OpenWISP Controller versions 1.0.0 through 1.3.1 contain a missing-authorization flaw (CWE-862) in the CRL download endpoint: an authenticated organization manager holding CA view permission can retrieve another organization's certificate revocation list by iterating sequential CA identifiers. The exposure is limited to CRL contents - CA issuer information and revoked certificate serial numbers - with no private-key disclosure and no ability to modify data, and the sibling CA detail endpoint remains correctly scoped. It is fixed in 1.3.2; exploitation requires a multi-tenant deployment, and because django-x509 serves CRLs publicly by default unless DJANGO_X509_CRL_PROTECTED is enabled, only deployments that explicitly protect CRLs gain a new confidentiality exposure. Severity is low (CVSS 5.3 as published; assessed 3.1 vector AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:N), no public exploit code is identified, and there is no KEV listing.
The Payment Integration Wompi plugin for WordPress (through version 4.0.1) accepts incoming payment notifications without proving they came from Wompi when the merchant secret has not yet been set, because the callback checksum is keyed on that secret while it is still empty; an unauthenticated remote attacker can therefore forge a notification and flip an arbitrary order to paid. The exposure is conditional on configuration: on a stock, unconfigured install any remote user with no credentials and no user interaction can perform the forgery, but once the operator sets a non-empty merchant secret the checksum verification works and the forged callback is rejected. Scoring is modest (CVSS 5.3, integrity-only, EPSS 0.17% at the 7th percentile), there is no CISA KEV entry, and no public exploit code was identified at the time of analysis; the vuln.today decision label is this_month.
The country/region configuration interface of the TP-Link RE305 V3 range extender lacks an authorization check, so anyone who can reach it from the same local network - joined to its Wi-Fi or on the same LAN - can rewrite the wireless regulatory region and force the device to reapply its wireless configuration, with no credentials or user interaction needed. The practical result is limited to wireless availability: interrupted connections, unavailable channels, client disconnections and unstable service, plus operating parameters meant for a different regulatory domain; there is no loss of confidentiality or of device-data integrity. Exploitation is constrained to the adjacent network segment rather than the internet, so exposure depends on who can join that local wireless or wired network. No public exploit code was identified at time of analysis, and TP-Link has released fixed firmware.
Reflected cross-site scripting in Semantic MediaWiki 3.0.0 through 7.3.0 lets remote attackers execute JavaScript in a wiki's origin by luring a victim to a crafted Special:Ask link that carries an unescaped separator parameter (sep, propsep, valuesep, delim, or separator) alongside request_type=raw or request_type=special_page. No attacker authentication is required, but the victim must click the link, so the practical blast radius depends on whether an administrator or other privileged user is targeted; successful script execution can read session cookies, exfiltrate CSRF tokens and perform actions as the victim. The vulnerable remote request path is enabled by default, and the flaw is a bypass of the earlier table-printer fix (GHSA-7xv3-gf2g-498h), which hardened only TableResultPrinter while six equivalent sinks in the list, category, DSV and CSV printers were left unescaped; the vendor has released 7.3.1, and a proof-of-concept request is published in the GitHub advisory. This is a genuine but moderate-priority issue for this month's patch cycle rather than a drop-everything emergency, and no confirmed active exploitation has been reported.
Reflected cross-site scripting in Semantic MediaWiki (Composer package mediawiki/semantic-media-wiki) versions 3.1.0 through 7.3.0 lets an unauthenticated attacker execute arbitrary JavaScript in a wiki's origin by convincing a victim to open a crafted Special:Ask link that forces a zero-result JSON export and reflects an unescaped 'default' parameter. Because the injected script runs with the wiki's origin, it can read the victim's session context and CSRF tokens and perform actions as that user; exploitation is deterministic once the link is followed but depends on that user interaction (CVSS v3.1 AV:N/AC:L/PR:N/UI:R/S:C/C:L/I:L/A:N). No confirmed active exploitation exists (CISA KEV not listed, SSVC exploitation: none), though a public proof-of-concept request is documented in the upstream GHSA-574v-j8hf-88h3 advisory; the vendor-released fix is 7.3.1 and the priority signal is 'this_month'.
Stored cross-site scripting in Semantic MediaWiki (2.5.0 through 7.3.0) lets a user with ordinary page-edit rights - and, on wikis that permit anonymous editing, an unauthenticated visitor - store HTML inside a Code-typed property value, which is later rendered unescaped by Special:Browse and executes JavaScript in the browser of anyone who views that page. The assessed vector (CVSS:3.1/AV:N/AC:L/PR:L/UI:R/S:C/C:L/I:L/A:N) puts this at medium severity: the payload never executes server-side and fires only after a victim passively opens Special:Browse for the annotated subject, so impact is bounded to viewer sessions, with escalation to administrator actions described as plausible rather than demonstrated. A reproduction is published in the vendor's GitHub advisory, and the fix ships in Semantic MediaWiki 7.3.1; there is no CISA KEV listing and no confirmed report of active exploitation.
Stored cross-site scripting in the SemanticMediaWiki extension for MediaWiki (5.0.0 through 7.3.0) lets any user with page-edit rights plant a .smw-highlighter span whose data-content, data-title, data-top, data-bottom or data-theme attributes are concatenated unescaped into tippy tooltip content rendered with allowHTML:true, so arbitrary JavaScript runs in the session of any viewer - administrators included - who hovers the element. There is no unauthenticated or zero-interaction path: an editor account is needed to store the payload and the victim must hover the planted element for the onShow handler to write the decoded attribute into innerHTML, which is why the assessed vector is PR:L/UI:R/S:C (the vendor's CVSS 4.0 score of 5.3 does not credit the editor prerequisite). The issue is not in CISA KEV and no active exploitation is confirmed, but the upstream GitHub advisory publishes a proof-of-concept wikitext payload (not verified as working exploit code) and a fixed release, 7.3.1, is available.
Unauthenticated attackers can forge payment notifications and mark any WooCommerce order as paid in stores running the Payex Payment Gateway for WooCommerce plugin through 1.2.10, because the plugin authenticates incoming callbacks against a validation secret that ships empty by default. The out-of-the-box configuration is therefore exposed with no setup required, although an administrator who has set a non-empty secret is not affected, and the impact is confined to order and payment state rather than site compromise or data disclosure. Scoring is moderate (CVSS 5.3, integrity-only) with a low EPSS of 0.13% (3rd percentile) and no public exploit identified at time of analysis; the vuln.today priority is this_month.
Denial of service in Tanium Enforce (2.9 before 2.9.736, 2.10 before 2.10.778, and 3.0 before 3.0.373) lets unauthenticated remote attackers degrade the Enforce service by consuming a resource that is not bounded or throttled (CWE-770). The CVSS 3.1 score of 5.3 reflects network reachability with low attack complexity and no privileges or user interaction, but impact is confined to low availability with no confidentiality or integrity effect, so successful abuse causes partial service degradation rather than system compromise or full outage. No public exploit code was identified at time of analysis and no CISA KEV entry exists; Tanium has released fixed builds and the advisory is TAN-2026-052.
A remote, unauthenticated denial-of-service condition in Tanium Server lets an attacker exhaust server resources by sending crafted requests to the reachable management service, degrading availability only (CVSS 5.3, A:L). Deployments are affected if they run builds below 7.7.3.8351, 7.8.2.1263, or 7.8.4.1399, and patched builds are published for all three branches. No public exploit code or active exploitation has been identified in the supplied sources, so the practical risk is highest for internet-exposed or poorly segmented Tanium Server instances.
A race condition in the A2A Java SDK (a2aproject/a2a-java) at version 1.2.0.Final and earlier lets an authenticated, network-reachable client corrupt the state of a task by issuing overlapping state-changing requests for the same task id - most clearly two simultaneous cancel calls, or a cancel racing a concurrent completion - because the cancel path performs an unserialized check-then-act against the task store. Successful exploitation affects task-state integrity and availability (the task may be finalized on a stale view, and a synthesized failure event may be rejected once the task is already terminal), but exposes no confidentiality per the assessed vector CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:L. Exploitation requires valid authenticated access plus precise timing against a shared task, so single-threaded or non-concurrent usage will not trigger it; no public exploit was identified at time of analysis, and vendor-released fix version 1.3.0.Final is available.
Unbounded registration of push notification configurations in a2a-java up to 1.2.0.Final allows an authorized caller to exhaust server resources. Because no per-task cap existed in the Push Notification Config Store (both the in-memory and JPA-database implementations), each new distinct config ID consumes memory or database rows and can trigger an outbound HTTP callback on every task event, degrading availability. Exploitation requires the privilege to invoke setInfo (CVSS PR:L, not anonymous), impact is availability-only, and no public exploit code was identified at time of analysis; this is a medium-severity resource-consumption issue rather than an emergency.
Authenticated users of MrDoc through 1.1.1 can re-enable disabled document shares and strip access codes from other users' shares, because the manage_doc_share type 3 branch updates DocShare records filtered only on the share token with no ownership check. The token itself is the enabling weakness: it is computed as the MD5 of '{doc_id}_{username}', so anyone who knows or can enumerate the document ID and the owner's username can derive a valid token offline and use it to flip is_enable to true or set share_type to 0, exposing documents that were meant to be private. Severity is medium (CVSS 4.0 base 5.3, assessed 3.1 5.3, CWE-639), exploitation requires a valid account plus a pre-existing share record on the target document, and no public exploit has been identified at time of analysis.
Event guest policy bypass in Admidio 5.0.16 and earlier lets an ordinary logged-in member consume an event's entire remaining seat count by tampering with the hidden additional_guests field, so registration closes for everyone else. A single modified POST to modules/events/events_function.php?mode=participate is enough: the handler never checks the event's dat_additional_guests setting before counting and storing the client-supplied value as mem_count_guests, so guest registration being disabled provides no protection. Impact is confined to integrity of the participant roster and denial of signup for other members (CVSS 5.3, our assessed CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:N) with no confidentiality loss or code execution; there is no CISA KEV listing, the linked GitHub security advisory includes public reproduction steps (a PoC reference) that were not independently verified here as working exploit code, and a vendor patch exists in 5.0.17.
Missing restrictions on repeated authentication attempts in İdvlabs Ontime (all versions through 2026-10-27) let unauthenticated remote attackers send unlimited login guesses against the application's authentication interface, which is the classic CWE-307 brute-force condition. Successful exploitation requires the attacker to actually guess or otherwise obtain valid credentials, so per the assessed vector the worst confirmed outcome is a limited integrity impact (C:N/I:L/A:N) on a compromised account rather than direct code execution or data exposure. No public exploit code has been identified at time of analysis, there is no CISA KEV listing, and the vendor was contacted about the disclosure but did not respond. Rated medium severity (CVSS 5.3) with a this_month priority from TR-CERT reporting.
Stored cross-site scripting in Semantic MediaWiki 3.0.0 through 7.3.0 allows an authenticated user with ordinary page-edit rights to store an event-handler payload in a Keyword-typed property value that Special:Browse renders without HTML escaping, so the script executes in the session of any user who opens that page, administrators included. Exploitation depends on the default Keyword configuration (no display-format URI), the ability to declare and annotate the property, and a victim who actively views the affected Browse page; the advisory documents a proof-of-concept, and there is no CISA KEV listing or other confirmation of active exploitation. The vendor-released patch is 7.3.1, and the assessed CVSS 3.1 score of 5.3 with a this-month priority reflects the edit-rights and user-interaction prerequisites rather than an emergency.
Stored cross-site scripting (CWE-79) in Semantic MediaWiki 2.5.0 through 7.3.0 allows any user holding ordinary page-edit rights - or an anonymous editor on wikis that permit anonymous editing - to store a script payload inside a property annotated as an 'External identifier' value, which Special:Browse later emits into the page without escaping. The payload executes in the browser session of any user who views Special:Browse for the annotated subject, including administrators, giving the attacker access to whatever data and actions that viewer's session permits under the wiki origin. No public exploit is flagged as confirmed working and there is no KEV listing or known in-the-wild exploitation; a public PoC reference exists in the upstream advisory but it is not verified as working exploit code, EPSS is not available, and the vendor has released a patch in 7.3.1. Exploitation is bounded by passive user interaction - a second user must actually open the Browse page for the poisoned subject - which keeps the observed CVSS at 5.3 and the priority decision at this_month rather than act_now.
Reflected cross-site scripting in Semantic MediaWiki (3.1.0 through 7.3.0) lets an unauthenticated attacker inject markup through the unescaped id parameter of the Special:SMWAdmin ID lookup, which is rendered verbatim into an HTML warning box. Because SpecialAdmin::execute overwrites wpEditToken on the request with the current user's own token before the handler validates it, a bare GET link with no CSRF token passes the only request-time guard, so the attacker only needs to lure a logged-in user holding the smw-admin right into opening a crafted URL that supplies a non-numeric id matching no stored entity. Successful exploitation yields script execution same-origin with the wiki in the privileged administrator's session, enabling authenticated requests and privileged wiki actions; a proof-of-concept payload is described publicly in the vendor advisory, though it is not verified as packaged working exploit code, active exploitation is not confirmed (no CISA KEV), and the vendor-released fix is version 7.3.1.
Stored cross-site scripting in Semantic MediaWiki 3.0.0 before 7.3.1 lets an attacker who can introduce a malicious semantic property value inject markup into the property-value autocomplete dropdown rendered for another user. The vendor-assigned CVSS 4.0 score is 5.3 (medium); exploitation requires the autocomplete widget to be enabled and a victim to open it with an empty query, and the confirmed impact is markup injection rather than demonstrated JavaScript execution. A safe public proof-of-concept reference shows the markup-injection behavior, but it is not verified as working exploit code, and no CISA KEV active exploitation is reported.
Authenticated users with low privileges can crash the web management service on Moxa NPort W2150A-W4 and W2250A-W4 wireless device servers running firmware 1.0 through 1.5 by supplying oversized input to a management-interface field that is copied into a fixed-size stack buffer without adequate bounds checking. The result is a management-plane denial of service: the web service becomes unresponsive and needs a power cycle, while other device functions remain operational, and the supplied assessment records no confidentiality or integrity impact. No public exploit was identified at time of analysis, and Moxa rates the issue CVSS 4.0 base 5.3 (medium).
Reflected cross-site scripting in Semantic MediaWiki's Special:FacetedSearch page (versions 4.2.0 through 7.2.0) is scored MEDIUM (CVSS 4.0 base 5.1) and lets unauthenticated remote attackers execute script in a victim's browser when the victim opens a crafted URL containing malicious cstate parameters. The flaw is a residual of CVE-2025-10354: cstate values are concatenated into hidden-input markup without escaping and inserted unescaped into the page template, so a payload can break out of the value attribute. A public PoC reference exists; it is not verified as working exploit code, and no KEV entry is present in the supplied intelligence; the vendor released a fix in 7.2.1.
Stored cross-site scripting in KGS Software tia Viewer prior to version 1.2.8 lets an authenticated user with access to the annotation API inject arbitrary JavaScript into a stickynote annotation; the payload executes in the browser of any user who later opens the document, putting that user's session at risk. Exploitation requires an attacker account able to create or edit annotations plus a victim who opens the document, so impact is confined to the victim's browser session. Severity is moderate (CVSS 5.1; assessed CVSS:3.1 7.1-style scope-changed vector AV:N/AC:L/PR:L/UI:R/S:C/C:L/I:L/A:N), no public exploit code or active exploitation has been confirmed, and version 1.2.8 contains the fix.
Octopus Server deployments in the 2026.1, 2026.2 and 2026.3 release branches (plus 2023.0.0 through 2026.1.674) allow an authenticated user who can upload a NuGet package to configure and deploy a script step that executes outside the intended parent directory. Because the actor must already hold elevated project and deployment privileges (PR:H), the realistic threat is a malicious insider or a compromised privileged account rather than an anonymous remote attacker, and the assessed impact is limited to low integrity and availability effects with no confidentiality loss. No public exploit code has been identified at time of analysis, no confirmed active exploitation (CISA KEV) is recorded, and the priority decision is 'this_month' - a moderate-severity, low-urgency issue rather than a top priority.
CSS injection and remote-reference filtering bypasses in enshrined/svg-sanitize before 1.0.0 let an attacker who can upload SVG files embed external stylesheets, tracking pixels and unquoted url() references that survive sanitization. The library never inspects the text content of <style> elements, its removeRemoteReferences option only matches quoted url('...') patterns in attributes, and bare href attributes on <image> are not checked at all, so a low-privileged uploader can exfiltrate host-page attribute data via CSS selectors or disclose visitor IP addresses and browsing activity. Impact is limited to confidentiality and tracking rather than code execution, and the CSS-selector data-exfiltration path only applies where the sanitized SVG is rendered inline in the DOM, not when loaded through <img src>. Proof-of-concept payloads are documented in the upstream advisory; no confirmed exploitation in the wild is reported, and the vendor fix is version 1.0.0.
REDAXO core releases before 5.21.3 render the backend media pool detail page's opener_link request parameter into the panel title as raw HTML, producing a reflected cross-site scripting condition (CWE-79/CWE-116) that an unauthenticated attacker can trigger by getting a logged-in backend user with media permissions to open a crafted link. The injected script executes in the backend origin with the victim's session; because the session cookie is HttpOnly it cannot be exfiltrated, but the CSRF token is present in the DOM, so the payload can perform authenticated actions limited to that user's role - escalating to code execution via template or module edits when the victim is an administrator. The advisory includes an illustrative payload URL; no separate public exploit release or confirmed in-the-wild exploitation was identified, and the priority signal places remediation in the this_month window rather than act-now.
REDAXO through 5.21.2 permits cross-site request forgery in its structure/history plugin, letting an attacker who lures a logged-in backend user onto a hostile page permanently wipe every stored article snapshot (func=clearall) or silently roll an article back to an earlier stored version (rex_history_function=snap). Both actions were triggered by plain GET requests without a CSRF token, and the default SameSite=Lax session cookie is still sent on a top-level navigation, so neither a form submission nor target-side JavaScript was required; the destructive clearall path needs an administrator victim, while rollback only needs a victim holding the history[article_rollback] permission. This is a genuine but modest-priority issue: user interaction and victim privilege level constrain it, the attacker gains no content disclosure, no permission change and cannot inject arbitrary content, and no public exploit code has been identified at time of analysis. The vendor released a fix in 5.21.3 that enforces token validation on both actions.
Stored cross-site scripting in REDAXO's media, medialist, link and linklist backend input widgets (versions before 5.21.4) lets an authenticated content editor persist a crafted REX_INPUT_MEDIA, REX_INPUT_MEDIALIST, REX_INPUT_LINK or REX_INPUT_LINKLIST value that breaks out of the form field's value attribute and runs script in the backend session of any administrator who later opens a form containing that widget, enabling session and CSRF token theft and account takeover. The widgets build their HTML by string concatenation without rex_escape(), so the value is stored unfiltered and the readonly attribute on the field is only a client-side hint. Rated 5.1 (Medium); no public exploit identified at time of analysis and a vendor-released patch (5.21.4) is available.
Cross-site request forgery in REDAXO CMS core 5.21.4 and earlier allows an attacker to flip the global safe-mode setting by convincing a logged-in backend administrator to open a crafted link. Because index.php processes the safemode GET parameter without a CSRF token, the request writes safe_mode: false into config.yml when global safe mode was enabled, or activates session safe mode that disables all backend addons except backup, be_style and install for that session. Confidentiality and privileges are unaffected, exploitation requires an authenticated administrator victim plus user interaction and a non-live installation, and no public exploit code has been identified at time of analysis; the issue is rated medium severity and prioritised as this_month rather than urgent.
Stack-based buffer overflow in Dell PowerProtect Data Domain (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) allows an authenticated attacker who already holds administrative-level privileges to overflow a stack buffer over the network and crash the appliance, producing a denial of service with no confidentiality or integrity impact. Dell has released fixes in 7.13.1.90, 8.9.0.0, 8.3.1.50 and 8.6.1.30, and the CVSS 4.9 score is driven mainly by the PR:H prerequisite rather than by unauthenticated reachability. No public exploit identified at time of analysis.
A stack-based buffer overflow (CWE-121) in Dell PowerProtect Data Domain allows a high-privileged, network-reachable attacker to crash the affected service or appliance, resulting in denial of service; there is no confidentiality or integrity impact. 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. The PR:H prerequisite (CVSS 4.9, AV:N/AC:L/PR:H/UI:N/S:U/C:N/I:N/A:H) is the main limiting factor - this is not reachable by an unauthenticated or low-privileged remote user - and no public exploit has been identified at time of analysis.
A stack-based buffer overflow in Dell PowerProtect Data Domain appliances allows an attacker who already holds administrative-level privileges on the appliance, reachable over the network, to corrupt memory and crash the affected process, resulting in denial of service. Only availability is impacted (C:N/I:N/A:H) and the high-privilege prerequisite is the main limiting factor - this is not exploitable by an unauthenticated or low-privileged user, and no public exploit has been identified at time of analysis. Affected releases are LTS2024 7.13.1.0 through 7.13.1.80, LTS2025 8.3.1.0 through 8.3.1.40, LTS2026 8.6.1.0 through 8.6.1.20 and Feature Release 7.7.1.0 onward; Dell has published fixed builds and the issue is triaged for remediation within the current monthly cycle.
Arbitrary file read in Supermicro BMC/IPMI firmware 01.07.20 allows an authenticated IPMI Administrator to retrieve files from the BMC filesystem, disclosing information beyond the intended IPMI management role. The CVSS 3.1 score is 4.9 with a vector of AV:N/AC:L/PR:H/UI:N/S:U/C:H/I:N/A:N, so exploitation requires high-privilege IPMI credentials and affects confidentiality only. EPSS is low at 0.23% (14th percentile), there is no CISA KEV confirmation, and no public exploit was identified at time of analysis; real-world exposure is further limited where BMC/IPMI interfaces are isolated on management networks.
Unauthenticated attackers can permanently delete arbitrary WordPress media library attachments through the Advanced Custom Fields: Extended plugin (hwk-fr) in all versions up to and including 0.9.2.6, because the Email form action fails to verify that the submitted attachment ID belongs to the submitter or that the user holds any capability. The attack surface is narrow: exploitation only works on sites where an administrator has published a public-facing ACFE form whose Email action attachment row uses a dynamic {field:...} source bound to a submittable field with the Delete option enabled, and the attacker must supply a valid numeric attachment ID. No public exploit has been identified at time of analysis, and the assessed impact is limited to integrity and availability of media attachments (no disclosure, no code execution), which is why the assessed vector carries AC:H and the priority signal places remediation within this month rather than immediately.
Stored cross-site scripting in Semantic MediaWiki 3.2.0 through 7.3.0 lets a user holding the elevated smw-schemaedit right save a property group schema whose canonical_name contains script markup, which the Special:Browse group formatter emits unescaped as a section heading and executes in the browser of anyone who later browses a page carrying a member property of that group, including administrators. Exploitation is authenticated and requires the victim to open the affected Special:Browse page, so impact scales with the privilege of the viewer rather than being reachable by anonymous visitors. A detailed public proof-of-concept reproduction is published in the GitHub advisory; no confirmed active exploitation or KEV listing exists, and this is assessed as a genuine but moderate-priority issue. The vendor has released SemanticMediaWiki 7.3.1 with the fix.
Vorota-ai's nmap-mcp MCP server does not enforce its own NSE script-category blocklist in the scan_vulnerabilities tool, so callers can pass categories such as exploit, dos, brute, or intrusive straight through to nmap --script. Anyone able to invoke that tool on the deployment, whether a legitimate MCP client or an agent steered by prompt injection, can therefore turn an ordinary scan into intrusive probing, brute-force attempts, or denial-of-service script runs against whatever targets the server is pointed at. The flaw is rated 4.8 (CVSS 4.0, AV:L/PR:L, with scope change and low integrity/availability impact on downstream systems), no public exploit code and no CISA KEV entry were identified at time of analysis, and the vulnerable host itself is not compromised; harm depends on the server being aimed at systems an attacker wants to disrupt.
Stored cross-site scripting in Semantic MediaWiki versions from 3.1.0 before 7.3.1 lets an authenticated user with the smw-schemaedit permission or an equivalent trusted schema-editing role save HTML in a schema description that is rendered unescaped when another user views the schema page. The confirmed issue is raw HTML insertion that can alter pages and target viewers; escalation to active script execution depends on the deployment permitting an executable HTML context, and a victim must interact by viewing the affected schema. A safe proof-of-concept marker demonstrates the insertion but is not verified as working exploit code, and no active exploitation is reported; the CVSS 4.0 base score is 4.8 (Medium).
Stored HTML injection leading to cross-site scripting in Semantic MediaWiki 4.2.0 through 7.3.0 lets an authenticated profile or schema editor plant markup in a FacetedSearch exploration-link label, which ExploreListBuilder::buildHTML() passes unescaped into Html::rawElement() so the content renders as raw anchor markup on Special:FacetedSearch for anyone who opens that exploration list. The prerequisites are narrow: the attacker needs a privileged editor role (PR:H in both the CVE 4.0 and the assessor's CVSS 3.1 vector), the victim must open the affected exploration list (UI:P / UI:R), and per the vendor the injected markup only becomes script-capable where a specific deployment permits an executable HTML context - otherwise the realistic impact is editor-controlled UI alteration or defacement. Vendor patch 7.3.1 is available and there is no confirmed active exploitation (SSVC exploitation: none, not in CISA KEV); the advisory ships only an inert proof-of-concept marker, so a public PoC reference exists but it is not verified as working exploit code, and the issue is prioritized as this_month.
Server-side request forgery in the Nextcloud Whiteboard app lets an authenticated low-privileged user force the recording backend to reach an internal Chrome DevTools Protocol API and leak another user's JWT authentication token, granting read/write access to that victim's whiteboard. The attacker must hold a valid account on the same Nextcloud instance and repeatedly poll the SSRF while a victim starts a recording, so the victim's own action is a prerequisite (UI:R) and successful capture is timing-dependent. Scored 4.6 (CVSS:3.1/AV:N/AC:L/PR:L/UI:R/S:U/C:L/I:L/A:N) and rated this_month by vuln.today; no public exploit code was identified at time of analysis and the issue is not in CISA KEV, so the realistic risk is targeted, conditional abuse rather than mass scanning.
Nextcloud Tables 0.6.4 through 1.0.0 contain a moderate-severity authorization (access-control) flaw, not a high-priority emergency, that lets an authenticated member of a group with an administrator-configured link-sharing restriction create share links for tables they can already access, bypassing that group restriction. The issue is only exploitable where a group-level link-sharing restriction has actually been set, and impact is limited to data in reachable tables, capping confidentiality and integrity impact at Low. No public exploit identified at time of analysis; the only supplied reference is a HackerOne report.
Psono Server CE and EE versions 4.0.33 through 7.2.3 authorize the bulk-secret-read endpoint against a hard-coded unprefixed request path, so in deployments configured with URL_PREFIX the check fails and the request is granted under write scope instead of read scope. The result is that an authenticated remote attacker holding an atypical API key with read disabled but write enabled can pull encrypted secret data and nonces through the prefixed endpoint despite lacking read permission; only non-default, URL_PREFIX-configured installations are exposed, and the data obtained remains encrypted. The assessed severity is medium (CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:U/C:L/I:N/A:N), materially lower than a naive reading of a read-bypass would suggest, EPSS is 0.32% (25th percentile), no public exploit code has been identified at time of analysis, and the vendor has released a fix in v7.2.4.
Information disclosure in Nextcloud Server 32.x through 33.x lets any authenticated user who is not a member of a team configured as 'visible to everyone' enumerate that team's internal resource names and identifiers by calling the resources API endpoint directly, even though the web UI correctly returns a 'You are not a member of ...' block. The disclosure is limited to the existence, names and identifiers of resources - their contents remain protected - and there is no integrity or availability impact, which is consistent with the assessed CVSS 3.1 vector AV:N/AC:L/PR:L/UI:N/C:L/I:N/A:N (4.3). This is a low-to-moderate priority authorization-consistency bug rather than an urgent threat: no CISA KEV listing or public exploit code is confirmed in the available sources, and per the vuln.today decision label it is triaged for handling this month.
Authenticated denial of service in russh 0.60.0 through 0.64.1: a client can open a session channel advertising a maximum packet size of zero, and any server write through Channel::data(), Channel::make_writer(), or Channel::into_stream() then remains pending forever instead of returning an error. A malicious server can cause the same silent stall in a client by confirming a channel with a zero maximum packet size. The vendor fixed the defect in 0.65.0 by rejecting zero at parse time; the advisory includes a verified end-to-end reproducer, no CISA KEV entry is reported, and this is assessed as a moderate-priority availability problem rather than a critical emergency.
Out-of-bounds read in ThorVG's SFNT/TTF font loader (all versions before 1.0.7) lets an attacker who can supply a font buffer smaller than four bytes through the in-memory loading API, such as tvg::Text::load(..., data, size, "ttf", true), trigger a heap over-read in SfntLoader::gen. The result is limited to adjacent heap memory disclosure or an application crash, with no code execution path; the same malformed font delivered by file path is largely masked because mmap page alignment pads the read. This is not in CISA KEV and there is no confirmed active exploitation, but the GitHub advisory publishes a reproduction payload (a 1-byte buffer) that trips an ASan heap-buffer-overflow, and the issue is fixed in 1.0.7; overall it is a genuine but modest-priority parsing defect that matters mainly to applications that accept untrusted font bytes from memory.
Nextcloud Files Confidential (versions 3.3.0 through 5.0.0) fails to apply its content-based confidential classification when an uploaded docx or odt file has a '#' character in the filename, allowing an authenticated user to store sensitive content such as a credit card number without it being marked confidential. The flaw is a low-severity access-control weakness (CVSS 4.3, confidentiality-only impact, AV:N/AC:L/PR:L/UI:N) that matters only where the app is deployed and relied upon as a marking control; exploitation requires a valid account and upload capability. No public exploit code or confirmed active exploitation was identified at time of analysis, and the vuln.today priority is this_month.
An out-of-bounds read in Qt's qt5compat QTextCodec TSCII encoder can crash an application or cause it to return misinterpreted text when it converts attacker-influenced input to the legacy TSCII encoding. The issue is rated CVSS 4.0 4.3 (medium) and is constrained by a local attack vector (AV:L) plus required user interaction (UI:P); exploitation needs an application that links the Qt 5 Compatibility APIs and actually performs TSCII conversion, a rare code path. EPSS is 0.12% (2nd percentile), no public exploit was identified at time of analysis, and a vendor patch is available.
Local, low-privileged users on a Windows host running Veeam Backup for Microsoft 365 can steer the installer into writing arbitrary files with SYSTEM privileges, creating a privilege-escalation or denial-of-service risk. Affected builds are 8.0 through 8.6.0.1102, and exploitation is constrained to the installation or upgrade window with administrator-initiated interaction; the assessed severity is low-to-moderate (CVSS 4.3 Medium), and no public exploit was identified at time of analysis.
Memory exhaustion in the russh Rust SSH library affects every release before 0.65.0: a malicious SSH server can send repeated pre-authentication EXT_INFO packets encoded with russh's private message number 7, and the client's handle_ext_info appends each parsed extension name to a per-session vector that is never cleared, deduplicated, or bounded, growing client heap for the lifetime of the connection. The victim must initiate a session with a hostile or compromised server (UI:R, PR:N), and the advisory reports that retention persists even after the client has verified the server's host key; impact is availability only (C:N/I:N/A:L, CVSS 4.3). The vendor shipped a fix in 0.65.0 and the advisory includes a reproduction harness exercising the real packet parser; there is no CISA KEV entry and no confirmed active exploitation, and no public exploit code is verified beyond that harness.
Information disclosure in Tanium Server lets an authenticated low-privilege user expose limited sensitive data over the network, with no integrity or availability impact. The issue affects deployments running versions below 7.7.3.8351, 7.8.4.1399, and 7.8.2.1263, and Tanium has released fixed builds for each branch. Rated 4.3 with an AC:L/PR:L vector, it is a moderate-severity, lower-priority item - no public exploit identified at time of analysis and no CISA KEV listing.
An uncontrolled resource consumption flaw (CWE-770) in Tanium Enforce lets an authenticated user drive the Enforce component into resource exhaustion, degrading only the availability of that component. Deployments running Enforce 3.0 before 3.0.373, 2.9 before 2.9.736, or 2.10 before 2.10.778 are affected; patched builds are not exploitable. The CVSS score is 4.3 (AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:L), reflecting a low-impact, availability-only issue that requires a valid Enforce session - no public exploit code has been identified at time of analysis, and the issue is not confirmed actively exploited (CISA KEV).
Tanium Comply builds 2.32 before 2.32.263, 2.35 before 2.35.323, and 2.37 before 2.37.326 enforce authorization incorrectly, allowing a low-privilege authenticated user with network access to read data outside their permitted scope. The flaw is rated CVSS 4.3 with confidentiality-only, low impact, and requires valid Comply credentials, so it is not exploitable without authentication; no public exploit was identified at time of analysis, and Tanium has released fixed builds.