Missing rate limiting on second-factor (TOTP) code verification in the Roundcube twofactor_gauthenticator plugin lets someone who already holds a user's valid Roundcube password submit unlimited 6-digit codes until one is accepted, defeating the second factor and granting unauthorized access to the account. All plugin versions below commit 4640889 are affected per EUVD (0 < 4640889). The flaw only bypasses the second factor, so prior full password compromise is mandatory; the vendor's CVSS:4.0 score of 6.3 reflects AC:H/AT:P, and both the CVSS 4.0 and the assessed CVSS:3.1 vectors carry PR:N while exploitation in practice requires the victim's first factor. EPSS is low (0.37%, 30th percentile), there is no CISA KEV entry, and no public exploit code was identified at time of analysis; a fix is published as commit 4640889.
Insufficient session termination in the Roundcube twofactor_gauthenticator plugin (all builds prior to commit 4640889) lets an attacker who already holds a victim's valid account password ignore the client-side logout redirect after a failed second-factor check and continue using the still-valid authenticated session to brute-force the TOTP code. The plugin fails to destroy the session server-side when verification fails, so the 2FA layer is weakened rather than bypassed outright - the stolen password remains a hard prerequisite, and success is bounded by TOTP entropy and any rate limiting, which is why the assessment rates attack complexity high with low confidentiality, integrity and availability impact. No public exploit has been identified at time of analysis and EPSS is low (0.35%, 28th percentile); the vendor fix is commit 4640889 and the prioritisation label is this_month - a moderate, not emergency, priority.
Query string parameter injection in the npm qs library: when an application calls qs.stringify with the non-default option allowEmptyArrays: true, the key of an empty array is emitted without encoding, so an attacker-influenced key containing &, =, # or the configured delimiter becomes query string syntax rather than data - letting the caller add parameters to the serialized query string or truncate it at a fragment. All releases from 6.12.0 through 6.16.1 are affected (allowEmptyArrays did not exist before 6.12.0); the issue only bites when untrusted key names reach stringify and at least one of them maps to an empty array, and deployments that leave the option at its default are unaffected. The vendor has released patches in 6.12.6, 6.13.4, 6.14.3, 6.15.4, 6.16.2 and 6.17.0; EPSS is low at 0.23% (14th percentile), there is no CISA KEV entry, and the advisory includes a public PoC reference that is not verified as working exploit code. The assessed vector (CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:L/A:N) rates impact as integrity-only and modest.
Improper privilege management in Dell PowerProtect Data Domain allows a high-privileged attacker with local access to the appliance to escalate privileges, with assessed impact to integrity and availability rated High and confidentiality Low (CVSS 6.3). Affected builds are Feature Release 7.7.1.0 through 8.7, LTS2024 7.13.1.0 through 7.13.1.80, LTS2025 8.3.1.0 through 8.3.1.40 and LTS2026 8.6.1.0 through 8.6.1.20; because exploitation requires both an existing high-privileged account and local access, real-world exposure is largely confined to insiders or already-compromised privileged accounts. No public exploit code or confirmed active exploitation was identified at time of analysis, and vendor fixes are available.
Cleartext transmission in fclairamb/ftpserverlib through 0.32.4 lets an on-path attacker read or modify transferred file contents and directory listings even when the server enforces encrypted data transfers. The flaw is reachable by a client that prepares a data connection with PASV or PORT while a cleartext protection level is in force and only then issues PROT P, because TransferOpen trusts the transferTLS flag instead of inspecting whether the prepared connection is actually a TLS connection; no authentication or user interaction is needed, but the attacker must already hold a man-in-the-middle position on the data channel. Severity is medium (assessed CVSS 3.1 AV:N/AC:H/PR:N/UI:N/C:L/I:L/A:N), EPSS is negligible at 0.09% with no KEV listing and no public exploit identified at time of analysis, and an upstream fix is proposed in pull request 685.
Improper link resolution in the Tanium Threat Response Client Extension (threat-response-cx and the Endpoint Configuration Toolset Solution) allows an authenticated local attacker with low privileges to follow a manipulated link and read or write files outside the intended location, producing high confidentiality and integrity impact with no availability effect. Affected builds include threat-response-cx releases prior to 1.17.802, 1.18.2732, 1.19.2213, 1.20.2271 and 1.21.1673, and Endpoint Configuration Toolset Solution releases prior to 1.51.60, 1.59.50 and 1.71.25. Exploitation is local-only and additionally depends on an unspecified high-complexity condition, so realistic abuse is limited to attackers who already hold a foothold on the endpoint; no public exploit code has been identified at time of analysis.
Authenticated SQL injection in Tanium Provision can expose or modify data and cause limited service disruption for a low-privileged user who already holds a valid session. Provision 1.17 before 1.17.150, 1.23 before 1.23.161, and 1.24 before 1.24.199 are affected, and Tanium has released fixed builds for each branch. The vendor rates the issue CVSS 6.3 with PR:L and UI:N, the vulnerable endpoint or parameter is not identified in the available sources, and no public exploit was identified at time of analysis.
GFI KerioConnect Webmail before 10.0.8 Patch 2 is affected by stored cross-site scripting in which a crafted SVG email attachment bypasses sanitization of event-handler attributes and executes JavaScript when the victim opens it, with the temporary-download endpoint's insufficient Content-Security-Policy sandbox failing to block script execution. Successful exploitation can steal credentials and mailbox contents from an authenticated Webmail session; the supplied CVE description attributes an exploitation observation to Proofpoint Threat Research, while no CISA KEV status is included in the provided intelligence. The independent assessment rates this as a Medium-severity issue (CVSS 4.0 6.2) because the victim must actively open the attachment and impact is scored against the victim's browser/Webmail session rather than the mail server itself.
Path traversal in the model persistor of crownpku Rasa_NLU_Chi (versions 0.12.0 through 0.12.2) lets an attacker who can influence a model archive that the Persistor later decompresses write outside the intended extraction directory, with low confidentiality, integrity and availability impact. The CVSS 4.0 vector (2.1, AV:N/AC:L/PR:L/UI:N) requires low-privilege access and the ability to supply or alter the model artifact being loaded or restored rather than simple anonymous reach. A public PoC reference exists; it is not verified as working exploit code, and there is no confirmed active exploitation (not in CISA KEV). The project was notified through GitHub issue #138 and has not responded, and no vendor-released patch has been identified at time of analysis, so the practical exposure is limited to deployments running this unmaintained fork.
Stack-based buffer overflow in WatchGuard Fireware OS lets an authenticated mobile VPN user crash the VPN Network Access Enforcement process and potentially run arbitrary code as an unprivileged process by sending a crafted request. Mobile VPN must be configured and valid VPN credentials are required; the independent assessment rates this as moderate priority, not emergency-tier, with denial of service as the primary assured impact and code execution only potential. Affected releases are Fireware OS 2025.1 before 2026.4 and 12.9 before 12.12.4; a vendor patch is available, and no public exploit identified at time of analysis.
MISP Guard releases before 1.4 enforce required-taxonomy rules incompletely, so a feed proxy configured to distribute only lower-classification data can be made to pass events and attributes marked with a higher classification. An authenticated user able to author or modify events in a proxied feed source can attach an allowed tag (for example tlp:clear) together with a non-allowed tag of the same taxonomy (tlp:red or tlp:amber) to an event, an attribute, an attribute nested in an object, or a shadow attribute, and the guard's policy check still permits distribution - disclosing more sensitive data to feed consumers than policy intends. No public exploit has been identified at time of analysis; this is a moderate-severity data-handling bypass rather than a high-urgency emergency, and the upstream fix is commit 6127d94c350e4dfe31fd4426ba26f51294ed3867 with the affected range listed as misp-guard 0 to <1.4.
Administrator account takeover in TP-Link RE305 V3 Wi-Fi range extenders: the device's unauthenticated password-recovery verification-code interface can be driven by anyone positioned on the same LAN or Wi-Fi to generate an administrator password-reset code, and the code can then be obtained or brute-forced. Two weaknesses enable this - the interface has no effective attempt counter or lockout, and the verification code is produced by a predictable generation process (CWE-338). Exploitation requires an adjacent-network position (the vector is AV:A, so this is not reachable from the internet) and a brute-force or prediction effort (AC:H); no public exploit or PoC has been identified at time of analysis and the CVE is not in CISA KEV. Successful abuse lets the attacker reset the admin password and reach privileged device-management functions, with high integrity impact but limited confidentiality and availability impact.
Improper validation of an integrity check value in VideoLLaMA2's model-loading routine (load_pretrained_model in videollama2/model/__init__.py) allows a tampered model artifact to be accepted when an operator loads a model from an attacker-influenced path supplied through the _name_or_path argument. Versions 1.0 and 1.1 are affected; because the flaw lives in a library function rather than an always-on service, exploitation requires passive user interaction - someone must actively initiate a load from the hostile source - and impact is bounded to low integrity and low availability with no confidentiality effect. A public reference to the issue exists (the upstream GitHub issue report), but it is not verified as working exploit code, there is no CISA KEV listing, and EPSS is 0.15% (4th percentile), so all available signals point to a low real-world priority and a monitor-level decision.
Chitchatter's embedded SDK mode can be tricked into accepting forged configuration and identity keys from an attacker-controlled parent frame, because the postMessage CONFIG handler in src/Bootstrap.tsx fails to properly authenticate the sender when the supplied parentDomain is manipulated (CWE-287). Only deployments running Chitchatter inside an iframe with the embedded/SDK flags set are exposed; a victim has to load or interact with the attacker's embedding page (CVSS 3.1 AV:N/AC:L/PR:N/UI:R/S:U/C:L/I:L/A:N, low severity), while standalone Chitchatter use is not affected. The flaw was fixed in upstream commit e3babb5cd83bfe3b99f1d82358ac5c1ea8e8a688 (PR #635), and a public PoC reference exists but is not verified as working exploit code; there is no CISA KEV entry, and the current triage decision is monitor.
Path traversal in the Dexbotic gr00tsonic Checkpoint Loader (dexbotic/model/gr00tsonic/gr00tsonic_arch.py, function load_pretrained_gr00t) allows an attacker who controls the weight_map argument to cause file paths outside the intended checkpoint location to be used during checkpoint loading. Dexbotic 0.1 and 0.2.0 are listed as affected (CPE cpe:2.3:a:dexmal:dexbotic:*), and exploitation is remote but requires the victim to actually load an attacker-influenced pretrained checkpoint - per the assessed CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:L/I:L/A:L vector no authentication is needed, while passive user interaction is the primary limiting factor. There is no CISA KEV entry and no confirmed active exploitation; a public PoC reference exists but it is not verified as working exploit code, and the project has been notified through an issue report without a response, so the issue is assessed as low priority (monitor).
Reflected cross-site scripting in Kulga News Software from v26.7.4 before v26.10.0 allows remote attackers to execute script in a victim's browser session when the victim opens an attacker-crafted URL, requiring user interaction but no authentication. The flaw is rated CVSS 6.1, with low confidentiality and integrity impact and no availability impact, and exploitation is limited to the victim's browser session. No public exploit identified at time of analysis; the vendor fix is v26.10.0.
Missing authorization in REDAXO core's package management API (rex_api_package, reached via rex-api-call=package) allows a backend user without admin rights to uninstall, deactivate, install, activate, or delete packages as long as a valid CSRF token for that API function is present in their session. The most damaging outcome is uninstalling a system addon such as structure, whose uninstall routine drops its database tables, thereby destroying all articles and their content. Exploitation is conditional rather than broadly reachable: the token is only generated on administrator-restricted backend pages, so the realistic attacker is a demoted admin whose session is still active, or someone who obtained the token through a separate flaw, and live-mode installations are not affected at all. Severity is medium (vendor CVSS 4.0 base 6.1); no public exploit code and no confirmed active exploitation were identified at time of analysis, and the vendor has released 5.21.5 as the fix.
Cross-site scripting in the chat.php endpoint of Saruf-Ratul's HR-Payroll-Management-System allows an attacker to inject script through the ID argument, with execution occurring in a logged-in victim's browser after that user is induced to open a crafted request or link. The flaw is remotely reachable but requires a low-privileged authenticated session (PR:L) plus victim interaction (UI:P), and the assessed impact is limited to script execution in the victim's browser context with no confidentiality or availability impact as scored. A public PoC reference exists in the project's issue tracker; it is not verified as working exploit code, CISA KEV does not list the product, EPSS is 0.20% (10th percentile), SSVC rates exploitation as 'poc' with partial technical impact, and the tracking decision is monitor.
Reflected cross-site scripting in the chat.php script of MD-MAFUJUL-HASAN Online-Payroll-Management-System allows an authenticated low-privileged user to manipulate the ID argument so that attacker-supplied script runs in a victim's browser when the victim follows a crafted link. Scoring is low: the CVE carries CVSS 4.0 base 2.0 and our independent assessment rates it CVSS 3.1 AV:N/AC:L/PR:L/UI:R/S:C/C:N/I:L/A:N, with impact confined to limited integrity change and no confidentiality or availability effect. A public PoC reference exists at the project's GitHub issue tracker; it is not verified as working exploit code, EPSS is 0.20% (10th percentile), there is no CISA KEV entry and no confirmed active exploitation, and the maintainer has not responded to the report, so no fixed release is available for this rolling-release project.
NetApp ONTAP 9 (before 9.18.1), StorageGRID (through 12.2.0) and affected E-Series SANtricity OS Controller Software builds can still trust eight eMudhra intermediate CA certificates whose private keys were compromised and which eMudhra revoked on October 18, 2026. An attacker holding one of those compromised intermediate keys and positioned to intercept traffic could mint a certificate that chains to a trusted-but-revoked intermediate and have it accepted by the product, exposing confidential data (confidentiality-only impact). The assessed path is non-trivial - it requires an interception position, a victim connecting to the forged endpoint, and AC:H - and no public exploit code has been identified at time of analysis. NetApp tracks the required trust-store changes under this CVE and has published an advisory with vendor-provided updates.
WatchGuard Fireware OS leaves the FireCluster state-synchronization exchange without authentication, so an attacker who can place crafted traffic on the cluster sync segment can force premature termination of VPN tunnels carried by the firewall. Only deployments with FireCluster high-availability enabled and a sync link reachable by the attacker are exposed; an unauthenticated, no-interaction attacker on that adjacent segment is sufficient, per the assessed vector (AV:A/AC:L/PR:N/UI:N). Impact is availability-only (forced tunnel drops plus limited integrity impact), confidentiality is unaffected, and no public exploit has been identified at time of analysis - the practical barrier is network segmentation, since sync links are normally pinned to a dedicated isolated interface or VLAN.
Dell PowerProtect Data Domain appliances are exposed to a path traversal flaw (CWE-28) that a high-privileged local attacker can use to disrupt the appliance, with availability impact as the dominant outcome. Affected trains are Feature Release 7.7.1.0 through 8.8, LTS2024 7.13.1.0 through 7.13.1.80, LTS2025 8.3.1.0 through 8.3.1.40, and LTS2026 8.6.1.0 through 8.6.1.20. Exploitation requires local access to the appliance plus administrative privileges (AV:L/PR:H, no user interaction), so the realistic threat model is an existing administrator or an attacker who already holds a privileged account; no public exploit has been identified at time of analysis, and the vendor has released fixed builds.
A heap buffer overflow in the Codec2 transcoding module (codec_codec2) of Asterisk can be triggered by an authenticated SIP caller who negotiates the Codec2 codec and then sends a crafted audio frame, resulting in an out-of-bounds write that crashes the channel/process and denies service. Affected are Asterisk 20.x before 20.20.1, 21.x before 21.12.3, 22.x before 22.10.1, 23.x before 23.4.1, plus Certified Asterisk before 20.7-cert11 and before 22.8-cert3; exploitation requires the codec_codec2 module to be loaded and Codec2 explicitly configured on the targeted endpoint, which is not a default configuration. The assessed impact is availability-only (crash/denial of service) against a medium-severity baseline; no public exploit code and no confirmed active exploitation were identified at time of analysis, and no EPSS score was supplied in the available data.
Improper access controls in Tanium Comply allow a low-privileged authenticated network user to reach functionality or data outside their intended permissions, with limited confidentiality, integrity, and availability impact and a scope change. Affected builds are Comply 2.32 before 2.32.263, 2.35 before 2.35.323, and 2.37 before 2.37.326; Tanium has released fixed builds in each branch. No public exploit has been identified at time of analysis, and the vendor's high attack-complexity rating keeps this a moderate-severity, lower-urgency issue.
Arbitrary file write in Veeam Backup & Replication 13.0 (before 13.0.3) and 13.1 (before 13.1.2) lets an authenticated user who holds the Restore Operator role and who also controls the contents of a NAS share that the server backs up plant attacker-chosen files at an arbitrary path on the backup server via relative path traversal. Both preconditions are confirmed attack requirements: a privileged, non-default backup role and write influence over backed-up NAS data, which confines realistic exposure to malicious insiders or attackers who have already compromised a privileged backup account or a NAS data source. No public exploit has been identified at time of analysis; the flaw is rated moderate severity (CVSS 4.0 base 5.9, assessed CVSS 3.1 6.4-equivalent vector AV:N/AC:H/PR:H/UI:N with high integrity impact and limited availability impact), and the vendor has released fixed builds.
Arbitrary file write in Veeam Backup & Replication 13.x lets an authenticated holder of the Restore Operator role, who also controls the contents of a NAS share that the backup server protects, place attacker-authored files at an arbitrary path on the backup server via relative path traversal. Only Backup and Replication 13.0 before 13.0.3 and 13.1 before 13.1.2 are affected, and the assessed vector (CVSS:3.1/AV:N/AC:H/PR:H/UI:N/S:U/C:N/I:H/A:L) reflects a high-barrier integrity impact with no confidentiality loss and limited availability impact, since exploitation needs both the delegated role and a position of control over backup source data. This is a medium-severity (CVSS 4.0 base 5.9), high-barrier integrity flaw, not an emergency; no public exploit was identified at time of analysis and the issue is not in CISA KEV, with vendor fixes available in 13.0.3 and 13.1.2.
Cross-origin credential leakage in follow-redirects 1.16.0 through 1.16.1 lets a custom credential header named in the sensitiveHeaders option be forwarded to an attacker-controlled redirect destination when that option is supplied as a genuine cross-realm array. Only direct callers or wrappers that pass a cross-realm array (for example one created via node:vm) are affected - same-realm arrays are handled correctly, built-in Authorization, Proxy-Authorization and Cookie stripping remained effective, and callers such as Axios 1.20.0 that normalize the array did not reproduce. The vendor released a patch in 1.16.2, and the advisory describes a loopback proof-of-concept harness that is held privately; there is no confirmed active exploitation (not in CISA KEV), and this is best treated as a conditional medium-severity disclosure rather than an urgent default-configuration issue.
Code injection in the Remotedev Server component of redux-devtools 5.0.0 allows a locally positioned attacker with low-level privileges to inject code through a crafted value passed to the action argument of the evalAction function. Because the attack vector is local (AV:L) and requires pre-existing privileges (PR:L), the practical exposure is confined to developer and debugging environments where the Remotedev Server is reachable by an already-present user; no user interaction is needed. The flaw was reported to the project through a GitHub issue (reduxjs/redux-devtools#2125) but no vendor response or patch has been identified, and a public PoC reference exists that is not verified as working exploit code - no confirmed active exploitation (CISA KEV) was recorded, and the priority assessment is 'monitor'.
An out-of-bounds write in the chan_unistim channel driver lets a remote attacker crash Asterisk by sending a crafted request packet that overflows the internal phone_number field during DIALPAGE digit handling; the flaw is contained to a denial of service with no confidentiality or integrity impact (C:N/I:N/A:H). Only deployments that load chan_unistim and configure a reachable UNISTIM UDP port are exposed, and the attacker must also be able to register as a configured phone MAC or reach the main page through an equivalent autoprovisioning path, which keeps the assessed severity at 5.9 with AC:H. The issue is fixed in Asterisk 20.20.1, 21.12.3, 22.10.1, 23.4.1 and Certified Asterisk 20.7-cert11 and 22.8-cert3; no public exploit code or CISA KEV entry was identified at time of analysis.
Eaton easySoft releases below 8.46 fail to properly validate the TLS certificate presented by easyE4 devices, allowing an attacker who already holds a man-in-the-middle position on the network path between the engineering client and the controller to intercept or modify that session. The assessed vector (AV:A/AC:H/PR:N/UI:R/S:U/C:L/I:L/A:N) reflects the real gating factors - an adjacent MITM position plus an engineer initiating the connection - which keeps impact to partial confidentiality and integrity loss rather than device takeover. No public exploit code was identified at time of analysis, the vendor has released easySoft 8.46 as the fix, and the independent assessment rates this a moderate, not an emergency, priority.
Server-side request forgery in Vaultwarden 1.36.0 and earlier allows a remote unauthenticated attacker to steer the server's built-in icon downloader at loopback, link-local (169.254.169.254) or private-range addresses by embedding an IP-literal favicon URL such as http://127.0.0.1:PORT/favicon.ico in an otherwise benign public page. The flaw is a bypass of the fix for CVE-2026-47160: the initial icon host and DNS answers are validated, but discovered favicon URLs are fetched through get_page_with_referer() without should_block_host(), and IP-literal URLs never invoke reqwest's custom DNS resolver, so the post_resolve() allowlist is skipped entirely. Exposure is limited to instances with internal icon downloading enabled, impact is capped at limited confidentiality (internal responses that resemble supported images, plus reachability inference via timing, status and cache behaviour), and no public exploit code or confirmed active exploitation was identified at time of analysis; a vendor patch exists in 1.37.0.
Server-side request forgery in Dell PowerProtect Data Domain lets a remote attacker who can reach the appliance's exposed service make the system issue requests on their behalf, with no authentication or user interaction required (CVSS 3.1: AV:N/AC:L/PR:N/UI:N). The flaw carries a changed-scope rating because the impact falls outside the vulnerable component's own security boundary, but the direct consequence is limited to a low confidentiality effect (C:L, I:N, A:N), with CVSS base score 5.8. Affected branches are Feature Release 7.7.1.0 through 8.7, LTS2024 7.13.1.0 through 7.13.1.80, LTS2025 8.3.1.0 through 8.3.1.40 and LTS2026 8.6.1.0 through 8.6.1.20; vendor-released fixes exist and no public exploit has been identified at time of analysis.
Reflected cross-site scripting in the WebSub intent-verification flow lets an attacker execute script in a subscriber's origin by getting a subscriber callback endpoint to echo an attacker-supplied hub.challenge (hub_challenge) value back with a dangerous media type - text/html, the default in many web frameworks, or application/octet-stream when MIME sniffing is still enabled. Affected are WebSub subscriber implementations conforming to W3C recommendation REC-websub-20180123, including the reference demonstration hub at websub.rocks; the attacker must first register a subscription with their own session to obtain a valid callback URL, and the victim must hold an active session on that subscriber and follow a crafted verification URL (UI:R, PR:L). Vendor CVSS 3.1 is 5.7 (medium band) and the vuln.today decision label is 'this_month', making this a schedule-and-fix item rather than an emergency; there is no CISA KEV entry and no confirmed active exploitation, and while a public PoC reference exists - reflection was reproduced on the WebSub Rocks validator, which the advisory says does not establish practical impact for that site - it is not verified as working exploit code against a production subscriber.
Session hijacking in WatchGuard Fireware OS allows a local, unauthenticated attacker on the appliance to take over another user's in-progress AuthPoint MFA session by sending a crafted request and persuading the victim to approve an attacker-triggered push notification. The flaw affects Fireware OS 12.7 before 12.12.4 and 2025.1 before 2026.4 where AuthPoint MFA is configured, and exploitation is constrained by the need for local access, a live victim authentication attempt, and user interaction. No public exploit was identified at the time of analysis, and the issue is assessed as medium severity rather than an emergency.
WatchGuard Mobile VPN with SSL Client for Windows before 2026.4 exposes VPN usernames and passwords by transmitting them in cleartext over the loopback interface during authentication, allowing a local attacker who can capture loopback traffic to recover credentials. Exploitation requires an existing foothold on the endpoint and capture during an active VPN login, and no public exploit identified at time of analysis; the vendor has released version 2026.4 to address the issue.
Docker Desktop before 4.97.0 can write third-party API keys and OAuth access tokens in cleartext to local application logs, crash dumps, and diagnostics bundles because IPC payload redaction is insufficient. An attacker who can read those files - or receives a shared diagnostics bundle - may recover credentials that the victim actually used in Docker Desktop; Docker account tokens are not affected. No public exploit has been identified at time of analysis, and the vendor CVSS 5.7 / assessed CVSS:3.1 AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:N/A:N reflects a local, confidentiality-only exposure; Docker Desktop 4.94.0 and 4.95.0 addressed known logging paths but did not remove previously logged secrets, while 4.97.0 fixes the issue and cleans existing host logs and crash dumps.
SourceCodester Online Food Ordering System 1.0 ships an administrative endpoint, admin/delete_admin.php, that performs no authentication or authorization check before acting on a supplied ID parameter, so an unauthenticated remote attacker can delete administrator records outright. The CVSS 4.0 vector scores it 5.5 with PR:N/UI:N, and the independent assessment concurs that no login, no user interaction and no special configuration are required - only network reachability of the admin path. A public proof-of-concept reference has been published (a GitHub repository addressing authentication on the admin delete endpoints); it is not verified as working exploit code, and there is no CISA KEV entry, while EPSS sits at 0.57% (47th percentile) and the vuln.today decision label is 'this_month', reflecting a genuine but modest-priority issue limited to integrity and availability of admin records.
An out-of-bounds read (CWE-125) in Adobe Substance 3D Painter, affecting releases up to and including 12.1.5, can leak sensitive contents of process memory when a user opens a crafted file. It is scored CVSS 5.5 (AV:L/AC:L/PR:N/UI:R, confidentiality-only impact), meaning disclosure is gated by a local, unauthenticated attacker plus a victim opening a malicious asset; there is no network-triggerable path and no code execution, data modification, or denial of service is established. No public exploit code and no CISA KEV entry are associated with this CVE at time of analysis, and the vuln.today priority decision is 'this_month'.
Local denial of service in the Tanium Performance client extension (performance-cx) and the Tanium Endpoint Configuration Toolset Solution lets a user who already holds low-level privileges on an endpoint exhaust resources in the agent, degrading or halting its availability with no confidentiality or integrity impact. The flaw is availability-only (CVSS 5.5, AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H) and requires local access to the host, so it is a moderate-severity issue rather than a top-tier priority. Tanium has released fixed builds and no public exploit was identified at time of analysis; priority is this_month.
ML-KEM key-pair lookups in Eclipse KeyPont crypto11 v2 (2.0.0-rc3 through 2.0.0-rc8) bind a public key to a private key without verifying that the two actually belong together, so Encapsulate() can encrypt a fresh shared secret to a public key whose owner is not the holder of the private key the caller looked up, letting that third party recover the secret. Two paths lead there: an attacker able to open a session on the same PKCS#11 token can plant a CKK_ML_KEM public object carrying the victim's CKA_ID/CKA_LABEL, or, when an ML-KEM private key has a label but no CKA_ID and no matching public object, an 'ID only' fallback search degrades to a class-and-type template and silently returns whichever ML-KEM public key the token lists first. The advisory documents a reproduction on SoftHSMv2 and SoftHSMv3; the issue is not listed in CISA KEV, the CVSS vector is local, low-privilege (AV:L/PR:L) with high confidentiality impact, and v1 module paths are unaffected because they have no ML-KEM support. Overall the risk is moderate and conditional: it requires a shared token, an application invoking the ML-KEM lookup and encapsulation path, and either a planting-capable session or a label-only key layout.
OS command injection in the Eclipse ioFog iofogctl CLI lets whoever can alter the ioFog Controller API response slip shell metacharacters into the provisioning key or CA certificate, which the CLI concatenates into the `edgelet provision` / `edgelet config cert` (or legacy `iofog-agent provision` / `iofog-agent cert`) commands it runs over SSH on the agent host and in the local shell. The issue affects iofogctl 1.0.0 through 3.7.0 and 3.8.0-rc.1 through 3.9.1, and triggering it requires an operator to actively deploy or attach an agent (user interaction) while talking to a Controller whose response has been tampered with; a trusted, authentic Controller is not vulnerable to this path. There is no CISA KEV entry and no public exploit identified at time of analysis, and the vendor fixed the flaw in iofogctl 3.9.2.
Psono Server (Community and Enterprise editions) 1.0.0 through 7.2.3 fails to enforce the direction of file-transfer credentials, letting an authenticated user who holds only read-only download access to a file-repository object obtain a signed upload URL for an attacker-chosen object key and append data to the owner's file. The result is limited-integrity and limited-availability impact - repository objects can be uploaded over or overwritten and storage consumed - bounded by server-side transfer size, chunk-count and lifetime limits (CVSS 5.4, PR:L, no user interaction). Exploitation requires an existing Psono account, prior read access to the targeted object, and the file-repository feature being in use; no public exploit code was identified at time of analysis, EPSS is low at 0.34% (27th percentile), the issue is not in CISA KEV, and vendor-released patch 7.2.4 is available, which places this in a this-month rather than emergency remediation window.
Server-side request forgery (CWE-918) in Psono Server CE and EE versions 0.1.0 through 7.2.3 allows an authenticated remote user to supply an arbitrary host when configuring a personal Duo MFA device, which the server then passes to the Duo client without a host allowlist or explicit timeout. Because connections still require valid TLS certificates and are confined to HTTPS on port 443, an attacker gains only a failure-mode oracle for internal DNS resolution and service reachability, plus the ability to tie up a server worker against unresponsive targets until the operating-system timeout. No public exploit is identified at time of analysis, EPSS is low at 0.24% (15th percentile), and the risk posture aligns with patching on a normal cycle; the vendor has released 7.2.4 with the fix.
An out-of-bounds read in the QAuthenticator NTLM handling of the Qt Network module lets a malicious or on-path server or proxy return a crafted NTLM challenge that leaks heap memory contents back toward the attacker or crashes the application outright. Only applications built on Qt Network that actually perform NTLM authentication through QNetworkAccessManager or QTcpSocket are exposed, and the victim application must itself initiate the authentication (UI:A / AT:P), so no attacker credentials are needed (PR:N) but a passive target cannot be triggered. Qt 6.8.0 through 6.8.9 and 6.9.0 through 6.11.1 are affected, with fixes released in 6.8.10 and 6.11.2; there is no CISA KEV listing, EPSS is 0.46% (39th percentile), and the CVSS 4.0 vector carries an exploit maturity rating of Proof-of-Concept (E:P), though no confirmed working exploit code has been identified at time of analysis.
CSS sanitizer bypass in Crass, the Ruby CSS parser, affects versions prior to 1.0.8 and allows unauthenticated remote attackers to smuggle declarations, style rules, or at-rules past sanitizers that parse untrusted CSS with Crass and re-serialize it without rejecting bad URL tokens. Confirmed downstream impact requires a vulnerable sanitizer such as Sanitize through 7.0.0, Loofah's scrub_css, or rails-html-sanitizer, and a victim rendering the sanitized output; the result is limited to UI redress, external resource loading, or CSS-based data exfiltration, not script execution in modern browsers. No public exploit identified at time of analysis, and the vendor has released Crass 1.0.8 as the fix.
Heap out-of-bounds read in zchunk before 1.5.5 allows unauthenticated remote attackers to crash an application that parses a crafted .zck file, with user interaction required because the victim must ingest the malicious archive. The flaw is in read_preface() in src/lib/header.c, which passes the total header length instead of the remaining byte count to compressed-integer parsing, so a header truncated inside a compressed integer causes reads past the heap header buffer. Moderate severity (CVSS 5.3), availability-only impact; no public exploit identified at time of analysis, and vendor-released patch 1.5.5 is available.
Unbounded memory growth in the a2a-java SDK's per-task ChildQueue lets a remote, low-privileged authenticated client exhaust server memory by enqueuing streaming events faster than a slow or stalled subscriber drains them, degrading availability of the affected a2a-java server. Versions up to and including 1.2.0.Final are affected; the flaw is CWE-400 resource consumption with a low availability-only impact and no data exposure. A vendor fix, PR #1040 (commit be663f806efcfdddd2805d214fae4998379ac5f2), ships in release 1.3.0.Final; no public exploit code was identified at time of analysis and no EPSS score was supplied.
Nextcloud Server 32.0.0 through 34.0.0 accepts an application-specific app password as sufficient proof for the password-confirmation (re-authentication) gate that protects sensitive endpoints, so anyone holding a valid app-password token can perform integrity-affecting privileged actions that should have required the real account password. The concrete prerequisite is possession of such a token (the flaw is rated PR:L), which typically means a leaked, over-scoped or stale app password rather than an unauthenticated attacker. Severity is moderate (CVSS 5.3 vendor score, 5.3 equivalent under our assessment, which raises AC:H to AC:L because no unusual conditions beyond holding the token are needed); there is no CISA KEV entry and no public exploit code identified at time of analysis, only the original HackerOne report, so this is a this-month priority rather than an emergency.
Cross-site request forgery in Phoca Download for Joomla 3.1.0 through 6.1.5 allows a remote attacker to induce a logged-in administrator or category-management user to submit forged category-management actions, with limited integrity and availability impact and no confidentiality disclosure. The CVSS vector is network-reachable with no attacker privileges, but exploitation depends on a privileged victim's active browser session and interaction; EPSS is 0.21% (12th percentile), no public exploit is identified at time of analysis, and the priority is this_month. This is a moderate, low-priority issue rather than an emergency.
Reflected cross-site scripting in SemanticMediaWiki 3.1.0 through 7.3.0 lets an unauthenticated remote attacker execute script in a wiki's origin when a victim opens a crafted Special:Search URL that uses the smw profile with an SVG event-handler payload, because ProfileForm::profile_sheet writes the search query into a pre element with Html::rawElement instead of escaping it. Exposure is conditional: the operator must have enabled SemanticMediaWiki's optional extended search engine ($wgSearchType = SMW\MediaWiki\Search\ExtendedSearchEngine::class or the legacy SMWSearch alias), so wikis running MediaWiki's default search are not affected on this path, and the attack is reflected rather than stored, so a visitor must follow the attacker-supplied link. A public PoC reference exists in the form of a raw HTTP request shape plus a marker-only JSDOM validation in a disposable fixture; it is not verified as working exploit code beyond that confined test, there is no CISA KEV entry, and the moderate CVSS 4.0 score of 5.3 reflects limited confidentiality and integrity impact inside the victim's browser rather than a demonstrated account takeover.