Authenticated Dolibarr administrators can run arbitrary operating-system commands as the web server user through the database export tool (htdocs/admin/tools/export.php) in every release before 24.0.0, because the restrict_os_commands allowlist can be bypassed by injecting a crafted mysqldump value. The attacker must already hold administrator rights (CVSS 4.0 base 8.6; assessed CVSS 3.1 AV:N/AC:L/PR:H/UI:N/S:U/C:H/I:H/A:H) and exploitation is confined to the web server user's privileges, so there is no scope change to other systems; there is no CISA KEV entry and no public exploit identified at time of analysis, and the vendor rates the fix as this-week rather than emergency priority.
Stack-based buffer overflow in FLTK's BMP/ICO image decoder (Fl_BMP_Image::load_bmp_()) affects FLTK through 1.4.5: a crafted BMP or ICO file whose colors_used field exceeds 256 overflows a fixed 768-byte stack colormap array, corrupting stack memory and causing crashes with potential code execution. Exploitation requires a victim to open or process a malicious image through an application linked against FLTK's image loading path - the assessed vector is local with user interaction and no authentication, and no remote or zero-click exposure is described. A vendor fix commit is available and an independent public write-up with PoC-level detail is referenced; EPSS is low (0.16%, 5th percentile) and there is no KEV listing, so signals point to a moderate, patch-this-week priority.
Heap buffer overflow in FLTK through 1.4.5 lets a crafted BMP or ICO file corrupt the application heap when the toolkit's image loader (Fl_BMP_Image::load_bmp_()) decodes a file whose declared width and height overflow the pixel-buffer size calculation. Exploitation requires the target application to use FLTK's BMP/ICO loading path and a victim to open or render the malicious file - a local attack vector with user interaction (CVSS 8.5; assessed 3.1 vector AV:L/AC:L/PR:N/UI:R, C/I/A:H) - so there is no remote, interaction-free path, no authentication is needed, and no prior privileges are required. EPSS is low at 0.15% (4th percentile), the issue is not listed as actively exploited in CISA KEV, and although a public technical write-up of the overflow exists it is not verified as working exploit code; the vendor fix is available via upstream commit.
Improper file permissions in Japan's Ministry of Land, Infrastructure, Transport and Tourism Electronic delivery item Inspection Support System (versions 0 through Ver.4.0.36) let a low-privileged local user overwrite files inside the application's installation folder, which are then executed in a more privileged context, yielding arbitrary code execution. The flaw is purely local (CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H, vendor score 8.5), so an attacker must already hold a foothold on the host and a higher-privileged user or service must subsequently invoke the tampered files; it is an integrity and privilege-escalation problem rather than a remotely reachable emergency. No public exploit code was identified at the time of analysis, and the issue is not in CISA KEV; JPCERT reporting places it in the 'this_week' remediation queue, and no fixed version is stated in the available data - consult JVN#75668103 for update status.
Use of a hard-coded cryptographic key in Johnson Controls Kantech EntraPass before 9.20 enables a padding oracle crypto attack. Exploitation requires local access and low-level privileges (AV:L/PR:L per the vendor CVSS 4.0 vector), so the attacker must already have authenticated or local foothold on the EntraPass system rather than operating as a remote unauthenticated adversary; impact is high across confidentiality, integrity and availability of the affected system. No CISA KEV entry, public exploit code, or EPSS score was provided, and the product's fixed 9.20 release resolves the condition.
SQL injection in Crocoblock JetEngine through version 3.9.0 allows an authenticated low-privilege user on a WordPress site to inject SQL and read sensitive database data, with a low availability impact and no user interaction required. The CVSS 3.1 score is 8.5 (AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:N/A:L), and the issue is prioritized for this week rather than as an emergency because exploitation is gated by authentication. No public exploit code has been identified at time of analysis.
Blind SQL injection in the WordPress plugin Media Library Assistant through version 3.42 allows an authenticated low-privileged user to read sensitive data from the site database; the assessed impact is confidentiality-only. The issue is network-reachable with no user interaction required, but exploitation is gated by a valid account (PR:L) and the specific injectable endpoint or parameter is not named in the available sources. No public exploit code has been identified at time of analysis, and the issue carries a this_week priority rather than an emergency designation.
Read-level project members in OpenPanel before 3.0.0 can retrieve plaintext integration secrets - Slack bot tokens, incoming-webhook URLs and webhook authorization headers - through the notification.rules tRPC query, which eager-loads integration rows and spreads them into the response without applying the redactIntegration step that integration.get and integration.list enforce. The flaw is access-gated: the attacker needs an existing authenticated account with bare project membership (the lowest role, which cannot create or edit integrations) on a project where a write/admin user has attached at least one integration to a notification rule, and a single query call with no user interaction then exposes those credentials for use against external systems. EPSS is 0.30% (23rd percentile) with no CISA KEV listing, and the vendor advisory contains a benign manual proof-of-concept walkthrough; it is a public PoC reference that is not verified as working exploit code.
A local privilege escalation in Apple macOS allows an attacker who can already run code on the host to gain elevated privileges, affecting Macs running macOS versions prior to the fixed releases 26.7.2 (Tahoe), 27.2 (Golden Gate) and 15.8.2 (Sequoia). Apple describes the root cause as a logic issue in a protection mechanism, corrected with improved checks. Exploitation has not been observed and no working exploit code is publicly confirmed; EPSS is low at roughly 0.24% (15th percentile) and SSVC marks exploitation as none and non-automatable, though technical impact is total once local access is achieved. Priority is this_week rather than emergency-grade.
Opening a crafted V7 project file in Fuji Electric V-SFT (the HMI screen-configuration and programming software, listed in the CPE as Fuji Electric / Hakko Electronics V-SFT) overflows a stack buffer and can result in information exposure or arbitrary code execution with the privileges of the user running the application. The attack is local rather than network-reachable and requires the victim to actively open the malicious file, so compromise depends on social engineering a legitimate V-SFT user; no authentication is required of the attacker. No public exploit code has been identified at time of analysis, and the vulnerability is tracked by JPCERT/CC under JVNVU96549652 with a priority of 'this_week' rather than emergency-level.
Out-of-bounds read in Fuji Electric V-SFT can be triggered when a user opens a crafted V7 project file, potentially exposing information or enabling arbitrary code execution on the affected system. The flaw is local and requires user interaction-the victim must open the malicious file-and no privileges are required, so social engineering is the primary attack vector. This is a moderate-priority, not emergency, issue; no public exploit identified at time of analysis.
Opening a crafted V7 project file in Fuji Electric/Hakko V-SFT can cause an out-of-bounds read in VS6EditData!CDataRomErrorCheck::CheckSymbolMacroParam, potentially leading to information exposure or arbitrary code execution in the context of the user running the application. The issue is assessed as high severity (CVSS 4.0 8.4) with a local, user-assisted file-open path, and affected-product data uses wildcard CPE version fields rather than a specific version range. No CISA KEV designation is present in the supplied data, and no public exploit was identified at time of analysis; the independent assessment rates it this-week priority, not emergency-grade.
Opening a crafted V7 project file in Fuji Electric V-SFT can trigger an out-of-bounds read in the macro-command check path, leading to information disclosure or arbitrary code execution on the workstation running the product. The flaw is local to the host, requires no attacker privileges, and depends on the victim opening the malicious file, so exploitation is constrained by file delivery and user interaction rather than remote network access. No public exploit was identified at time of analysis, and the vendor and JVN advisories are the available sources for affected-version and update details.
Opening a crafted V7 project file in Fuji Electric's V-SFT screen-creation/engineering software can drive an out-of-bounds read in the VS6MemInIF!set_temp_type_default routine, exposing process memory and, per the vendor, potentially enabling arbitrary code execution on the workstation running the application. Rated 8.4 (CVSS 4.0, CWE-125) and tracked by JPCERT/CC, the issue is conditional rather than mass-exploitable: the attack vector is local and the victim must actively open an attacker-supplied project file, while no authentication or special privileges are required for the payload to be processed. No public exploit identified at time of analysis, no CISA KEV listing, and the vendor advisory plus JVN note are the primary sources for version and fix guidance.
Out-of-bounds read in Fuji Electric V-SFT's VS6ComFile!load_gr_item routine can expose memory contents or enable arbitrary code execution when an operator opens a crafted V7 file. The attack is local and requires user interaction: an attacker must deliver a malicious V7 file and convince a V-SFT user to open it. No public exploit identified at time of analysis, and the issue is not confirmed as actively exploited (CISA KEV).
An out-of-bounds write in Fuji Electric's V-SFT screen-creation software (also published under Hakko Electronics) is triggered when a user opens a crafted V7 project file, allowing memory corruption that can disclose information or lead to arbitrary code execution on the engineering workstation running the tool. Exploitation is local and interaction-gated: the attacker must first deliver the malicious V7 file to a victim who then opens it in V-SFT, so there is no network-facing attack surface and no credentials are needed for the file itself to carry the payload. No public exploit code and no confirmed active exploitation (CISA KEV) are indicated in the available intelligence, but the JVN-published issue is rated high severity and prioritized for remediation this week.
An out-of-bounds write in libspeex's speex_bits_read_whole_bytes() (libspeex/bits.c) affects Speex through 1.2.1 and triggers only when an application decodes into a caller-provided, non-owned SpeexBits buffer created with speex_bits_init_buffer() or speex_bits_set_bit_buffer(); crafted Speex data then overflows that buffer because length truncation is mishandled, corrupting heap or stack memory and crashing the process. Remote, unauthenticated attackers can reach it wherever such streams are processed, but exposure is conditional on the vulnerable buffer APIs being in use - reflected as AT:P in the CVSS 4.0 vector and as moderate real-world priority in the provided assessment, with EPSS at 0.41% (35th percentile), no CISA KEV listing, and a public PoC reference (a HackMD writeup) that is not verified as working exploit code. The overall rating is denial-of-service-weighted (A:H, I:L, no confidentiality impact), and a fix has been proposed upstream.
Replay of previously signed ActivityPub requests against YesWiki's public /api/forms/{formId}/actor/inbox endpoint lets an attacker re-execute captured Delete, Update, Follow or Undo actions at any later time, because HttpSignatureService::verifySignature() never validates the Date header and keeps no record of seen signatures. YesWiki from 4.6.2 through 4.6.7 is affected, and the 5.0.0-alpha1 prerelease also lacks the guard; the vendor fix is 4.7.0. Exploitation is gated by the need to first obtain a valid signed request (reverse-proxy logs, WAF/IDS logs, application debug output or shared-hosting co-tenancy), which is consistent with the low EPSS score of 0.22% (13th percentile) and the absence of a CISA KEV listing; a public PoC reference exists but it is not verified as working exploit code, and the attacker cannot forge new actions - only replay captured ones.
Memory corruption in SOEM (Simple Open EtherCAT Master) 2.0.0 and earlier lets a malicious or compromised device on the EtherCAT segment drive manipulated message length values through unvalidated arithmetic, producing negative or oversized results that bypass bounds checks and cause writes outside allocated memory. The flaw is rated CVSS 8.3 (AV:A/AC:L/PR:N/UI:N, C:L/I:H/A:H) and affects masters built on the library during normal message processing - no authentication or user interaction is needed, but the attacker must already have a presence on the same fieldbus segment as the master, so internet-facing remote exploitation is not indicated. No public exploit code or CISA KEV entry was identified at time of analysis, and the vuln.today decision label is 'this_week', reflecting a genuine priority for EtherCAT automation operators rather than a drop-everything emergency.
Server-side request forgery in the Bazar CSV import action of YesWiki (all versions before 4.7.0, Composer package yeswiki/yeswiki, confirmed through 4.6.7) lets an authenticated administrator force the web server to issue arbitrary outbound requests by placing crafted URLs in CSV file or image field values, with the fetched response written into the web-accessible files/ directory and retrievable by any unauthenticated visitor at a predictable path. Because the raw URL reaches curl_init() without the SsrfUrlValidator::curlPin() guard used by every other outbound-request sink in the codebase, an attacker can reach cloud metadata endpoints (169.254.169.254 on AWS/GCP/Azure, exposing IAM credentials), internal network services, or local files via file://, gopher:// and dict://; the filename-extension allowlist is defeated by a '?.ext' parser differential between PHP basename() and curl or by a redirect chain under CURLOPT_FOLLOWLOCATION. The upstream GHSA advisory publishes a step-by-step proof-of-concept walkthrough, but no confirmed in-the-wild exploitation is reported and EPSS is low at 0.34% (28th percentile); the principal limiting factor is the administrator-level privilege requirement, which keeps this off the mass-exploitation path.
Denial of service in the Datadog dd-trace-py Python APM client, versions 3.13.0 through 4.15.4, where the IAST instrumentation for old-style % string formatting fails to contain an internal error and can terminate the entire Python process, interrupting whatever work it was handling. Three conditions must all hold for exploitation: IAST must be explicitly enabled (DD_IAST_ENABLED=true, which is not the default), the monitored application must apply a %-formatting operation to attacker-controlled input, and that input must contain a value the taint-tracking modulo aspect cannot process - the vendor release note specifically identifies a lone UTF-16 surrogate character that fails native UTF-8 conversion. This is a pure availability issue with no confidentiality or integrity impact; the advisory withholds reproduction steps, and no public exploit or active exploitation has been identified at time of analysis (EPSS 0.33%, not in CISA KEV).
Server-side request forgery in YesWiki's page duplication API (DuplicationManager::downloadFile) lets an authenticated administrator drive the server into fetching arbitrary URLs - including file:// paths and internal or cloud-metadata endpoints - and writes the fetched bytes into the web-accessible files/ directory with no filename extension restriction. All releases up to and including 4.6.7 are confirmed affected and the vendor has released 4.7.0 as the fix. Exploitation is gated by the admin privilege requirement, which is enforced by both a route-level @admins ACL and an in-method denyAccessUnlessAdmin() check; there is no CISA KEV listing and SSVC records exploitation as none, EPSS is low at 0.24% (15th percentile), while a public proof-of-concept request sequence is published in the vendor advisory (a PoC reference exists; it is not verified as working exploit code).
An authentication issue was addressed with improved state management. This issue is fixed in macOS Golden Gate 27.2. A remote attacker may be able to cause arbitrary code execution.
Missing authentication in OmniRoute 3.8.50 through 3.8.51 lets an unauthenticated remote caller POST to /api/sync/tokens and mint a sync token whenever the instance runs in open mode (requireLogin=false), then spend that token against GET /api/sync/bundle to pull decrypted provider credentials and plaintext API keys - the same material /api/settings/export-json already denies to anonymous callers. The flaw is conditional: when login is enabled (the default), token issuance already required a dashboard session or the local CLI, so only open-mode deployments are reachable, which is why the assessed CVSS:3.1 vector is AV:N/AC:H/PR:N/UI:N with confidentiality-only impact (C:H/I:N/A:N) even though the vendor's CVSS:4.0 score is 8.2. Tokens minted this way were recorded as issuedBy:"open" and kept opening the bundle after login was re-enabled; version 3.8.52 fixes issuance, routing /api/sync/tokens into the always-protected tier and refusing the legacy "open" tokens. No public exploit code was identified at time of analysis and the CVE is not in CISA KEV; priority is this_week.
Routinator, NLnet Labs' RPKI relying party validator, mishandles RRDP notification files by accepting delta entries with gaps and by not safely handling an excessive number of delta entries. A malicious or compromised RPKI publication point, or a network-path attacker able to serve crafted RRDP content, can cause incomplete RPKI data (integrity impact) or memory exhaustion and a crash (availability impact). No public exploit code or confirmed active exploitation is identified in the supplied data.
netty-tcnative 2.0.83.Final and earlier can be driven into denial of service or silent TLS message corruption by ordinary TLS 1.3 traffic, because the tcn_write_to_bytebuffer BIO_write callback in ssl.c reports more bytes consumed than the caller actually asked to write. The inflation occurs when the internal side buffer (populated by queued post-handshake records such as NewSessionTicket, KeyUpdate or alerts) is non-empty at the moment a write-sink BIO_write runs, and the outcome is backend-specific: BoringSSL silently corrupts handshake or application data on the wire, an adjacent BoringSSL post-handshake path can abort the JVM process outright, and LibreSSL permanently wedges the connection's write path. The independent assessment rates this CVSS:3.1 AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:L/A:H - unauthenticated remote and requiring the coincidence of queued side-buffer data with a write-mode flush - while a public proof-of-concept reproduction is documented in the upstream advisory; it is not verified as turnkey exploit code and there is no CISA KEV confirmation of active exploitation at time of analysis.
A use-after-free in netty-tcnative 2.0.66.Final through 2.0.83.Final builds linked against OpenSSL 3.0 or newer lets a TLS handshake crash the server process when it issues or resumes a session ticket at the same moment the operator rotates ticket keys through SSLContext.setSessionTicketKeys0(). The handshake side of the race is reachable by unauthenticated remote clients (PR:N, UI:N), but the limiting precondition is the operator-initiated rotation happening inside a window only a few CPU instructions wide, which is why the assessed vector carries AT:P and AC:H and the reported CVSS 4.0 score of 8.2 resolves to an assessed CVSS 3.1 base of 5.9 with confidentiality and integrity impact unconfirmed and only availability (A:H) demonstrated. No CISA KEV entry exists and no weaponized exploit code has been identified, though the public advisory documents a reporter-side, repeatable SIGSEGV reproduction on a real compiled openssl-dynamic build using a race-widened test harness; patched builds are available and the priority label is this_week.
Heap buffer overflow in netty-tcnative's custom OpenSSL BIO wrapper (all releases through 2.0.83.Final, including netty-tcnative-boringssl-static) allows a remote TLS peer to crash the JVM process by pushing handshake continuation, renegotiation, or TLS 1.3 post-handshake bytes through the internal nonApplicationBuffer ring buffer until its un-normalized write index exceeds the buffer size, after which a memcpy runs with an out-of-bounds pointer and a negative length cast to size_t. No authentication or user interaction is required (PR:N/UI:N), but reliability is limited by the CVSS 4.0 AT:P flag: the overflow only materializes when the integrating application chose a comparatively small nonApplicationBufferSize via SSL.bioNewByteBuffer and traffic arrives in write/partial-drain cycles that never fully empty the buffer. The assessed impact is availability-only (VA:H) - a deterministic process crash - with no confirmed confidentiality or integrity effect; no CISA KEV listing exists and no released working exploit code was identified at time of analysis, although the upstream advisory documents a deterministic reproduction PoC built from the public JNI API.
Memory exhaustion in the Erlang hackney HTTP client (2.0.0-beta.1 through 4.8.x) lets any server the client requests force the connection process to buffer an entire response body with no size bound; when {auto_decompress, true} is set, a small compressed body is additionally inflated in one zlib call, multiplying the allocation. The default hackney:request/5 and hackney:body/1 paths buffer the full body, so unauthenticated remote attackers who control or influence the target server (directly, via redirect, or via an SSRF-style URL) can crash or stall the calling BEAM process, giving availability-only impact. No public exploit code was identified at time of analysis and there is no KEV confirmation; 4.9.0 introduces a max_body_size option (default 512 MiB) that caps buffering and returns {error, body_too_large}.
HTTP/2 resource-limit bypass in the Elixir Bandit web server (1.9.0 through 1.13.x) lets an unauthenticated remote client exceed the configured max_concurrent_streams and max_requests caps by packing multiple HEADERS frames into a single TCP write where one frame raises a StreamError, causing the handler to continue from pre-chunk connection state. Because stream counts, last_stream_id, rapid-reset timestamps and the HPACK decoder state are all rolled back while the already-started Plug calls keep running, an attacker can drive roughly five times the intended number of concurrent Plug invocations per connection (50 running against a limit of 10 in the reporter's test), exhausting server processes, CPU and memory. Impact is strictly availability - no confidentiality or integrity loss - HTTP/2 with a finite max_concurrent_streams is the default for Bandit's TLS/h2 listeners, no public exploit code is confirmed, and a vendor patch is available.
Stored cross-site scripting in Mautic 7.0.0 through 7.2.1 allows anyone who can write a value into an HTML-type custom contact field - through contact editing, the contact API, or a form submission mapped to that field - to plant script that executes in a victim's browser session when the affected contact's detail page is opened, potentially letting the attacker create administrator accounts, change system configuration, or read contact and campaign data. The independent assessment records the write path as authenticated (PR:L), with an unauthenticated route existing only where a public form is deliberately mapped to an HTML field, whereas the vendor scored the same issue PR:N; exploitation additionally requires a configured HTML-type field and a second, more-privileged user opening the record (UI:R). No public exploit code or PoC was identified in the available intelligence, and no active exploitation is indicated; Mautic 7.2.2 is the only complete fix, with 4.x, 5.x and 6.x confirmed unaffected.
HAVELSAN Liman MYS releases from 2.3.4-1124 up to but not including 2.3.5-1131 contain a weak forgotten-password recovery mechanism (CWE-640) that lets a network attacker abuse the account-recovery workflow to take over a targeted account, with high confidentiality and integrity impact and no effect on availability. The CVSS 3.1 vector (AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:N, base 8.1) indicates the flaw is network-reachable and low complexity and needs no attacker credentials, but it is not fully autonomous because a victim must perform some recovery-related action (UI:R); the exact weakness in the recovery flow is not described in the available sources. No CISA KEV entry and no exploit code reference were provided, so no public exploit identified at time of analysis; TR-CERT rates it for remediation within this week.
PHP object injection in Mautic's segment filters and custom field properties can escalate to server-side code execution because stored serialized values are deserialized without class restrictions and Mautic's dependencies may supply usable gadget chains. The flaw affects Mautic core versions before 5.2.12, 6.0.10 and 7.2.0; exploitation requires high privileges or an existing ability to write the affected stored data, and the payload fires when Mautic later evaluates segments, exports contacts or serves custom field API requests. No public exploit was identified at the time of analysis, and the vendor advisory lists fixed releases for each maintained branch.
Kernel-level code execution is possible on Apple iOS/iPadOS and macOS because of a use-after-free in kernel memory handling, allowing an app that is already running on the device to execute arbitrary code with kernel privileges (CVSS 7.8, CWE-416). The flaw is fixed in iOS 26.7 and iPadOS 26.7, iOS 27 and iPadOS 27, macOS Golden Gate 27, macOS Sequoia 15.8 and macOS Tahoe 26.7.2, so unpatched devices across all those branches are exposed. Impact is total once triggered, but the attack path is local - the attacker must first get an app onto the device (AV:L, PR:L, UI:N) - and no public exploit has been identified at time of analysis; EPSS is 0.20% (10th percentile) and CISA SSVC records exploitation: none, automatable: no, which places this in a this-week rather than emergency remediation queue.
This issue was addressed with improved file handling. This issue is fixed in macOS Golden Gate 27.2, macOS Sequoia 15.8.2, macOS Tahoe 26.7.2. A malicious application may be able to execute arbitrary code with system privileges.
Windows command injection in the systeminformation Node.js library before 5.33.16 lets an attacker who controls the service-name argument to services() run arbitrary PowerShell expressions with the Node.js process account's privileges. The flaw is in the Windows-only branch that builds a Get-CimInstance Win32_Service -Filter command; a downstream application must forward attacker-controlled input to services() or getDynamicData() because the package exposes no network service itself. The advisory includes a proof of concept that the reporter reproduced on Windows, but EPSS is low at 0.14% (4th percentile) and there is no CISA KEV listing, so risk is genuine but bounded rather than an emergency.
Crafted GES project files processed by GStreamer Editing Services before 1.28.8 can embed an arbitrary GStreamer pipeline description inside an effect definition that is parsed and instantiated with no restriction on elements or properties, letting an attacker read or write arbitrary local files, force outbound requests to attacker-controlled servers, or crash the host application, with file writes potentially chained into code execution. The vulnerable component is local (AV:L) and requires the file to be opened or auto-processed (UI:R), but no privileges are needed (PR:N); the autoplugging/thumbnailing path means a media player or file-manager thumbnailer can instantiate the pipeline without a deliberate user open. No public exploit code was identified at time of analysis, and the vendor has released a fix in gst-editing-services 1.28.8; this is rated a this-week priority rather than an inflated high-CVSS item.
Feed filtering bypass in MISP mispguard allows a compromised or attacker-controlled MISP feed host to push unfiltered event data into the protected MISP instance by answering with any 2xx status other than 200 (201 Created, 203 Non-Authoritative Information, 206 Partial Content), because the pre-patch process_feed_response logic only reasoned about the non-200 case and let every other successful code through untouched. The attacker must already control or have compromised a feed endpoint configured and polled by that MISP instance; from there no authentication or user interaction is required beyond the feed polling loop already in place. Upstream fix commit b5f606a4e57a49cc4dbf9b3ef0c6ddd19f866735 changes these responses to a 403 block, no public exploit was identified at time of analysis and active exploitation is not confirmed (CISA KEV), and vuln.today priority is this_week.
Forescout SecureConnector for Windows up to and including 11.3.20 can be induced to establish a session with an attacker-controlled server because trusted-server entries in the agent configuration may omit TLS certificate fingerprints once the list grows past three entries, allowing a network-adjacent attacker to redirect agent traffic and then abuse the agent's built-in command and file features to run arbitrary commands as SYSTEM. The prerequisites are narrow: the non-default 'Additional Appliance Connections' setting must be enabled, the configuration file must hold more than three trusted-server entries with at least some fingerprints missing, and the attacker must control the victim's wired or wireless network to perform the redirect - this is not internet-facing exploitation, and the assessed CVSS 3.1 vector (AV:A/AC:H/PR:N/UI:N, 7.1) sits below the headline 7.7 for exactly that reason. No public exploit code and no CISA KEV entry were identified at time of analysis, EPSS is 0.15% (4th percentile), and the priority signal is 'this_week'; PR:N in the vector means no credentials or user interaction are required, but the adjacency and multi-trusted-server misconfiguration sharply limit who can realistically trigger it.
Type confusion in Timescale's pgvectorscale extension (all versions before 0.9.1) lets a low-privileged database user who holds CREATE on the vectorscale schema pre-create shadow objects that impersonate pgvector's vector type or operators, then build and query a DiskANN index over crafted values to drive out-of-bounds reads and writes. Successful exploitation can disclose process memory, crash the PostgreSQL server, or escalate privileges within the database; both pgvector and a vulnerable pgvectorscale build must be installed, and the source CVSS 4.0 score of 7.7 reflects full confidentiality, integrity and availability impact confined to the vulnerable system. No public exploit code has been identified at time of analysis and the issue is not in CISA KEV; the vendor-priority decision is 'this_week' and fixed builds ship in 0.9.1.
Bucket squatting in Google Cloud Run MCP before 1.11.0 lets an unauthenticated remote attacker who pre-registers a predictable Cloud Storage bucket intercept proprietary source code staged by the tool, tamper with deployments to execute arbitrary code, and exfiltrate data. Real-world urgency is tempered below the raw CVSS 4.0 base score of 7.7 because the attacker must claim the predictable bucket before the tool does and a developer must then run a deployment through the vulnerable instance. No CISA KEV entry or public exploit was identified at time of analysis, and the vendor's stated fix is upgrading to 1.11.0 or later.
Filesystem jail escape in Syncplify Server! before 8.8.0 lets an authenticated low-privileged user read restricted files outside the virtual filesystem root by placing a symbolic link or, on Windows, a directory junction in their home folder. It requires a valid account, the ability to create such links, and knowledge or discovery of a reachable sensitive target outside the jail. No public exploit was identified at time of analysis; the vendor patch is 8.8.0 and the issue is prioritized for this week.
Cross-tenant record takeover in the Ash framework (Elixir) for versions 1.22.0 up to 3.37.2: when a resource uses attribute multitenancy and exposes a create action with upsert? true that accepts the primary key (or upserts on an all_tenants? identity), an authenticated user of one tenant can supply another tenant's record ID and have that record updated, its full attribute set returned, and - with default upsert fields - its tenant attribute rewritten, moving the record into the attacker's tenant. Read and update policies do not run on the upsert path, and tenant-scoped create policies pass because the tenant on the request is the caller's own, so the only meaningful preconditions are that such an action is exposed and that the attacker knows or can enumerate a foreign key (for example a serial integer ID or an ID leaked through a shared link or export). The vendor has released a fix in ash 3.37.2; no public exploit code was identified at time of analysis, and the priority signal is to patch this week rather than an emergency response.
MongoDB Shell (mongosh) versions prior to 2.14.0 bundle a Node.js runtime whose TLS root store includes recoverable eMudhra intermediate CA certificates, so an unauthenticated on-path attacker who can present a certificate chain under that CA could intercept and alter TLS connections when a victim initiates a mongosh TLS session. The issue is CWE-295 improper certificate validation and is rated CVSS 4.0 7.6 (High); exploitation requires a MITM network position plus victim-initiated interaction, and npm-based installs instead depend on the host Node.js runtime. The vendor fix is mongosh 2.14.0 bundling Node.js 24.22.1 or later, and no public exploit identified at time of analysis.
Mendix SAML modules for Mendix 9, 10 and 11 fail to block replay of SAML assertions in IdP-initiated authentication flows, allowing an unauthenticated remote attacker to gain unauthorized application access by re-presenting a valid SAML response. Three module branches are affected: the Mendix 10 and Mendix 11 compatible modules below V4.2.4 and the Mendix 9 compatible module below V3.6.29. Per the assessed CVSS 4.0 vector (7.6, High), the attack is network-reachable and requires no credentials (PR:N), but difficulty is raised by AC:H (the attacker must first obtain a valid, unexpired SAML response) and UI:P (a legitimate user's IdP-initiated authentication must occur to produce that response); no public exploit was identified at time of analysis and no KEV entry exists.
Deserialization in Seroval prior to 1.6.8 accepts sequence nodes whose done and throw indices are unvalidated, so a crafted payload yields an iterator that never terminates; a related array-node flaw lets a node-supplied length drive the decoding loop. Affected are any services that pass untrusted input to fromJSON or fromCrossJSON, which a remote unauthenticated attacker can drive to block the event loop, exhaust memory, or take the service down (CVSS 7.5, availability-only). A public advisory, commit and test case are available; no public exploit code is confirmed and EPSS is low at 0.45% (39th percentile), with no CISA KEV listing.
Denial of service in the Go JOSE library (go-jose) allows remote attackers to burn excessive CPU by supplying a JWK containing crafted RSA parameters, because versions up to 4.1.5 pass the attacker-controlled modulus and primes straight into crypto/rsa.PrivateKey.Validate, which on Go 1.24+ runs in cubic time relative to the size of those values. Any service that calls ParseSigned, ParseEncrypted or unmarshals untrusted JWK/JWK Set data on a Go 1.24+ toolchain is exposed; exposures are limited to availability (CVSS 7.5, AV:N/AC:L/PR:N/UI:N, C:N/I:N/A:H), and the issue is not known to be exploited (CISA SSVC exploitation: none; EPSS 0.37%, 30th percentile). The vendor advisory ships a public proof-of-concept reproducer; go-jose/v4 is fixed in 4.1.6, while the v3 branch (<= 3.0.5) has no patched release yet, and the current priority signal places remediation in this week's patch cycle.
Denial of service in the seroval npm package (versions 1.4.1 through 1.6.7) lets an unauthenticated remote attacker who can get a crafted payload into fromJSON or fromCrossJSON exhaust process memory or block the decoding thread indefinitely. The flaw is a residual bypass of the fix for CVE-2026-23957, because the deserializer still accepted an array-like object such as {"length":2000000} in place of a genuine element array, so allocation and loop counts followed the forged length rather than the actual bytes on the wire. It is fixed in 1.6.8; there is no confirmed active exploitation (not in CISA KEV, no public exploit identified at time of analysis) and EPSS is low at 0.36% (30th percentile), so exposure depends almost entirely on whether a service feeds untrusted input to these decoders.
Denial of service in Esri ArcGIS Server 12.0 and earlier can occur when crafted requests exhaust all available worker threads, denying service to legitimate users. The issue is remotely reachable without authentication or user interaction, but Esri describes the thread exhaustion as occurring only under exceptional circumstances and does not specify the exact trigger. Patches are available for the General Availability and Extended support branches 12.0, 11.5, 11.3, and 11.1; no public exploit was identified at time of analysis.
Unauthenticated remote attackers who can reach a Consul agent's HTTP API can exhaust the agent's memory, causing a denial of service; per the CVSS vector the attack needs no credentials, no user interaction and is low-complexity, but the impact is confined to availability of the agent process with no confidentiality or integrity effect. Both Consul Community Edition and Consul Enterprise are affected across the version ranges covered by the CPE entries, and the primary limiting factor is exposure of the HTTP API - deployments that bind it to loopback or gate it behind network controls have a much smaller reachable attack surface. The only complete fix stated by the vendor is an upgrade to Consul Community Edition 2.1.0, or Consul Enterprise 2.1.0 or 1.21.19; no public exploit code or PoC reference was included in the supplied intelligence, and no CISA KEV listing was reported.