2026-11-03
An infinite loop in the ShapeRecord::loadFromFile parser of phpMyAdmin ShapeFile (versions up to and including 4.0.1) allows remote attackers to stall or hang the importing process by supplying a crafted ESRI .shp file, consuming CPU until the process is killed or times out. No authentication is required, but the crafted shapefile must actually be fed into the SHP import/parse path for the condition to fire, so exposure is limited to deployments that accept externally supplied shapefiles. Impact is confined to availability (VA:L) with no confidentiality or integrity effect; a public PoC reference exists, but it is not verified as working exploit code, and there is no confirmed active exploitation (no CISA KEV listing).
Out-of-bounds stack read in NTPsec's NTS-KE client response parser (nts_client_process_response_core) affects versions before 1.2.4 and lets a malicious NTS-KE server leak client stack memory - including pointers useful for defeating ASLR - which the client then transmits in cleartext as NTS cookies. Per the authoritative assessment, exploitation requires the NTPsec client to connect to an attacker-controlled NTS-KE server, either by configuration or by impersonating a legitimate server with a TLS certificate the client trusts, and NTS-KE's TLS layer blocks passive or certificate-less MITM attempts; impact is limited to low confidentiality with no code execution. No public exploit code is confirmed in the available data and the issue is not in CISA KEV, with the priority decision scored as monitor.
Lightdash 0.2609.0 through 3.136.x contains a cross-project insecure direct object reference in RenameService that lets an authenticated user holding update permission on any single project tamper with dashboards in projects they cannot access. By pairing a projectUuid they control in the URL with a foreign dashboardUuid, an attacker can persist rewritten filters on another team's dashboard (breaking its charts for every viewer) or read that project's field metadata through the /rename/dashboard/fields endpoint. Exploitation requires a valid credential such as a Personal Access Token and prior knowledge of a non-enumerable foreign dashboard UUID, which is why the vendor's advisory and this analysis both treat it as a low-priority, monitor-level issue rather than an emergency; no confirmed public exploit code or KEV entry exists, though the vendor advisory publishes a proof-of-concept walkthrough.
Django's server-side per-page caching layer (UpdateCacheMiddleware and the cache_page() decorator) can generate identical cache keys for two requests that should be cached separately: one that omits a request header named in the response's Vary header, and one that sends that header with an empty value. Because a missing header was substituted with an empty string during key generation, a response cached for one of those requests may be served to the other, potentially disclosing private or per-request data to a different client. Exposure requires that the application has opted into Django's caching middleware or decorator and caches responses that vary on a header clients can omit or blank; the vendor rates the issue low severity, no public exploit has been identified at time of analysis, and fixes are available in Django 6.1.3, 6.0.10, and 5.2.19.
AuditPilot's knowledge-retrieval endpoints let an authenticated user who holds read:knowledge on a single project read knowledge chunks belonging to other projects in the same tenant by supplying a foreign context.project_id to GET /api/knowledge/query or POST /api/research/answer. Builds of the intelligent-audit-system repository prior to commit 4669c93f3a7403a0e242bef14d771beef0b4356b are affected; the exposure is confidentiality-only and stays inside one tenant, and the attacker must know or guess a valid in-tenant project identifier before the bypass returns anything. No public exploit code and no CISA KEV entry were identified at time of analysis, and the low severity score is consistent with the monitor priority label assigned to this CVE.
Parse Server's POST /upgradeToRevocableSession endpoint will mint a fresh session for any valid session token - including one that is already revocable - because it does not enforce the create class-level permission on the _Session class, so a holder of a user's session token can obtain extra sessions with new expiration dates that remain valid after the original session is logged out. The attacker gains no new privileges or data, only persistence for an account they can already access, and only deployments that restrict the _Session create CLP are affected, per the advisory. Vendor fixes are released in 8.6.112 and 9.10.6-alpha.4; no public exploit code was identified at time of analysis and the issue is not listed in CISA KEV.
FreeRDP X11 clients through 3.32.1 can be terminated by a malicious or compromised RDP server that supplies oversized target dimensions or output origin in an RDPGFX MAPSURFACETOSCALEDOUTPUT PDU, because the values are stored verbatim and later reach a checked UINT16 cast or a destination-stride assertion. Impact is availability-only client abort (no confidentiality or integrity effect) and the assessed vector is AV:N/AC:H/PR:N/UI:R with the victim required to initiate a connection to the attacker's server; two documented trigger paths depend on specific build, assert and X visual configurations rather than a stock default build. No active exploitation is confirmed (not in CISA KEV), the priority label is monitor, and a public PoC reference in the vendor advisory is a focused harness rather than verified working exploit code.
FreeRDP clients built with active WinPR assertions can be aborted by a malicious or attacker-controlled RDP server that sends a malformed RDPDR device-list announcement declaring a DeviceDataLength larger than the bytes actually delivered. The crash requires a narrow combination of conditions: device redirection must be enabled so the RDPDR channel is active, the RDPDR logger must be at TRACE level so the diagnostic packet dumper runs, and the client build must carry active WinPR assertions - with INFO logging or assertions disabled the same packet left the client running. Affected releases are 3.27.0 through 3.32.1 (fixed in 3.32.2), the demonstrated impact is availability only - no memory corruption, data disclosure, modification, or code execution - and the vendor advisory publishes a working relay script that reproduced the abort end to end, though there is no indication of active exploitation in the field.
mongorestore in MongoDB Database Tools can redirect restored documents into a different collection when a truncated dump filename causes the tool to URL-decode the original collection name from metadata, with colliding names potentially omitting backed-up documents or, when --drop is used, replacing the unintended destination. The issue affects versions 100.0.2 through before 100.20.1 and requires an authenticated low-privileged database user who can create a crafted collection plus a later operator-initiated directory restore with write access to the destination. The vendor released 100.20.1 as the fix; no public exploit identified at time of analysis.
Password-hash disclosure in the Django admin affects custom user-model deployments: staff users who hold only the 'view' permission on a user model can read the entire stored password hash from the read-only change form when that model is not the configured AUTH_USER_MODEL, or when AUTH_USER_MODEL inherits its `password` field from a concrete parent through multi-table inheritance. The flaw is in `django.contrib.admin.utils.display_for_field()`, which only masked the password field when the field's model was exactly `get_user_model()`; the fix broadens that check to any `AbstractBaseUser` subclass and to models that are ancestors of the configured user model. Supported branches 6.1 before 6.1.3 and 6.0 before 6.0.10 are affected, and Django rates the issue severity "low"; there is no CISA KEV listing and no public exploit identified at time of analysis, and the operator decision label is monitor. Exposure requires an attacker who already holds valid staff credentials and admin access (PR:H per the assessed vector), and what is disclosed is a hash rather than plaintext, so offline cracking is still needed to make the information useful.
Heap buffer over-read in the HDF5 library through 2.2.0 lets a crafted HDF5 file push an object-header message body past the end of its decoded chunk, so opening the file through H5Fopen() can terminate the process or disclose a small amount of process memory through a decode callback such as H5O__mtime_new_decode(). The flaw is in H5O__chunk_deserialize(), which validates the declared message size before the remainder of the message header (flags and, for version 2 object headers, the optional creation index) has been consumed, leaving up to four bytes of unchecked slack. Impact is bounded by the assessed CVSS:3.1 vector AV:L/AC:L/PR:N/UI:R/S:U/C:L/I:N/A:L - exploitation is local and requires a victim to open a malicious file, with only low confidentiality and availability consequences and no integrity or code-execution impact; there is no CISA KEV entry, the CVSS 4.0 vector carries an exploit-maturity value of proof-of-concept only, and the priority decision is monitor. The vendor fix is an upstream pull request (HDFGroup/hdf5#6589) that closes issue #6401.
HDF5 through 2.2.0 discloses process heap memory when an application opens and reads a dataset that uses the scaleoffset filter from a crafted file whose filtered chunk record is truncated. The decode path in H5Z__filter_scaleoffset() validates reads against the allocated size of the chunk buffer instead of the number of bytes actually stored, and because H5D__chunk_lock() sizes that buffer to the larger of the stored and unfiltered chunk sizes while filling only the stored bytes, the truncated chunk passes the bounds check and the uninitialized remainder is decompressed and returned as dataset data. The vulnerability is local (CVSS:3.1/AV:L/AC:L/PR:N/UI:R), requires the victim to open and read the crafted file, and is limited to confidentiality impact with no code execution, modification, or crash; an upstream fix exists in PR 6585 but no released patched version is confirmed, and no public exploit has been identified.
Silent message loss in whatstk 0.3.0 through 0.8.x lets a party who supplies one of the chat exports cause messages to vanish from the output of merge_chats() (reached through BaseChat.merge()) without any warning or count of what was removed. The merge keeps whole the chat whose first message is oldest and appends from the other export only messages strictly newer than that chat's last message, so every message of the second export that falls inside the already-covered date range is discarded; a message held by only one export is dropped whenever its date sits inside the range of the other. Exploitation requires the victim to run a merge over attacker-influenced input (local, active user interaction, no privileges needed), there is no code execution or confidentiality impact, the assessed severity is low (CVSS 1.8), no public exploit has been identified at time of analysis, and the vendor has released a fix in 0.9.0.
MongoDB Database Tools' mongofiles utility in versions 100.2.0 through before 100.20.1 can be induced to create a file outside the operator's download working directory when get_regex retrieves a GridFS object after an attacker plants a symlink in that directory. Exploitation requires low-privilege GridFS upload access, local write access to the operator's working directory, and the operator later running the matching retrieval; the demonstrated default behavior creates a new file rather than overwriting one. Impact is limited integrity loss under the operator's filesystem permissions, and no public exploit has been identified at time of analysis.
Local file-write redirection and path disclosure in the nnn terminal file manager (through version 5.3) affect users who enable the NNN_FIFO preview/FIFO channel. An attacker holding an unprivileged local account and write access to /tmp can pre-create the documented, predictable /tmp/nnn.fifo path as a symlink or an attacker-readable regular file before the victim starts nnn, so that writes nnn makes to that path overwrite files the victim can already write to or expose the victim's browsed paths; exploitation requires the FIFO feature to be in use and depends on the victim launching nnn. No public exploit code or confirmed active exploitation was identified at time of analysis, and the assessed severity is low (CVSS 4.0 base 1.0; our CVSS 3.1 assessment AV:L/AC:H/PR:L/UI:R/S:U/C:L/I:L/A:N; vuln.today priority label: monitor).
Terminal escape sequence injection in Canop dysk before 3.8.2 allows a local low-privileged user who can create FUSE mounts to place ESC/OSC sequences in mount-point names or fsname, which dysk then renders unsanitized in its table and CSV output; when another user runs dysk in an interpreting terminal, output can be spoofed or hidden and window titles changed. The issue is local-only with victim interaction, integrity impact only, no code execution or data disclosure, and no public exploit identified at time of analysis (priority: monitor).
Denial of service in Apache HertzBeat 1.7.1 through 1.9.0 allows authenticated users with monitor-template creation privileges to consume excessive collector CPU by supplying XPath expressions through metrics.aliasFields that bypass the validation applied to parseScript. The flaw is a follow-up to an incomplete fix for CVE-2026-24343 and is fixed in Apache HertzBeat 1.9.1. No public exploit code was identified at time of analysis; the issue has low availability impact under CVSS 3.1 (AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:L) and the priority decision is monitor.