release lines (15.0-RELEASE, 15.1-RELEASE, and 14.4-RELEASE).
Stored cross-site scripting in FiftyOne (all versions before v1.21.0) allows a dataset contributor to execute arbitrary JavaScript in the browser of any user who opens a poisoned dataset and hovers over a field with a malicious description. The payload persists in the dataset schema across exports, publications, and shares, giving it a supply-chain propagation vector common in ML/data-science workflows. When the script fires, it runs in the FiftyOne application origin, where it can leverage the open-source server's unauthenticated media route to read arbitrary local files from the victim's machine and interact with dataset and operator APIs as the authenticated user. No public exploit identified at time of analysis; a vendor-released patch exists in v1.21.0.