Overview 176,750 matches · 8,352 in Waratek scope · 532 protected · 1,021 exploited · 184 CISA KEV · 11,846 public PoC · 701 EPSS ≥ 0.5
176,750matches
8,352in Waratek scope4.7%
532protected0.3%
1,021exploited0.6%
184CISA KEV0.1%
11,846public PoC6.7%
701EPSS ≥ 0.50.4%
Protected by Waratek
534
0.3%
Rule in development
2
0.0%
Mitigation candidate
218
0.1%
No exploit published
7,598
4.3%
Outside Waratek scope
1,746
1.0%
Not applicable
166,652
94.3%
Withdrawn CVE
0
0.0%
split by
peak 38,000 / year
19913
Not applicable3
Medium3
not blocked3
No public exploit3
19920
19933
Not applicable3
Medium3
not blocked3
No public exploit3
199410
Not applicable10
Medium10
not blocked10
Working exploit published2
No public exploit8
19956
Not applicable6
Medium6
not blocked6
Working exploit published2
No public exploit4
199621
Not applicable21
Medium21
not blocked21
Working exploit published5
No public exploit16
199789
Not applicable89
Medium89
not blocked89
Working exploit published18
No public exploit71
199885
Not applicable85
Medium85
not blocked85
Working exploit published11
No public exploit74
1999357
Outside Waratek scope1
Not applicable356
Medium357
not blocked357
Working exploit published119
No public exploit238
2000467
Mitigation candidate4
No exploit published3
Not applicable460
Medium467
not blocked467
Exploited2
Working exploit published195
No public exploit270
2001706
Mitigation candidate3
No exploit published5
Outside Waratek scope5
Not applicable693
Medium706
not blocked706
Exploited3
Working exploit published219
Forecast only2
No public exploit482
2002980
Protected by Waratek2
Mitigation candidate6
No exploit published16
Outside Waratek scope1
Not applicable955
Medium980
blocked by ARMR today2
not blocked978
Exploited3
Working exploit published246
Forecast only3
No public exploit728
2003747
Protected by Waratek2
Mitigation candidate5
No exploit published18
Not applicable722
Medium747
blocked by ARMR today2
not blocked745
Working exploit published243
No public exploit504
20041,265
Protected by Waratek1
Mitigation candidate1
No exploit published16
Outside Waratek scope4
Not applicable1,243
Medium1,265
blocked by ARMR today1
not blocked1,264
Exploited10
Working exploit published414
Forecast only1
No public exploit840
20052,430
Protected by Waratek3
Mitigation candidate4
No exploit published40
Outside Waratek scope5
Not applicable2,378
Medium2,430
blocked by ARMR today3
not blocked2,427
Exploited8
Working exploit published622
Forecast only6
No public exploit1,794
20063,324
Protected by Waratek4
Mitigation candidate4
No exploit published49
Outside Waratek scope2
Not applicable3,265
Medium3,324
blocked by ARMR today4
not blocked3,320
Exploited12
Working exploit published1,135
Forecast only2
No public exploit2,175
20073,101
Protected by Waratek3
Mitigation candidate11
No exploit published67
Outside Waratek scope42
Not applicable2,978
Medium3,101
blocked by ARMR today3
not blocked3,098
Exploited14
Working exploit published1,052
Forecast only1
No public exploit2,034
20082,582
Protected by Waratek4
Mitigation candidate8
No exploit published73
Outside Waratek scope23
Not applicable2,474
Medium2,582
blocked by ARMR today4
not blocked2,578
Exploited7
Working exploit published987
Proof of concept only2
Forecast only1
No public exploit1,585
20092,786
Protected by Waratek5
Mitigation candidate9
No exploit published121
Outside Waratek scope25
Not applicable2,626
Medium2,786
blocked by ARMR today5
not blocked2,781
Exploited17
Working exploit published1,122
Proof of concept only2
Forecast only3
No public exploit1,642
20102,241
Protected by Waratek6
Mitigation candidate10
No exploit published118
Outside Waratek scope33
Not applicable2,074
Medium2,241
blocked by ARMR today6
not blocked2,235
Exploited14
Working exploit published547
Proof of concept only1
Forecast only2
No public exploit1,677
20112,101
Protected by Waratek4
Mitigation candidate4
No exploit published156
Outside Waratek scope33
Not applicable1,904
Medium2,101
blocked by ARMR today4
not blocked2,097
Exploited9
Working exploit published294
Proof of concept only3
Forecast only1
No public exploit1,794
20123,035
Protected by Waratek5
Mitigation candidate20
No exploit published187
Outside Waratek scope35
Not applicable2,788
Medium3,035
blocked by ARMR today5
not blocked3,029
not established1
Exploited20
Working exploit published475
Proof of concept only2
Forecast only42
No public exploit2,496
20132,914
Protected by Waratek12
Mitigation candidate13
No exploit published234
Outside Waratek scope144
Not applicable2,511
Medium2,914
blocked by ARMR today12
not blocked2,901
not established1
Exploited23
Working exploit published237
Proof of concept only1
Forecast only6
No public exploit2,647
20145,314
Protected by Waratek32
Mitigation candidate9
No exploit published224
Outside Waratek scope150
Not applicable4,899
Medium5,314
blocked by ARMR today32
not blocked5,282
Exploited24
Working exploit published415
Proof of concept only7
Forecast only12
No public exploit4,856
20153,501
Protected by Waratek20
Mitigation candidate2
No exploit published222
Outside Waratek scope107
Not applicable3,150
Medium3,501
blocked by ARMR today20
not blocked3,481
Exploited24
Working exploit published254
Proof of concept only7
Forecast only8
No public exploit3,208
20162,446
Protected by Waratek26
Mitigation candidate3
No exploit published297
Outside Waratek scope45
Not applicable2,075
Medium2,446
blocked by ARMR today26
not blocked2,419
not established1
Exploited16
Working exploit published122
Proof of concept only6
Forecast only9
No public exploit2,293
20175,687
Protected by Waratek24
Mitigation candidate5
No exploit published336
Outside Waratek scope110
Not applicable5,212
Medium5,687
blocked by ARMR today24
not blocked5,663
Exploited18
Working exploit published332
Proof of concept only21
Forecast only2
No public exploit5,314
20186,163
Protected by Waratek54
Mitigation candidate13
No exploit published444
Outside Waratek scope82
Not applicable5,570
Medium6,163
blocked by ARMR today54
not blocked6,109
Exploited38
Working exploit published348
Proof of concept only27
Forecast only7
No public exploit5,743
20197,117
Protected by Waratek35
Mitigation candidate17
No exploit published427
Outside Waratek scope116
Not applicable6,522
Medium7,117
blocked by ARMR today34
partly blocked1
not blocked7,082
Exploited48
Working exploit published296
Proof of concept only30
Forecast only8
No public exploit6,735
20207,359
Protected by Waratek35
Mitigation candidate17
No exploit published517
Outside Waratek scope131
Not applicable6,659
Medium7,359
blocked by ARMR today35
not blocked7,324
Exploited58
Working exploit published197
Proof of concept only37
Forecast only10
No public exploit7,057
20218,137
Protected by Waratek29
Mitigation candidate12
No exploit published301
Outside Waratek scope159
Not applicable7,636
Medium8,137
blocked by ARMR today29
not blocked8,108
Exploited88
Working exploit published240
Proof of concept only90
Forecast only43
No public exploit7,676
20229,908
Protected by Waratek58
Mitigation candidate3
No exploit published575
Outside Waratek scope113
Not applicable9,159
Medium9,908
blocked by ARMR today58
not blocked9,850
Exploited73
Working exploit published270
Proof of concept only953
Forecast only34
No public exploit8,578
202311,824
Protected by Waratek29
Mitigation candidate10
No exploit published441
Outside Waratek scope96
Not applicable11,248
Medium11,824
blocked by ARMR today29
not blocked11,795
Exploited100
Working exploit published236
Proof of concept only2,407
Forecast only26
No public exploit9,055
202418,998
Protected by Waratek19
Mitigation candidate5
No exploit published322
Outside Waratek scope110
Not applicable18,542
Medium18,998
blocked by ARMR today19
not blocked18,979
Exploited129
Working exploit published114
Proof of concept only3,597
Forecast only10
No public exploit15,148
202523,043
Protected by Waratek37
Rule in development2
Mitigation candidate14
No exploit published473
Outside Waratek scope78
Not applicable22,439
Medium23,043
blocked by ARMR today37
not blocked23,006
Exploited134
Working exploit published170
Proof of concept only5,242
Forecast only5
No public exploit17,492
202638,000
Protected by Waratek85
Mitigation candidate6
No exploit published1,916
Outside Waratek scope96
Not applicable35,897
Medium38,000
blocked by ARMR today84
partly blocked1
not blocked37,915
Exploited129
Working exploit published126
Proof of concept only6,960
No public exploit30,785
199319961999200220052008201120142017202020232026
| CVE | Severity | Published▾ | Status | Exploitation | Description |
|---|---|---|---|---|---|
| CVE-2026-133958 | Medium | 2026-11-21 | Not applicable | No public exploit | A flaw was found in Quay. Insufficient validation of repository notification webhook URLs allows an authenticated administrator to trigger Server-Side Request Forgery (SSRF). By specifying an internal URL or a redirect to an internal destination, an attacker can force the Quay worker to issue unauthorized HTTP requests to internal network locations and cloud metadata services. |
| CVE-2026-133471 | Medium | 2026-11-21 | Not applicable | No public exploit | The Fastcache by Host.it plugin for WordPress is vulnerable to Reflected Cross-Site Scripting via the 's' parameter in all versions up to, and including, 1.7.14 due to insufficient input sanitization and output escaping. This makes it possible for unauthenticated attackers to inject arbitrary web scripts in pages that execute if they can successfully trick a user into performing an action such as clicking on a link. This code path is only reachable when both the Light Images feature and the DOMDocument entity-decode branch are enabled (lightimgs_status = 1 and img_processing_entity_decode = 1), which reflect the default plugin configuration. |
| CVE-2026-133439 | Medium | 2026-11-21 | Not applicable | No public exploit | The Coupon Creator plugin for WordPress is vulnerable to Stored Cross-Site Scripting via the 'post_title' parameter in all versions up to, and including, 3.6.5 due to insufficient input sanitization and output escaping. This makes it possible for authenticated attackers, with author-level access and above, to inject arbitrary web scripts in pages that will execute whenever a user accesses an injected page. This vulnerability is only exploitable when an administrator has enabled the 'Disable Print View' option (cctor_hide_print_link), as this setting is required to reach the vulnerable else branch; the default configuration is not affected. |
| CVE-2026-132781 | Medium | 2026-11-21 | Not applicable | No public exploit | The Booking Calendar plugin for WordPress is vulnerable to Stored Cross-Site Scripting via the plugin's 'booking' shortcode in all versions up to, and including, 12.0 due to insufficient input sanitization and output escaping on user supplied attributes. This makes it possible for authenticated attackers, with contributor-level access and above, to inject arbitrary web scripts in pages that will execute whenever a user accesses an injected page. |
| CVE-2026-131651 | Medium | 2026-11-21 | Not applicable | No public exploit | The rtMedia for WordPress, BuddyPress and bbPress plugin for WordPress is vulnerable to Stored DOM-Based Cross-Site Scripting via the 'media_title' parameter in all versions up to, and including, 4.7.18 due to insufficient input sanitization and output escaping. This makes it possible for authenticated attackers, with subscriber-level access and above, to inject arbitrary web scripts in pages that will execute whenever a user accesses an injected page. The vulnerability is only triggerable via the Backbone.js-rendered "Load More" pagination path; the server-side PHP gallery rendering correctly escapes media_title with esc_attr() and is not affected. |
| CVE-2026-125932 | Medium | 2026-11-21 | Not applicable | No public exploit | The Coupon Creator plugin for WordPress is vulnerable to Stored Cross-Site Scripting via the 'metavalue (with metakeyinput=cctor_expiration)' parameter in all versions up to, and including, 3.6.5 due to insufficient input sanitization and output escaping. This makes it possible for authenticated attackers, with contributor-level access and above, to inject arbitrary web scripts in pages that will execute whenever a user accesses an injected page. The exploit is possible because the cctor_coupon post type advertises 'custom-fields' support, enabling WordPress core's Custom Fields metabox for Contributors, and the unregistered cctor_expiration meta key causes sanitize_meta() to be a no-op, allowing raw HTML to be stored verbatim in the database via a single AJAX request. |
| CVE-2026-119607 | Medium | 2026-11-21 | Not applicable | No public exploit | The Salon Booking System – Appointment Booking for Salons, Barbershops & Spas plugin for WordPress is vulnerable to authorization bypass in all versions up to, and including, 10.31.16. This is due to the plugin not properly verifying that a user is authorized to perform an action. This makes it possible for authenticated attackers, with subscriber-level access and above, to overwrite arbitrary users' user meta fields, including the _sln_phone and _sln_sms_prefix meta of administrator accounts, and create bookings attributed to arbitrary users. This is exploitable on default installations because reCAPTCHA is disabled by default, causing the booking endpoint's permission check to return true for any authenticated user without a meaningful authorization gate. |
| CVE-2026-95827 | Medium | 2026-11-20 | Not applicable | No public exploit | The IPv6 neighbour discovery code in subsys/net/ip/ipv6_nbr.c keeps packets that are waiting for address resolution on a per-neighbour pending queue. A queued packet holds two references: the one handed over by the sender and a second one taken by net_ipv6_send_ns() when the packet is put on the queue. The Router Advertisement handler handle_ra_input() popped each packet off the queue and passed it to net_send_data(), relying on a trailing nbr_clear_ns_pending() call to release the queue reference — but that call then found the queue already empty and released nothing, so one reference was leaked for every pending packet flushed on an RA. Router Advertisements are unauthenticated link-local ICMPv6 messages, so any node on the same link can trigger the flush. An attacker can make the leak deterministic by advertising itself as a router without a source link-layer address option, so the victim's neighbour entry stays INCOMPLETE, then inducing the victim to transmit through that next hop (the packet is queued and a neighbour solicitation is sent), leaving the solicitation unanswered and sending a further RA to drain the queue. Each round leaks one struct net_pkt together with its net_buf chain. Leaked packets are never returned to the fixed-size pools, so the loss is permanent for the lifetime of the boot. With CONFIG_NET_PKT_TX_COUNT defaulting to 4 (14 with CONFIG_NET_PKT_LARGE_DEFAULTS), a small number of repetitions exhausts the transmit packet slab and the node stops being able to send any traffic. The impact is availability only: the reference count is always too high, never too low, so no use-after-free or memory corruption arises. The fix moves the flush into a helper, nbr_send_pending(), which releases the queue reference for every packet it takes off the queue before handing the remaining reference to net_send_data(), and releases that one too if the send is refused. |
| CVE-2026-95825 | Medium | 2026-11-20 | Not applicable | No public exploit | net_if_ipv4_addr_rm() and net_if_ipv6_addr_rm() in subsys/net/ip/net_if.c unconditionally called net_if_addr_unref() on the target address. When an earlier removal could not complete because another subsystem still held a reference (the TCP stack takes one on the local address of every connection, in tcp_conn_new() and on connect()), the address is left in place with is_added cleared and is_used still set. A second removal of such an address decremented the reference count again, consuming a reference owned by that other holder. When it was the last one, the address slot was released, its NET_EVENT_IPV*_ADDR_DEL event fired, and the slot became reusable while the holder still believed it owned a reference. Every address lookup helper matches on is_used and not on is_added, so a half-removed address stays visible to the rest of the stack and repeat removals are readily reachable from remote input. Under IPv6 SLAAC, an on-link node that sends a Router Advertisement with a short valid lifetime causes the address to expire into the deferred state; a refreshing Router Advertisement is handled by net_if_ipv6_addr_update_lifetime(), which re-arms the lifetime timer without restoring the reference, so the next expiry performs the second, reference-stealing removal. With CONFIG_NET_IPV4_ACD and DHCPv4, repeated spoofed ARP conflicts for the device's lease drive dhcpv4_acd_event_handler() into calling net_if_ipv4_addr_rm() more than once on the same address. Where CONFIG_USERSPACE is enabled, the net_if_ipv4_addr_rm_by_index() and net_if_ipv6_addr_rm_by_index() system calls let an unprivileged user-mode thread trigger the same kernel-side release deterministically. The impact is loss of a configured IP address while connections are still using it, and a corrupted reference count that propagates: the slot can be handed to a newly added address, and when the stale holder later tears down, net_if_addr_unref() resolves the address by value and drops the new incarnation's only reference, removing a live address a second time. The affected objects are static array slots and the counter is never decremented below zero, so there is no memory corruption or information disclosure; the consequence is denial of service against the device's network connectivity, repeatable at the attacker's pace. The fix resolves the address before touching the reference count and returns false without calling net_if_addr_unref() when the address is already marked as removed, so the public removal API can no longer consume a reference it does not own. |
| CVE-2026-95824 | Medium | 2026-11-20 | Not applicable | No public exploit | The IPv6 Neighbor Discovery code in subsys/net/ip/ipv6_nbr.c flushes a neighbour's queued packets at the end of handle_ra_input(). The Source Link-Layer Address Option (SLLAO) is optional in a Router Advertisement; when it is absent, handle_ra_input() still obtains (or creates) a neighbour cache entry for the RA source via net_ipv6_nbr_add() with a NULL link-layer address, so the entry stays in NET_IPV6_NBR_STATE_INCOMPLETE with nbr->idx == NET_NBR_LLADDR_UNKNOWN. The pending-packet drain loop was entered unconditionally for that unresolved entry. Each net_send_data() in the drain loop runs synchronously into net_ipv6_prepare_for_send(), which finds the neighbour still unresolved and hands the packet to net_ipv6_send_ns(), which puts it straight back onto the same pending_queue. The loop therefore never sees an empty queue. With a single queued packet it transmits a fresh Neighbor Solicitation on every iteration; with two or more queued packets it rotates the queue with no allocation and no transmission at all. The loop runs in the network RX thread with the global IPv6 neighbour mutex held. An unauthenticated attacker on the same link can trigger this deterministically: advertise itself as a router, induce the victim to queue off-link traffic toward that router (for example by sending echo requests with an off-link source address) while leaving the solicitation unanswered, then send a second Router Advertisement without an SLLAO option. Spoofing an SLLAO-less RA from the legitimate router's link-local address has the same effect whenever traffic is queued for it. Router Advertisements are processed with no authentication whenever CONFIG_NET_IPV6_ND is enabled, which is the default for IPv6-capable builds. The result is a permanent denial of service: the RX thread spins forever holding the neighbour lock, so all IPv6 transmission and every other user of the neighbour cache blocks, and the Neighbor Advertisement that would resolve the entry can no longer be processed. The ipv6_ns_reply_timer work item that would normally drop the pending packets also blocks on the same mutex. In the single-packet variant the loop additionally floods the link with Neighbor Solicitations. The fix restricts the flush to resolved entries (nbr->idx != NET_NBR_LLADDR_UNKNOWN), leaving the packets queued for the NS reply timeout. There is no memory-safety impact. |
| CVE-2026-94599 | Medium | 2026-11-20 | Not applicable | No public exploit | NukeViet CMS is a multi Content Management System. From 4.5.00 until 4.5.13 and 4.6.02, an unauthenticated attacker can place a multiply encoded value in the news module search q parameter. $nv_Request->get_title() decodes and sanitizes the input once, but modules/news/funcs/search.php subsequently applies urldecode() after sanitization. The resulting raw keyword reaches the {KEY} search-input attribute through search_theme() and the no-results element through search_result_theme() in the default and mobile themes, permitting attribute or tag injection and same-origin JavaScript execution when a victim opens the crafted search URL. This issue is fixed in versions 4.5.13 and 4.6.02. |
| CVE-2026-94566 | Medium | 2026-11-20 | Not applicable | Proof of concept only | NukeViet CMS is a multi Content Management System. In 4.6.00, an unauthenticated attacker can rotate X-Forwarded-For because vendor/vinades/nukeviet/Core/Ips.php accepts forwarded client-IP headers without first confirming that REMOTE_ADDR is a trusted proxy. The resulting NV_CLIENT_IP keys the administrator login blocker in includes/core/admin_login.php; changing only the header value bypasses that lockout and permits effectively unlimited password guesses. The same spoofable address is also consumed by the user login blocker in modules/users/funcs/login.php, IP restrictions, bans, flood controls, and security logs, but those additional code-level effects were not dynamically demonstrated. This issue is fixed in version 4.6.01. |
| CVE-2026-94563 | Medium | 2026-11-20 | Not applicable | No public exploit | NukeViet CMS is a multi Content Management System. From 4.5.00 until 4.6.01, an authenticated settings-module administrator can upload a PKCS#12 S/MIME certificate whose X.509 Subject CN contains path separators and traversal components. The smimeadd handler in admin/settings/smtp.php applies only an at-sign replacement before using the CN in .key, .crt, and .pem paths beneath data/certs. Traversal can select attacker-chosen paths outside data/certs, but writes remain constrained to already-existing writable directories and the permitted extensions. Within that constraint, attacker-controlled PEM data can overwrite certificate assets, including a DKIM private key; when the curl.cainfo PHP directive is empty, the attacker can also replace cacert.pem and undermine outbound TLS verification. This issue is fixed in version 4.6.01. |
| CVE-2026-86587 | Medium | 2026-11-20 | Not applicable | No public exploit | Certain EZVIZ devices are affected by an improper certificate validation issue (CWE-295). An attacker may intercept communication traffic to conduct a man-in-the-middle attack. |
| CVE-2026-86586 | Medium | 2026-11-20 | Not applicable | No public exploit | Certain EZVIZ devices are affected by a replay attack vulnerability (CWE-294). An attacker may intercept communication traffic and replay specific control commands, causing the device to execute unauthorized operations. |
| CVE-2026-82676 | Medium | 2026-11-20 | Not applicable | No public exploit | The GiveWP – Donation Plugin and Fundraising Platform plugin for WordPress is vulnerable to Cross-Site Request Forgery in all versions up to, and including, 4.16.7.2. This is due to missing nonce validation on the `give_render_donor_view()` function. This makes it possible for unauthenticated attackers to reassign an arbitrary GiveWP donor record to an arbitrary WordPress user via a forged request granted they can trick a site administrator into performing an action such as clicking on a link. The resulting association changes donor-account ownership and can satisfy the linked-donor precondition for a separately tracked account-takeover vulnerability (CVE-2026-82675). |
| CVE-2026-77390 | Medium | 2026-11-20 | Not applicable | No public exploit | pmo-platform is a modular PMO and release-management platform for Claude Code. Prior to 3.73, release/skills/pmo-skill-refiner/eval-viewer/generate_review.py and viewer.html contain three security weaknesses in the local evaluation viewer. During workspace scanning and generation, _find_runs_recursive and build_run follow symlinks and embed_file reads their targets without confining resolved paths to the workspace, although the server do_GET path is not a traversal surface. embed_file also reads output files without a size cap or redaction, allowing secrets or personally identifiable information to be embedded in a no-auth locally served page or shareable static file. Separately, do_POST /api/feedback writes caller-supplied JSON without authentication, an Origin check, or a cross-site request forgery token, allowing a malicious page or DNS-rebound localhost target to modify feedback while the server runs. This issue is fixed in version 3.73. |
| CVE-2026-77349 | Medium | 2026-11-20 | Not applicable | Proof of concept only | pmo-platform is a modular PMO and release-management platform for Claude Code. From 3.48 until 3.69.1, release/skills/pmo-skill-refiner/eval-viewer/generate_review.py reads untrusted model or skill output from workspace/run*/outputs/ and embeds it with json.dumps into viewer.html inside an inline script without escaping HTML metacharacters. A crafted output can close the script element and execute JavaScript when an operator generates and opens the viewer. The maintainer also confirmed that DOM-based injection through renderGrades and renderBenchmark innerHTML survives transport-only escaping, so affected render sinks must neutralize the content. Executed script can read embedded prompts and outputs, alter the viewer DOM, or modify saved feedback. This issue is fixed in version 3.69.1. |
| CVE-2026-77346 | Medium | 2026-11-20 | Not applicable | No public exploit | Pozeiden is a MermaidJS renderer written in Zig. Prior to 0.3.0, Pozeiden emits attacker-controlled Mermaid values into SVG attributes or markup without sufficient XML escaping or URL scheme validation, and first-party consumers such as zigmark and policypress embed the generated SVG raw into HTML. The click href syntax can introduce a dangerous URL scheme or break out of an attribute; style, classDef, ::: class assignment, and linkStyle color values can inject fill or stroke markup; color and font-weight values can escape text attributes; %%{init}%% themeVariables can inject global renderer attributes; and quadrant y-axis labels can introduce raw markup. These paths involve src/svg/writer.zig, src/renderers/flowchart.zig, src/renderers/quadrant.zig, and src/root.zig. An attacker who can influence rendered Mermaid source can therefore store script-bearing SVG that executes when another user views the published HTML. This issue is fixed in version 0.3.0. |
| CVE-2026-77328 | Medium | 2026-11-20 | Not applicable | No public exploit | MapServer is a system for developing web-based GIS applications. Prior to 8.6.5, MapServer's PostgreSQL JOIN implementation in src/mappostgresql.c copies the source layer attribute selected by JOIN FROM into joininfo->from_value and embeds it without parameterization or escaping into a dynamically generated PostgreSQL query executed by PQexec(). When PostgreSQL JOIN support is enabled, an attacker who can control that source feature attribute can inject quote characters and SQL operators that alter the intended WHERE predicate. This second-order injection can cause unintended rows and JOIN attributes to be returned, potentially exposing information from unrelated PostgreSQL records. This issue is fixed in version 8.6.5. |
| CVE-2026-77327 | Medium | 2026-11-20 | Not applicable | Proof of concept only | MapServer is a system for developing web-based GIS applications. Prior to 8.6.5, MapServer's MySQL JOIN implementation in src/mapjoin.c copies the source layer attribute selected by JOIN ... FROM into joininfo->target and inserts it into the query built by msMySQLJoinNext() without parameterization or SQL escaping. When MySQL JOIN support is enabled, an attacker who can control that source feature attribute, including through imported shapefile or DBF data or another untrusted spatial source, can inject SQL operators or expressions that change the intended JOIN predicate. This can bypass the intended key comparison and expose unrelated records from the MySQL or MariaDB JOIN table through output templates, query responses, or downstream application logic. This issue is fixed in version 8.6.5. |
| CVE-2026-77325 | Medium | 2026-11-20 | Not applicable | No public exploit | MapServer is a system for developing web-based GIS applications. From 6.0 until 8.6.5, MapServer's WMS OpenLayers output, selected by FORMAT=application/openlayers or template=openlayers, substitutes a request-derived online-resource value into a single-quoted JavaScript string in src/maptemplate.c. The prior remediation for GHSA-xqj6-vjqr-33vv used msEscapeJSonLikeString() to protect the JavaScript string but did not neutralize angle brackets, so attacker-controlled X-Forwarded-Host or X-Forwarded-Proto values can still terminate the script element at the HTML parser layer. Deployments that expose attacker-controlled PATH_INFO or map= values reaching the same string are also affected. An unauthenticated attacker can cause arbitrary JavaScript to execute in a victim's browser under the MapServer origin when the victim loads the crafted output. This issue is fixed in version 8.6.5. |
| CVE-2026-77323 | Medium | 2026-11-20 | Not applicable | No public exploit | MapServer is a system for developing web-based GIS applications. Prior to 8.6.5, an unauthenticated reflected cross-site scripting vulnerability in OGC API Features collection-items HTML output allows crafted offset or limit query parameter values from raw CGI request handling in src/mapogcapi.cpp to reach template.params.offset or template.params.limit. share/ogcapi/templates/html-bootstrap/collection-items.html inserts those values without quoting or escaping into a script element. A victim who follows a crafted request can have the script element terminated by attacker-controlled markup, causing arbitrary JavaScript to execute under the MapServer origin and enabling access to browser-session data, actions as the victim, and content manipulation. This issue is fixed in 8.6.5. |
| CVE-2026-59987 | Medium | 2026-11-20 | Not applicable | No public exploit | Avahi is a system which facilitates service discovery on a local network via the mDNS/DNS-SD protocol suite. In 0.9-rc4 and earlier, org.freedesktop.Avahi.Server.RecordBrowserNew accepts a domain-name value that avahi_is_valid_domain_name considers valid even though avahi_normalize_name_strdup cannot normalize it. The resulting null key reaches the key assertion in dbus_prepare_record_browser_object. A local low-privileged D-Bus client can trigger the mismatch and terminate the Avahi daemon, causing denial of service. The issue is fixed in 0.9-rc5. |
| CVE-2026-59968 | Medium | 2026-11-20 | Not applicable | Proof of concept only | LinkAce is a self-hosted archive to collect website links. Prior to 2.5.9, the authenticated GET /api/v2/lists/{list}/links endpoint in app/Http/Controllers/API/ListLinksController.php authorizes access only to the parent list and then returns its links without applying visibleForUser(). Because another user's INTERNAL or PUBLIC list may legitimately contain PRIVATE links, a normal authenticated user who may view the list can also receive those private link records, including their URLs, titles, descriptions, icons, and timestamps. Equivalent web, tag, notes, and feed handlers apply the per-link visibility scope, so this API handler is the affected exception. This issue is fixed in version 2.5.9. |
| CVE-2026-58406 | Medium | 2026-11-20 | Not applicable | No public exploit | Avahi is a system which facilitates service discovery on a local network via the mDNS/DNS-SD protocol suite. In 0.9-rc4 and earlier on NetBSD architectures such as x86_64 where char is signed and the ctype guard-page change is present, avahi_domain_hash can pass a negative value to tolower when a resource-record name contains a byte at or above 128. The invalid ctype input reliably raises SIGSEGV, allowing a remote peer on the local network to crash the Avahi daemon. This issue is fixed with 0.9-rc5. |
| CVE-2026-54046 | Medium | 2026-11-20 | Not applicable | Proof of concept only | LibreChat is an enhanced ChatGPT clone that supports multiple AI providers. Prior to 0.8.7, unauthenticated POST /api/user/verify requests in api/server/routes/user.js call api/server/services/AuthService.js verifyEmail and return distinct responses for unknown, verified, and unverified email addresses and the route lacks request throttling. A remote attacker can submit candidate addresses with invalid tokens to enumerate registered accounts and verification state, supporting targeted phishing or credential attacks. This issue is fixed in version 0.8.7. |
| CVE-2026-54045 | Medium | 2026-11-20 | Not applicable | No public exploit | LibreChat is an enhanced ChatGPT clone that supports multiple AI providers. Prior to 0.8.7, POST /api/auth/requestPasswordReset calls deleteTokens by user ID before creating a reset token, while password-reset and email-verification records in packages/data-schemas/src/methods/token.ts have no populated flow type. When password reset is enabled and email verification is enforced, an unauthenticated attacker who knows a target email can repeatedly delete pending verification tokens and prevent account verification until the unverified account expires. This issue is fixed in version 0.8.7. |
| CVE-2026-54043 | Medium | 2026-11-20 | Not applicable | No public exploit | LibreChat is an enhanced ChatGPT clone that supports multiple AI providers. Prior to 0.8.7, packages/data-schemas/src/methods/file.ts updateFileUsage accepts chat attachment files[].file_id values without restricting the lookup to the authenticated owner. An authenticated attacker who obtains another user's file identifier, including through GET /api/share/{shareId} for a shared conversation, can attach that file to a chat, causing extracted document text to enter the model context. The same operation can remove expiration metadata from the victim file. This issue is fixed in version 0.8.7. |
| CVE-2026-54042 | Medium | 2026-11-20 | Not applicable | No public exploit | LibreChat is an enhanced ChatGPT clone that supports multiple AI providers. Prior to 0.8.7, POST /api/auth/2fa/verify-temp in api/server/routes/auth.js calls verify2FAWithTempToken and accepts repeated TOTP or backup-code guesses without a dedicated attempt limiter. An attacker must first possess valid primary credentials and a valid five-minute temp token, but can then make repeated second-factor attempts and obtain a full session if a valid code is guessed. This issue is fixed in version 0.8.7. |
| CVE-2026-54038 | Medium | 2026-11-20 | Not applicable | No public exploit | LibreChat is an enhanced ChatGPT clone that supports multiple AI providers. Prior to 0.8.9, GET /api/permissions/search-principals passes the q parameter to JavaScript RegExp construction and MongoDB regex queries in packages/data-schemas/src/methods/userGroup.ts findGroupsByNamePattern and searchPrincipals without escaping metacharacters. An authenticated user with people-picker permission can submit a catastrophic-backtracking pattern that blocks the Node.js event loop and makes the service unresponsive to concurrent users; invalid patterns can also produce server errors. This issue is fixed in version 0.8.9. |
| CVE-2026-54032 | Medium | 2026-11-20 | Not applicable | No public exploit | LibreChat is an enhanced ChatGPT clone that supports multiple AI providers. Prior to 0.8.6, client/src/components/Chat/Input/MCPConfigDialog.tsx renders the remote MCP details.description value through dangerouslySetInnerHTML without sanitization, allowing a malicious MCP server to execute script when a user opens the configuration dialog. The secondary client/src/components/Banners/Banner.tsx sink similarly renders administrator-controlled banner.message HTML without sanitization for all users. Script execution occurs in the LibreChat browser origin and can expose browser-context data or alter displayed content. This issue is fixed in version 0.8.6. |
| CVE-2026-54031 | Medium | 2026-11-20 | Not applicable | No public exploit | LibreChat is an enhanced ChatGPT clone that supports multiple AI providers. Prior to 0.8.8, packages/api/src/mcp/oauth/handler.ts createOAuthFetch validates only the initial destination for MCP OAuth discovery, registration, token, refresh, and revocation requests, then uses redirect-following fetch behavior without validating each redirect target. An attacker who can initiate MCP OAuth against a controlled endpoint can redirect server requests to internal or private addresses, reaching services that the initial SSRF checks should block. The separate MCP transport path already validates redirects and is not affected. This issue is fixed in version 0.8.8. |
| CVE-2026-54026 | Medium | 2026-11-20 | Not applicable | No public exploit | LibreChat is an enhanced ChatGPT clone that supports multiple AI providers. Prior to 0.8.6, POST /api/messages/:conversationId authorizes req.params.conversationId in api/server/middleware/validateMessageReq.js but persists attacker-controlled req.body.conversationId through api/server/routes/messages.js. An authenticated attacker can use an owned conversation in the URL while naming a victim conversation in the body, causing an attacker-controlled message to be stored in the victim conversation. This issue is fixed in version 0.8.6. |
| CVE-2026-54023 | Medium | 2026-11-20 | Not applicable | No public exploit | LibreChat is an enhanced ChatGPT clone that supports multiple AI providers. From 0.8.5 until 0.8.6, GET /api/user in api/server/controllers/UserController.js getUserController serializes federatedTokens and tokenset to the authenticated browser because an earlier response-sanitization fix was not applied to this endpoint. Browser-context script can obtain OAuth access, ID, and refresh credentials that were intended to remain server-side. This issue is fixed in version 0.8.6. |
| CVE-2026-44908 | Medium | 2026-11-20 | Not applicable | No public exploit | Improper following of a certificate’s chain of trust in the remote-attestation verifier used to establish trust with Meta AI's confidential computing backend in WhatsApp for iOS before 2.26.24.72, WhatsApp Business for iOS and WhatsApp for Mac Desktop before v2.26.23.72, WhatsApp for Android before v2.26.23.74, and WhatsApp Business for Android before v2.26.23.70 could have allowed someone positioned within Meta infrastructure to bypass the hardware root of trust by presenting clients with a VCEK certificate signed by a key other than the hardware manufacturer’s. Any such VCEK Cert would first have to be published to the third-party binary transparency ledger used by WhatsApp Private Processing. The ledger shows that every published VCEK certificate chains to the hardware root of trust, confirming no exploitation. |
| CVE-2026-24205 | Medium | 2026-11-20 | Not applicable | No public exploit | NVIDIA TensorRT-LLM for any platform contains a vulnerability where a user may cause a race condition by running concurrent requests. A successful exploit of this vulnerability may lead to data tampering. |
| CVE-2026-19777 | Medium | 2026-11-20 | Not applicable | No public exploit | G DATA Total Security AVKProxy Link Following Denial-of-Service Vulnerability. This vulnerability allows local attackers to create a denial-of-service condition on affected installations of G DATA Total Security. An attacker must first obtain the ability to execute low-privileged code on the target system in order to exploit this vulnerability. The specific flaw exists within the G DATA ANTIVIRUS Proxy Service. By creating a symbolic link, an attacker can abuse the service to create a file. An attacker can leverage this vulnerability to create a persistent denial-of-service condition on the system. Was ZDI-CAN-28662. |
| CVE-2026-19030 | Medium | 2026-11-20 | Not applicable | No public exploit | Observable Discrepancy vulnerability in Drupal CAPTCHA for Drupal 7 allows unauthenticated attackers to test login credentials despite CAPTCHA protection. When CAPTCHA is enabled on Drupal login forms, normal form validation can still run after a missing or incorrect CAPTCHA response. This may expose credential validation feedback and allow an attacker to distinguish invalid credentials from valid credentials paired with a failed CAPTCHA. This issue affects CAPTCHA for Drupal 7: from 7.x-1.0 through 7.x-1.7. |
| CVE-2026-16132 | Medium | 2026-11-20 | Not applicable | No public exploit | A stored cross-site scripting (XSS) vulnerability exists in the Icon API: Menu submodule of the Drupal 7 Icon API module. When a menu item is configured with the Title Wrapper option enabled, the value of the Title Wrapper Class field is output directly into an HTML attribute without sanitization or escaping. A user with the "administer menu icons" permission can inject a payload such as `title" onmouseover="alert(document.cookie)` that is rendered verbatim on every page displaying the affected menu item, executing arbitrary JavaScript in the browser of any user who interacts with the menu link. This can lead to session theft, account hijacking, and further attacks against site users. This issue affects Icon API 7.1.0. |
| CVE-2026-15223 | Medium | 2026-11-20 | Not applicable | No public exploit | An improper export control defect in ChatEventDispatchActivity (com.vivo.agent.business.chatmode.activity.ChatEventDispatchActivity) of Vivo Assistant allows a local malicious application to execute voice commands without user interaction via a crafted Intent, which may lead to unauthorized phone calls, SMS sending, access to private data such as location and calendar entries, and modification of system settings |
| CVE-2026-14915 | Medium | 2026-11-20 | Not applicable | No public exploit | Exposure of sensitive system information to an unauthorized control sphere vulnerability in Tegsoft Yönetim ve Bilişim Hizmetleri Tic. Ltd. Şti. Tegsoft İletişim Merkezi allows Query System for Information. This issue affects Tegsoft İletişim Merkezi: before latest-202611161510. |
| CVE-2026-134171 | Medium | 2026-11-20 | Not applicable | No public exploit | Cesanta Mongoose 7.21 before 7.24 contains an out-of-bounds read vulnerability in rx_ip6() checksum validation that allows remote unauthenticated attackers to read past received packets by sending IPv6 extension headers. Attackers can send a single IPv6 TCP or ICMPv6 packet with extension headers, causing tcp6csum_ok() and icmp6csum_ok() to use the full payload length and read adjacent receive-buffer memory. |
| CVE-2026-134168 | Medium | 2026-11-20 | Not applicable | No public exploit | devpi-server before 6.21.0 contains a path traversal vulnerability that allows authenticated users to write files outside the +files storage root by naming a user and index '..'. Attackers can self-register a '..' user, create a '..' index, and upload release files with attacker-chosen filenames and contents into a parent of the server directory. |
| CVE-2026-134166 | Medium | 2026-11-20 | Not applicable | No public exploit | probe-image-size before 7.5.0 contains an unbounded memory buffering vulnerability in parseExif in lib/parse_stream/webp.js that allows remote attackers to exhaust memory via an attacker-declared EXIF chunk length. Attackers can supply a WebP stream or URL to probe(url) or probe(stream) declaring a huge EXIF length after VP8X, causing unbounded buffering and process memory exhaustion. |
| CVE-2026-134165 | Medium | 2026-11-20 | Not applicable | No public exploit | probe-image-size before 7.5.0 contains an unbounded memory allocation vulnerability in lib/parse_stream/avif.js that allows remote attackers to exhaust memory via attacker-declared ftyp or meta box sizes. Attackers can supply an AVIF stream or URL to probe(url) or probe(stream) declaring a huge box length, forcing the parser to buffer all received bytes and causing denial of service. |
| CVE-2026-134163 | Medium | 2026-11-20 | Not applicable | No public exploit | probe-image-size 7.4.0 contains a regular expression denial of service vulnerability in lib/common/svg_header.js that allows remote unauthenticated attackers to block the event loop via crafted SVG doctypes. Attackers can supply SVGs with an unclosed internal subset followed by many ']' characters to trigger quadratic backtracking in probe.sync, probe(stream) or probe(url), causing repeated event loop stalls. |
| CVE-2026-134136 | Medium | 2026-11-20 | Not applicable | No public exploit | OceanWP WordPress Theme before 4.2.9 contains a stored cross-site scripting vulnerability in the oceanwp_edd_terms_list() function in inc/edd/edd-helpers.php, which allows authenticated users with EDD download category creation permissions to inject malicious JavaScript by crafting a category name that breaks out of an unescaped HTML title attribute. Attackers can create a category name containing an event-handler payload that executes in the browser of visitors who interact with the category link on the public EDD archive page, potentially compromising an administrator's session if they view the affected page while authenticated. |
| CVE-2026-134051 | Medium | 2026-11-20 | Not applicable | No public exploit | Open Robotics RViz XYZPCTransformer PointField Out-Of-Bounds Read Denial-of-Service Vulnerability. This vulnerability allows network-adjacent attackers to create a denial-of-service condition on affected installations of Open Robotics RViz. Authentication is not required to exploit this vulnerability. The specific flaw exists within the XYZPCTransformer handling of PointCloud2 messages. The issue results from the lack of proper validation of user-supplied data, which can result in a read past the end of an allocated object. An attacker can leverage this vulnerability to create a denial-of-service condition on the system. Was ZDI-CAN-32642. |
| CVE-2026-134050 | Medium | 2026-11-20 | Not applicable | No public exploit | Open Robotics RViz AxisColorPCTransformer PointField Out-Of-Bounds Read Denial-of-Service Vulnerability. This vulnerability allows network-adjacent attackers to create a denial-of-service condition on affected installations of Open Robotics RViz. Authentication is not required to exploit this vulnerability. The specific flaw exists within the AxisColorPCTransformer handling of PointCloud2 messages. The issue results from the lack of proper validation of user-supplied data, which can result in a read past the end of an allocated object. An attacker can leverage this vulnerability to create a denial-of-service condition on the system. Was ZDI-CAN-32646. |
| CVE-2026-134049 | Medium | 2026-11-20 | Not applicable | No public exploit | Open Robotics RViz RGBF32PCTransformer PointField Out-Of-Bounds Read Denial-of-Service Vulnerability. This vulnerability allows network-adjacent attackers to create a denial-of-service condition on affected installations of Open Robotics RViz. Authentication is not required to exploit this vulnerability. The specific flaw exists within the RGBF32PCTransformer handling of PointCloud2 messages. The issue results from the lack of proper validation of user-supplied data, which can result in a read past the end of an allocated object. An attacker can leverage this vulnerability to create a denial-of-service condition on the system. Was ZDI-CAN-32644. |
| CVE-2026-134048 | Medium | 2026-11-20 | Not applicable | No public exploit | Open Robotics RViz RGB8PCTransformer PointField Out-Of-Bounds Read Denial-of-Service Vulnerability. This vulnerability allows network-adjacent attackers to create a denial-of-service condition on affected installations of Open Robotics RViz. Authentication is not required to exploit this vulnerability. The specific flaw exists within the RGB8PCTransformer handling of PointCloud2 messages. The issue results from the lack of proper validation of user-supplied data, which can result in a read past the end of an allocated object. An attacker can leverage this vulnerability to create a denial-of-service condition on the system. Was ZDI-CAN-32643. |
| CVE-2026-134047 | Medium | 2026-11-20 | Not applicable | No public exploit | Open Robotics RViz IntensityPCTransformer PointField Out-Of-Bounds Read Denial-of-Service Vulnerability. This vulnerability allows network-adjacent attackers to create a denial-of-service condition on affected installations of Open Robotics RViz. Authentication is not required to exploit this vulnerability. The specific flaw exists within the IntensityPCTransformer handling of PointCloud2 messages. The issue results from the lack of proper validation of user-supplied data, which can result in a read past the end of an allocated object. An attacker can leverage this vulnerability to create a denial-of-service condition on the system. Was ZDI-CAN-32645. |
| CVE-2026-134046 | Medium | 2026-11-20 | Not applicable | No public exploit | Open Robotics RViz MapDisplay OccupancyGridUpdate Out-Of-Bounds Read Denial-of-Service Vulnerability. This vulnerability allows network-adjacent attackers to create a denial-of-service condition on affected installations of Open Robotics RViz. Authentication is not required to exploit this vulnerability. The specific flaw exists within the handling of OccupancyGridUpdate messages. The issue results from the lack of proper validation of user-supplied data, which can result in a read past the end of an allocated object. An attacker can leverage this vulnerability to create a denial-of-service condition on the system. Was ZDI-CAN-32122. |
| CVE-2026-134045 | Medium | 2026-11-20 | Not applicable | No public exploit | Open Robotics RViz PointCloud2Display PointField Out-of-Bounds Read Denial-of-Service Vulnerability. This vulnerability allows network-adjacent attackers to create a denial-of-service condition on affected installations of Open Robotics RViz. Authentication is not required to exploit this vulnerability. The specific flaw exists within the handling of PointCloud2 messages. The issue results from the lack of proper validation of user-supplied data, which can result in a read past the end of an allocated object. An attacker can leverage this vulnerability to create a denial-of-service condition on the system. Was ZDI-CAN-32123. |
| CVE-2026-134044 | Medium | 2026-11-20 | Not applicable | No public exploit | Open Robotics ROS 2 sros2 load_policy XML External Entity Processing Information Disclosure Vulnerability. This vulnerability allows remote attackers to disclose sensitive information on affected installations of Open Robotics ROS 2. User interaction is required to exploit this vulnerability in that the target must visit a malicious page or open a malicious file. The specific flaw exists within the parsing of policy XML files. Due to the improper restriction of XML External Entity (XXE) references, a crafted document specifying a URI causes the XML parser to access the URI and embed the contents back into the XML document for further processing. An attacker can leverage this in conjunction with other vulnerabilities to execute arbitrary code in the context of the current user. Was ZDI-CAN-32121. |
| CVE-2026-133937 | Medium | 2026-11-20 | Not applicable | No public exploit | A flaw has been found in highlight highlight.js up to 11.12.0. The affected element is the function hljs.highlight of the file src/languages/livescript.js of the component Language Grammar Handler. Executing a manipulation of the argument text can lead to inefficient regular expression complexity. The attack can be launched remotely. The exploit has been published and may be used. The pull request to fix this issue awaits acceptance. |
| CVE-2026-133878 | Medium | 2026-11-20 | Not applicable | No public exploit | Mattermost versions 11.11.x <= 11.11.1, 11.7.x <= 11.7.14, 11.10.x <= 11.10.4 fail to restrict the broadcast scope of group WebSocket events to sessions permitted to read the group which allows an authenticated user, including guests, to obtain the display name, description, member count and member IDs of non-referenceable groups via the received_group event emitted when a group is created, updated, deleted or restored. Mattermost Advisory ID: MMSA-2026-00731 |
| CVE-2026-133876 | Medium | 2026-11-20 | Not applicable | No public exploit | Kartones flask-calendar before commit bfde72e contains a stored cross-site scripting vulnerability because task_details_for_markup returns unescaped content rendered with the |safe filter. Authenticated shared calendar members can store script in task details that executes for other members, letting attackers read and modify their private calendars. |
| CVE-2026-133875 | Medium | 2026-11-20 | Not applicable | No public exploit | one-hub through 0.14.27 contains a server-side request forgery vulnerability in GetImageFromUrl in common/image/image.go that fetches vision chat image_url values without destination restrictions. Attackers holding a standard non-admin API token can send crafted /v1/chat/completions requests via Claude, OpenRouter or Ollama providers to reach internal services or cloud metadata endpoints. |
| CVE-2026-133873 | Medium | 2026-11-20 | Not applicable | No public exploit | Cloud Mail before 3.6.0 contains a stored cross-site scripting vulnerability in attService.addAtt that allows unauthenticated attackers to store inline attachments with sender-supplied MIME types. Attackers can email SVG or HTML attachments, compute the content-hash URL under /oss/ or /attachments/, and steal session JWTs from localStorage to hijack accounts including administrators. |
| CVE-2026-133871 | Medium | 2026-11-20 | Not applicable | No public exploit | Pydici through 2026.11.1 contains a stored cross-site scripting vulnerability that allows authenticated users to inject script by storing HTML in contact names rendered unescaped in the contact detail mermaid diagram. Attackers with contact edit permission can store malicious markup that Contact.relationData() interpolates and contact_detail.html renders with the safe filter, executing script in victims' sessions. |
| CVE-2026-133855 | Medium | 2026-11-20 | Not applicable | No public exploit | flowintel contains an information disclosure vulnerability in its case history feature. When a user views the history of a case they are authorized to access, the history output includes references to other cases (linked, unlinked, or forked) without verifying whether the requesting user has permission to view those referenced cases. This exposes the identifiers and titles of cases that the user should not be able to discover. The vulnerability affects multiple output paths: - Browser-based case history view - API history endpoint - Plain-text history download - Markdown history download - Report timeline generation An authenticated user with access to any single case can enumerate the existence and identifiers of other cases by reading the history of a case they can access. This bypasses the intended per-case access control model and may reveal the existence of sensitive investigations, their identifiers, and associated metadata to unauthorized users. The raw audit storage is not modified; the issue is solely in the presentation layer failing to filter references to inaccessible cases before returning history data to the user. Affected versions: flowintel prior to 4.1.0. |
| CVE-2026-133853 | Medium | 2026-11-20 | Not applicable | No public exploit | FlowIntel's chatbot component used a single shared MCP (Model Context Protocol) session and process-wide environment variable to pass per-user API credentials to a child process. Because the session and the ambient FLOWINTEL_API_KEY environment variable were shared across all concurrent requests, a user's authenticated identity could be applied to another user's tool invocation, or a stale/ambient credential could be inherited by a new request. Preconditions: - An authenticated user with access to the chatbot interface. - Concurrent or overlapping chatbot tool invocations from different users, or a request made after another user's credentials were written to the process environment. Impact: - A lower-privilege user could read data belonging to a higher-privilege user (e.g., private investigation cases). - A user whose session was invalidated (deleted, password changed, key revoked) could still operate under a previously cached ambient credential. - The parent process environment was mutated with user-specific secrets, creating a persistent credential leak surface. Affected versions: FlowIntel prior to 4.1.0. |
| CVE-2026-133852 | Medium | 2026-11-20 | Not applicable | No public exploit | flowintel is a case management application that enforces an approval workflow for tasks in privileged cases. Tasks in such cases are required to start in a Requested status, meaning they must receive explicit approval before becoming active. A logic flaw in the task-creation path allowed tasks to be initialized with a Created status (bypassing the approval step) in two scenarios: - When a task was created from a template and no acting user was supplied, the privileged-case check was skipped entirely, leaving the task in Created status. - When a case was created from a template, all copied tasks were unconditionally assigned the Created status regardless of the case's privileged designation. An authenticated user with the ability to create tasks from templates or to instantiate cases from templates could therefore place tasks directly into an active state within privileged cases, circumventing the intended approval gate. This may expose restricted case content or permit actions that should require supervisory sign-off. Affected versions: flowintel prior to 4.1.0. |
| CVE-2026-133851 | Medium | 2026-11-20 | Not applicable | No public exploit | FlowIntel's alerting module contained an authorization bypass in the alert access control logic. When an alert was linked to a private case, the system relied solely on connector ownership (user or organization) to determine access, without verifying that the requesting user still held valid membership in the associated private case. This meant that a user who was a connector owner of an alert could continue to read alert details, review history, raw payloads, and action comments, and could perform review-status updates or case-creation actions, even after their access to the linked private case had been revoked or they were never a member of that case. Additionally, a time-of-check to time-of-use (TOCTOU) gap existed: after a review-status update or a competing case-creation claim was committed, the response could disclose private case identifiers or review history to a user whose access had been concurrently removed by another session linking the alert to a private case. Impact: - Unauthorized disclosure of private case data (review comments, raw payloads, action history) through alert endpoints. - Unauthorized modification of alert review status and creation of cases from alerts the user should no longer access. - Leakage of a competing case identifier to an unauthorized user during concurrent case-creation races. Affected versions: FlowIntel prior to 4.1.0. |
| CVE-2026-133850 | Medium | 2026-11-20 | Not applicable | No public exploit | flowintel contains an authorization flaw in its recurring reminder notification subsystem. When a user is subscribed to recurring notifications for a case, the system does not re-verify the recipient's current access rights to that case before generating or displaying the notification. As a result, a user who previously had access to a private case and later had that access revoked can still: - Receive recurring reminder notifications containing case-related content. - View stored reminder notifications in their notification list, including notification counts, even though they no longer have permission to view the underlying case. The vulnerability affects both the notification creation path (recurring reminder generation) and the notification display path (listing and counting stored reminders). The issue arises because the authorization check performed at subscription time is not re-applied at delivery or retrieval time. Security impact is limited to information disclosure: an unauthorized user may observe case titles, reminder messages, and associated metadata that they no longer have permission to access. No integrity or availability impact is indicated by the patch. Affected versions: flowintel prior to 4.1.0. |
| CVE-2026-133846 | Medium | 2026-11-20 | Not applicable | No public exploit | FlowIntel contains an authorization bypass in its task management workflow. When a user reopens a completed task (by toggling its completion state or changing its status to a non-finished value) or creates a new task within a completed privileged case, the parent case is automatically revived (un-completed, status reset to Created, finish date cleared). The system failed to enforce the required Admin or Case Admin privilege check before performing this case revival side effect. A user with basic task-level permissions (e.g., editor, queuer, or queue admin) could trigger the revival of a completed privileged case through the browser interface or the REST API, without holding the elevated role required for case-level operations. This could disrupt case lifecycle management, clear completion metadata, and potentially expose the case to further unauthorized modifications. The vulnerability affects both the browser-based task completion/status-change endpoints and the corresponding API endpoints, as well as the task creation flow. Affected versions: FlowIntel prior to 4.1.0. |
| CVE-2026-133844 | Medium | 2026-11-20 | Not applicable | No public exploit | FlowIntel, a threat intelligence case management platform, contains an information disclosure vulnerability in its case serialization logic. When a case is rendered via the REST API or browser views, all connector instances associated with that case were included in the response regardless of the requesting user's permissions. This means that any authenticated user with read access to a case could observe connector metadata belonging to other users (personal-scope connectors) or other organizations (org-scope connectors), including connector URLs, descriptions, UUIDs, and potentially embedded API credentials. Preconditions: - An authenticated user account with at least read access to the affected case - The case must have connector instances attached that belong to other users or organizations Impact: - Disclosure of personal connector endpoints and credentials of other users - Disclosure of organization-scoped connector metadata across organizational boundaries - Enumeration of connector infrastructure (URLs, UUIDs, descriptions) not intended for the viewer Affected versions: < 4.1.0 |
| CVE-2026-133843 | Medium | 2026-11-20 | Not applicable | No public exploit | flowintel contains a vulnerability in the MISP correlation feature that allows unauthorized disclosure of case identifiers. The application failed to enforce proper access controls when generating correlation lists for MISP attributes. Specifically, the system did not verify whether the requesting user had permission to view the source case or the matching cases returned in the correlation results. This lack of authorization checks allowed authenticated users to potentially discover the existence of cases they do not have access to, leading to information disclosure. The vulnerability affects the following components: - Case correlation logic - API endpoints for retrieving correlation attributes - MISP object handling Affected versions: < 4.1.0 |
| CVE-2026-133841 | Medium | 2026-11-20 | Not applicable | No public exploit | FlowIntel contains an authorization bypass in its case serialization logic that allows an authenticated user to view metadata of private cases they do not have access to. When a case is linked to another case, the JSON representation of the source case included the ID, title, and description of every linked case regardless of the requesting user's permissions. This affected multiple access paths: - Browser views (case detail, case info, search, calendar) - REST API endpoints (get case, list cases, search, title lookup, completed/not-completed lists) - Internal tooling and connector module context building An attacker with valid credentials and access to any case that contains a link to a private case could enumerate the private case's identifier, title, and description without holding a membership or role on that private case. The exposure occurred because the serialization method did not receive or evaluate the acting user's identity before including linked-case data in the response. Security impact: - Disclosure of private case identifiers, titles, and descriptions to unauthorized users - Facilitation of further targeted attacks against the exposed private cases - Bypass of the intended per-case access control model Affected versions: FlowIntel prior to 4.1.0 |
| CVE-2026-133840 | Medium | 2026-11-20 | Not applicable | No public exploit | flowintel contains an authorization bypass in the file-to-note conversion feature for case tasks. When a task is in a restricted workflow state (Requested, Rejected, or Request Review) within a privileged case, the conversion endpoint did not enforce elevated role requirements. Any authenticated user with basic case membership (e.g., editor or queuer) could invoke the conversion action, causing the system to read the attached file contents and create a note on the restricted task. This bypasses the intended workflow restriction that only Admin, Case Admin, or Queue Admin roles may modify tasks in those states. An attacker with a low-privilege account in a privileged case could: - Extract file contents into note format, potentially circumventing file-level access controls - Create unauthorized notes that alter the task's note count, modification timestamp, and case history - Disrupt the review/approval workflow on restricted tasks The vulnerability requires an authenticated user with at least editor-level access to the affected case and a task in one of the restricted states. Affected versions: flowintel < 4.1.0 |
| CVE-2026-133838 | Medium | 2026-11-20 | Not applicable | No public exploit | FlowIntel contains an authorization bypass that allows non-approver case members to modify restricted tasks (those in Requested, Rejected, or Request Review status) through several indirect code paths that lacked the required approver permission check. Affected operations include: - Self-unassignment from a restricted task - Reordering tasks, including indirectly moving restricted sibling tasks - Linking or unlinking MISP objects and attributes to/from restricted tasks - Creating MISP objects or standalone attributes with task IDs referencing restricted tasks - Deleting MISP objects or attributes that are linked to restricted tasks - Executing modules on restricted tasks A user with case member access (but without Admin, Case Admin, or Queue Admin privileges) could exploit these paths to alter the state of tasks that are supposed to be locked pending approval, effectively bypassing the intended approval workflow and potentially manipulating associated MISP data. Affected versions: FlowIntel prior to 4.1.0 |
| CVE-2026-133836 | Medium | 2026-11-20 | Not applicable | No public exploit | A vulnerability has been found in hankcs HanLP up to 2.1.6. The affected element is the function RNNLanguageModel.load_language_model of the file hanlp/layers/embeddings/contextual_string_embedding.py. Such manipulation of the argument model_file leads to deserialization. The attack may be performed from remote. The exploit has been disclosed to the public and may be used. Upgrading to version 2.1.7 is sufficient to fix this issue. The name of the patch is 6962907e37dfa467fa6dee164b5b1d0a904b24db. The affected component should be upgraded. |
| CVE-2026-133835 | Medium | 2026-11-20 | Not applicable | No public exploit | A flaw has been found in hankcs HanLP up to 2.1.6. Impacted is the function uncompress of the file hanlp/utils/io_util.py of the component TAR Archive Extraction Handler. This manipulation causes path traversal. The attack is possible to be carried out remotely. The exploit has been published and may be used. Upgrading to version 2.1.7 is recommended to address this issue. Patch name: feb0f82d12762cf4e687b9926efd90c672857780. You should upgrade the affected component. |
| CVE-2026-133730 | Medium | 2026-11-20 | Not applicable | No public exploit | A flaw was found in dnsmasq, where a memory handling issue exists in its Dynamic Host Configuration Protocol for IPv6 (DHCPv6) relay processing. A remote attacker can exploit this vulnerability by sending a sequence of network packets to a system configured as a DHCPv6 relay. This causes the service to write data to previously freed memory and crash, resulting in a Denial of Service (DoS). |
| CVE-2026-133709 | Medium | 2026-11-20 | Not applicable | No public exploit | C-Blosc2 before 3.4.1 contains a heap-based out-of-bounds read vulnerability in frame_decompress_chunk() because sparse frame chunk cbytes are not checked against the actual chunk file size. Attackers can supply a sparse frame directory with a truncated or crafted chunk file to crash applications or leak adjacent heap memory into decompressed output. |
| CVE-2026-133706 | Medium | 2026-11-20 | Not applicable | Proof of concept only | request-filtering-agent before 3.2.2 contains a server-side request forgery filter bypass because address validation in src/request-filtering-agent.ts treats IPv4-compatible IPv6 addresses in the ::/96 range as public unicast. Attackers controlling a fetched URL can use addresses like ::7f00:1 or ::a9fe:a9fe to reach loopback or link-local services, including cloud metadata endpoints. |
| CVE-2026-133704 | Medium | 2026-11-20 | Not applicable | Proof of concept only | Penpot 2.18.0 before 2.18.2 contains a stored cross-site scripting vulnerability that allows authenticated users with edit rights to bypass the sanitize-svg-element blocklist using namespace-prefixed SVG elements. Attackers can upload SVGs containing elements like x:script or attributes like x:onload, which are served inline as image/svg+xml to run JavaScript in victims' Penpot sessions. |
| CVE-2026-133703 | Medium | 2026-11-20 | Not applicable | Proof of concept only | Open5GS AMF through 2.8.0 contains a resource leak vulnerability that allows authenticated UEs to exhaust the session pool by triggering failed DNN or S-NSSAI validation. Attackers can repeat PDU Session Establishment Requests with different PDU Session Identity values, leaving stale amf_sess contexts that degrade service for other subscribers. |
| CVE-2026-133702 | Medium | 2026-11-20 | Not applicable | No public exploit | free5GC PCF through 1.4.6 contains a resource leak in postAppSessCtxProcedure that leaves PCC rules and QoS data in the SM policy decision after rejected Npcf_PolicyAuthorization create requests. Authenticated AFs knowing an active PDU session's UE address can send repeated failing requests with distinct AF application identifiers to grow PCF memory and pollute policy sent to the SMF. |
| CVE-2026-133696 | Medium | 2026-11-20 | Not applicable | No public exploit | A flaw has been found in continuedev continue up to 2.1.0. This affects the function DataLogger.logDevData of the file core/data/log.ts of the component Dev Data Logging. This manipulation causes information disclosure. The attack can be initiated remotely. The vendor was contacted early about this disclosure but did not respond in any way. |
| CVE-2026-133694 | Medium | 2026-11-20 | Not applicable | No public exploit | A security vulnerability has been detected in continuedev continue 2.0.0/2.1.0. Affected by this vulnerability is the function getSidebarContent of the file extensions/vscode/src/ContinueGUIWebviewViewProvider.ts of the component Mermaid Diagram Renderer. The manipulation leads to os command injection. It is possible to initiate the attack remotely. The vendor was contacted early about this disclosure but did not respond in any way. |
| CVE-2026-133693 | Medium | 2026-11-20 | Not applicable | No public exploit | A weakness has been identified in continuedev continue 2.0.0/2.1.0. Affected is the function validateSearchAndReplaceFilepath of the file gui/src/util/clientTools/editImpl.ts of the component Edit Tool. Executing a manipulation can lead to path traversal. The attack requires local access. The vendor was contacted early about this disclosure but did not respond in any way. |
| CVE-2026-133681 | Medium | 2026-11-20 | Not applicable | No public exploit | A vulnerability was determined in Soar Cloud System Human Resources up to 7.3.2026.0112. This impacts an unknown function of the file /client.aspx. Executing a manipulation can lead to cross site scripting. The attack can be executed remotely. The exploit has been publicly disclosed and may be utilized. The vendor was contacted early about this disclosure but did not respond in any way. |
| CVE-2026-133667 | Medium | 2026-11-20 | Not applicable | No public exploit | flowintel contains multiple related flaws that allow an authenticated attacker to inject arbitrary HTML or script into pages viewed by other users. - Timeline rendering: the timeline library's autolinking feature rewrote already-sanitised case notes and imported MISP attribute values into generated HTML link elements. Crafted text containing quote characters could break out of the generated attribute context, altering the surrounding markup and enabling script execution in the victim's browser. - Search highlighting: the highlight injection logic used a regular expression to locate text between angle brackets and spliced <mark> elements into the resulting string. Because the pattern could match inside HTML attribute values, the rewrite could corrupt or inject content into attributes of already-sanitised markup, again leading to DOM-based cross-site scripting. - Content-Type confusion: three analyser selection endpoints returned session-stored JSON strings with a text/html content type. A browser receiving such a response will parse and render the body as HTML, so any HTML or script embedded in the stored value is executed in the victim's context. Preconditions: the attacker must be an authenticated user with the ability to create or import case content (notes, MISP attributes) containing specially crafted quote or angle-bracket sequences. The victim must subsequently open the affected case timeline, run a search, or request one of the selection endpoints. Impact: successful exploitation yields DOM-based cross-site scripting in the victim's browser session, potentially allowing disclosure of case data, session manipulation, or execution of arbitrary actions under the victim's identity. Affected versions: flowintel prior to 4.1.0. |
| CVE-2026-133661 | Medium | 2026-11-20 | Not applicable | No public exploit | FlowIntel contains multiple authorization and credential management vulnerabilities affecting connector instances, MISP helper tools, and case templates. - Connector instance metadata, connectivity checks, and case listings accepted instance IDs without verifying if the acting user had visibility or ownership rights, allowing unauthorized access to sensitive configuration and data. - Case template serialization and creation processes exposed connector instances attached to templates regardless of the user's permissions, potentially leaking credentials or instance details to users who should not have access. - Editing connector instances allowed changes to URLs or sharing scopes without requiring re-authentication of API keys, enabling the reuse of another user's credentials in a new context or the leakage of shared keys to unauthorized endpoints. - Outbound MISP connector keys were incorrectly accepted for authenticating incoming alerts, bypassing the intended separation between outbound integration credentials and inbound alert source authentication. These issues allow authenticated users to access unauthorized connector data, manipulate credential scopes, and potentially exploit credential confusion between different security contexts. Affected versions: < 4.1.0 |
| CVE-2026-133658 | Medium | 2026-11-20 | Not applicable | No public exploit | Flowintel contains a privilege escalation vulnerability in its Matrix relay (bot) integration. When the relay forwarded API requests on behalf of a represented user, the authorization and permission checks evaluated the relay bot's elevated role while the request handlers acted as the lower-privileged represented user. This mismatch allowed users with restricted roles (e.g., Read Only) to perform write operations, directory lookups, calendar modifications, assignment changes, template creation, and import actions they were not authorized to perform. Additionally, relay authority was determined by matching editable user profile fields (name and email) rather than a protected, non-spoofable identifier. An ordinary user who modified their profile to match the expected bot name could be recognized as the trusted relay, enabling impersonation of the bot and potential access to its elevated privileges. Preconditions: - A valid Flowintel user account with a restricted role (e.g., Read Only) - Access to the Matrix relay integration or the ability to send API requests with the relay's API key and a MATRIX-ID header - For the impersonation vector: ability to edit one's own profile name fields Security impact: - Unauthorized write operations by users with read-only or otherwise restricted permissions - Potential impersonation of the trusted relay bot by ordinary users - Bypass of organization-level access controls and role-based restrictions - Unauthorized modification of cases, tasks, calendar events, templates, and user directory entries Affected versions: Flowintel prior to 4.1.0 |
| CVE-2026-133651 | Medium | 2026-11-20 | Not applicable | No public exploit | FlowIntel contained a Cross-Site Request Forgery (CSRF) vulnerability affecting fourteen web routes that performed destructive or state-changing operations. The affected routes accepted HTTP GET requests in addition to (or instead of) POST, which bypassed the Flask-WTF CSRF token validation that is only enforced on non-GET requests. The vulnerable operations included: - Case deletion and case merging - Organisation removal from a case - Task deletion (case and template) - Case template and note template deletion - Connector, connector instance, and icon deletion - Custom tag deletion and status change - Taxonomy and galaxy status changes An attacker who could induce an authenticated user to follow a crafted link or load a page containing a cross-site navigation (e.g., an image tag or redirect) could trigger any of these destructive actions using the victim's session cookie, without the victim's explicit consent or awareness. No special privileges beyond a valid authenticated session with the appropriate role (editor or admin) were required on the victim's account. Security impact: unauthorized deletion or modification of cases, tasks, templates, connectors, icons, custom tags, taxonomies, and galaxies belonging to the victim or the organisation, resulting in loss of data integrity. Affected versions: FlowIntel prior to 4.1.0 |
| CVE-2026-133630 | Medium | 2026-11-20 | Not applicable | No public exploit | The flowintel Docker build workflow used the GitHub Actions checkout action with default credential persistence, causing the job authentication token to be written into the repository's Git configuration file. This file became part of the Docker build context and could be retained in exported build cache layers, even though Git metadata was stripped from the final runtime image. An attacker with access to the CI/CD build cache or build environment could extract the persisted token and use it to gain unauthorized access to the private source repository. Impact: - Exposure of a valid authentication token to the private repository - Potential unauthorized read access to source code and repository metadata Affected versions: flowintel < 4.1.0 |
| CVE-2026-133629 | Medium | 2026-11-20 | Not applicable | Proof of concept only | A vulnerability was determined in Sunnet CTMS up to 10.11. Impacted is an unknown function of the file /cltcms/api/external-training/ext-course of the component External Training Application Form Feature. This manipulation of the argument memberids causes improper control of resource identifiers. Remote exploitation of the attack is possible. The exploit has been publicly disclosed and may be utilized. The vendor was contacted early about this disclosure but did not respond in any way. |
| CVE-2026-133627 | Medium | 2026-11-20 | Not applicable | No public exploit | flowintel contains a log injection vulnerability in its case history (audit log) subsystem. The case history is stored as a line-delimited file where each line represents a single event attributed to a specific user. User-controlled fields, including user names, event messages, and case titles used during case merging, were written to the history file without sanitizing embedded line separator characters (such as carriage return, line feed, vertical tab, form feed, and Unicode line/paragraph separators). An authenticated user with the ability to create or modify case history entries could embed line separators in these fields. Upon writing, the injected separators would split a single intended entry into multiple lines, each of which the history format interprets as a distinct event. This allows an attacker to forge additional history entries attributed to arbitrary users, corrupting the audit trail and undermining the integrity of case records. The vulnerability affects the case history append and merge operations. No authentication bypass or privilege escalation is required beyond normal case-management access. Affected versions: flowintel prior to 4.1.0. |
| CVE-2026-133625 | Medium | 2026-11-20 | Not applicable | No public exploit | iLBC audio codec crash in 4.6.0 to 4.6.9 and 4.4.0 to 4.4.19 allows denial of service |
| CVE-2026-133624 | Medium | 2026-11-20 | Not applicable | No public exploit | SCTP protocol dissector heap disclosure in 4.6.0 to 4.6.9 and 4.4.0 to 4.4.19 might enable denial of service or code execution |
| CVE-2026-133621 | Medium | 2026-11-20 | Not applicable | No public exploit | A flaw has been found in shader-slang slangpy up to 0.43.1. This vulnerability affects the function tinyexr::DecodePixelData in the library external/include/tinyexr.h of the component EXR Image Loader. Executing a manipulation can lead to out-of-bounds read. The attack may be performed from remote. |
| CVE-2026-133620 | Medium | 2026-11-20 | Not applicable | No public exploit | Flowintel's template repository subsystem did not adequately constrain file access to the configured repository base directory. Two related weaknesses existed: - Repository registration and editing accepted arbitrary local paths without verifying they resolved inside the configured REPOSITORY_BASE_PATH, allowing an authenticated user to point a repository at any readable directory on the host. - Manifest and template file lookups used normalized (non-canonical) path joins and a startswith-based guard that could be bypassed by symbolic links, permitting reads of files outside the intended repository. An authenticated user with the ability to add or edit a template repository could therefore cause the application to read and return the contents of arbitrary files readable by the service account, including configuration files, credentials, or other sensitive data. Affected: Flowintel versions prior to 4.1.0. |
| CVE-2026-133616 | Medium | 2026-11-20 | Not applicable | No public exploit | A vulnerability was found in sonos tract up to 0.23.4. The affected element is the function InferenceRulesOp::rules of the file onnx/src/ops/math/gemm.rs of the component Operand Arity Indexing. The manipulation of the argument inputs/outputs results in out-of-bounds read. The attack can be executed remotely. The exploit has been made public and could be used. Upgrading to version 0.23.5 is sufficient to fix this issue. The patch is identified as 44c6f4a7ef988a5bafe98c8bb465d8a253599395. It is recommended to upgrade the affected component. |
| CVE-2026-133615 | Medium | 2026-11-20 | Not applicable | No public exploit | A vulnerability has been identified in webauthn-kotlin versions 1.1.0 through 1.1.3, where an exported device-credential authentication activity launches attacker-supplied intents without validation, potentially allowing a malicious application on the same Android device to access application-protected data, depending on the host application’s content-provider configuration. |
| CVE-2026-133613 | Medium | 2026-11-20 | Not applicable | No public exploit | Flowintel's community data launcher script passes the admin API key as a command-line argument to a child Python helper process. On Unix-like systems, process arguments are readable by all local users via standard tools such as ps or by reading /proc/<pid>/cmdline, allowing any local user to observe the key while the helper process is running. An attacker with local access to the host can capture the admin API key during the brief window in which the helper process is active. Possession of the key grants full administrative access to the Flowintel REST API, enabling unauthorized creation, modification, or deletion of threat intelligence data, users, roles, and organisations. Preconditions: - The victim must invoke the community data import or deletion command (launch.sh -tdc or -dtdc) with the API key supplied as a positional argument. - The attacker must have a local user account on the same host and be able to observe process listings during the helper execution window. Impact: - Disclosure of the Flowintel admin API key to any local user. - Subsequent full administrative control over the Flowintel instance via its REST API. Affected: <4.1.0 |
| CVE-2026-133612 | Medium | 2026-11-20 | Not applicable | Proof of concept only | A weakness has been identified in SqueezeAILab SqueezeLLM up to a5fd71f353bf569feb7b55737c1c1493e78e8f31. This vulnerability affects the function vecquant4matmul_spmv_hybrid_nuq_perchannel of the file quant_cuda_kernel.cu of the component CUDA Matvec Kernel. Executing a manipulation of the argument full_row_indices can lead to out-of-bounds write. The attack can be launched remotely. The exploit has been made available to the public and could be used for attacks. This product does not use versioning. This is why information about affected and unaffected releases are unavailable. The project was informed of the problem early through an issue report but has not responded yet. |