Cleartext transmission of the Live UDP H.264 video stream in Yoosee GM8135S camera firmware 21.00.01.36 lets anyone positioned on the network path observe the endpoint at /npc/npc and read the unencrypted video feed, with confidentiality impact only (no integrity or availability effect). The camera requires no authentication (PR:N) and no user interaction (UI:N), but exploitation is rated high-complexity because it depends on first obtaining an on-path or shared-network vantage point (same LAN/Wi-Fi segment, upstream router, or other interception position) rather than simply sending a crafted request. A public PoC reference exists; it is not verified as working exploit code, no CISA KEV entry is associated with this CVE, and the vendor was contacted early but did not respond, so no patched firmware version is confirmed.
Reflected cross-site scripting in the 'max' argument of /book_flight.php in itsourcecode Airline Reservation System 1.0 allows attackers to execute script in a victim's browser session. The exploit is not authenticated (PR:N) but requires victim interaction (UI:P), so it is a targeted, social-engineering-dependent issue rather than a self-propagating or server-side compromise; impact is bounded to low integrity only, with no confidentiality or availability effect. A public PoC reference exists but it is not verified as working exploit code, EPSS is low at 0.35% (28th percentile), there is no KEV status, and the vuln.today decision label is 'monitor'.
Type confusion in the Lua sandbox of Sid Meier's Civilization IV 1.0.12.68 lets an attacker who convinces a player to load attacker-supplied content - a crafted mod, scenario/map, or save file - reach the loadstring function and corrupt the game's scripting state, producing limited confidentiality, integrity, and availability impact. No authentication is required (PR:N per both the CVE CVSS 4.0 vector and the assessed CVSS 3.1 vector), but the attacker has no silent trigger: the victim must actively choose to load the malicious file (UI:P/UI:R), and only that specific build is identified as affected. A public proof-of-concept reference exists (a GitHub repository with an Exploit.lua); it is not verified as working exploit code, there is no KEV listing, and the vulnerability-today decision label is 'this_month', indicating a genuine but modest priority rather than a top-tier emergency.
A type-confusion flaw in the Lua Sandbox of Civilization V 1.0.3.144 (function loadstring) lets remote attackers deliver crafted Lua content - a mod, map or scenario - that, once loaded by the victim, corrupts memory and can compromise confidentiality, integrity and availability of the game process. No authentication or privileges are required (CVSS 4.0 vector is PR:N), but the user must actively load the malicious content, which the independent assessment describes as a 'three-click' interaction flow; the reported CVSS 4.0 severity is 5.3 (medium) with Low impacts to each of C/I/A, while the independent CVSS 3.1 assessment rates all three as High. A public PoC reference exists (a VulDB entry and a blog plus GitHub exploit.lua linked from the advisory); it is not verified as working exploit code, and no CISA KEV listing or confirmed active exploitation is present in the supplied data. The vendor was contacted and did not respond, so no fixed version or official mitigation is available.
Server-side request forgery in the self-hosted bookmark manager Benotes (fr0tt benotes) through version 2.8.1 allows an authenticated, low-privileged user to make the server fetch an arbitrary URL they control through the bookmark metadata endpoint. Exploitation requires a valid account with access to the bookmark creation/metadata flow (CVSS PR:L) and only affects confidentiality at a low level, with no integrity, availability, or cross-system scope impact assessed; the overall severity is low (CVSS 4.0 score 2.1). A public PoC reference exists but is not verified as working exploit code, there is no CISA KEV listing, and the maintainer was notified via a GitHub issue but has not responded, so no vendor fix is currently identified.
An authenticated low-privileged user of BKWLD Decoy through version 5.11.1 can raise their own privileges by manipulating the 'role' argument handled by the Authorization Policy logic in Policy.php. Because the attacker must already hold a valid account (CVSS PR:L), anonymous remote users cannot exploit it, but no user interaction is needed and the attack path is simple. This is a medium-severity integrity-only issue (CVSS 4.0 base 5.3; assessed CVSS 3.1 5.3, C:N/I:L/A:N) that sits below emergency priority; a public PoC reference exists but is not verified as working exploit code, there is no confirmed active exploitation, and the vendor did not respond to the disclosure.
Authenticated price manipulation in Omega Solution Restro POS through 20261026 can corrupt sale-invoice business logic via the productUnitSalePrice argument, letting a low-privileged user alter invoice line values. The CVSS 4.0 base score is 2.1 and the independent CVSS 3.1 vector (AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:N) confirms integrity-only impact with no confidentiality or availability effect; exploitation requires a low-privileged authenticated session on the application. A public PoC reference exists but is not verified as working exploit code, active exploitation is not confirmed in the provided data, and the vuln.today priority is monitor.
Improper authorization in the backend component of Bdtask Garage 24 Garage Management Software 1.0 allows a remote user holding a low-privilege authenticated session to perform actions beyond their intended permissions, with limited confidentiality, integrity, and availability impact. The assessed vector requires valid authentication (PR:L) and shows no scope change, so the flaw is not reachable by fully unauthenticated attackers, and the specific backend feature or endpoint that triggers it is not described in the available sources. A public PoC reference exists but is not verified as working exploit code; the vendor did not respond to the disclosure, no fixed version is provided, and the current priority label is monitor, making this a low-priority issue despite being remotely reachable.
An authenticated low-privilege user of iFlytek astron-agent 1.1.2 and earlier can read knowledge-list data belonging to other spaces by tampering with the space-id argument on the Console Knowledge-List endpoint (FileInfoV2Service.listKnowledgeByPage). The flaw is an insecure direct object reference (CWE-639) with a CVSS 4.0 base score of 2.1 and only confidentiality impact; a public PoC reference exists, but it is not verified as working exploit code, and the vendor did not respond to the report. The available data does not identify a fixed release.
Code injection in AllenAI AllenNLP through 2.10.1 allows an attacker who controls the 'type' value in a configuration to drive unsafe reflective class instantiation through Registrable.resolve_class_name, potentially achieving code execution inside any host application that loads that configuration. Exposure is limited to deployments that pass untrusted FromParams-style configuration into the Registrable machinery and requires the victim to process the malicious config, so the library handling only trusted, local configurations is not affected. A public PoC reference exists (it is not verified as working exploit code), there is no CISA KEV entry, and the maintainer did not respond - the project is end-of-life, so no vendor fix is expected and the current priority is monitoring.
Improper authorization in JIZHICMS's front-end order handler (the Pay function in app/home/c/OrderController.php) allows a low-privileged authenticated user to supply an attacker-chosen orderno and act on orders belonging to other users, affecting releases up to and including 2.5.6. Exploitation requires a valid front-end account (CVSS 3.1 PR:L, AV:N, AC:L, UI:N) and is limited to low integrity impact with no confidentiality or availability effect. A public PoC reference exists; it is not verified as working exploit code, and no CISA KEV listing appears in the supplied intelligence, while the CVSS 4.0 vector carries E:P exploit maturity. The vendor was contacted about the disclosure and did not respond, and priority is set to this month's remediation cycle rather than immediate emergency action.
OS command injection in the Sed Command Filter component of Z.ai Zcode (versions up to and including 3.14.3/3.15.1) lets a remote attacker who already holds a low-privileged authenticated session push input into a shell invocation and obtain limited confidentiality, integrity and availability impact rather than full host takeover. The reporting source is VulDB and the vendor-facing reference scores it 5.3 under CVSS 4.0; 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). Risk is capped by the authentication prerequisite (PR:L in both the reported and independently assessed vectors), and the priority label is this_month rather than act_now.
Unrestricted file upload in GouGuOA through 6.1.2 lets an authenticated low-privilege user bypass upload extension restrictions and place a file of their choosing on the server through the disk module's chunkUpload API (app/disk/controller/Api.php). The CVSS 4.0 base score is 5.3 with PR:L and E:P, while the assessed CVSS 3.1 vector (AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H) reflects high confidentiality, integrity and availability impact once an account is held; no user interaction is needed. A public PoC reference exists in the form of a GitHub-hosted vulnerability report; it is not verified as working exploit code, and there is no KEV listing indicating active exploitation. The vendor was contacted and did not respond, so no fixed version has been published.
SQL injection in itsourcecode Online Cake Ordering System 1.0 lets a low-privilege authenticated remote user manipulate the ID argument of /deletetransac.php to alter database queries, with low confidentiality, integrity, and availability impact. A public PoC reference exists; it is not verified as working exploit code, and no active exploitation is confirmed. The issue is low-to-moderate and non-urgent, requiring a valid session and affecting a niche application with minimal real-world deployment.
Unrestricted file upload in SourceCodester Online Food Ordering System 1.0 allows an attacker holding a valid administrator session to upload arbitrary files through the 'image' argument of admin/add_food.php, potentially placing executable PHP content on the server. The precondition is significant: the CVSS vector carries PR:H, so a valid admin session on the admin interface is mandatory, which narrows realistic attackers to those who already hold or have compromised administrative credentials. Risk signals converge on low real-world priority - EPSS is 0.29% (22nd percentile), the vuln.today decision label is monitor, no KEV entry is present, and a public PoC reference exists but is not verified as working exploit code.
Stored cross-site scripting in Argilla 2.0 through 2.8.0 allows a user who can create or edit a dataset to inject script into the dataset Name field, which then executes in the browser of another user who opens that dataset's Settings page (SettingsInfo.vue). Because the injection point is the dataset name, an attacker needs an authenticated account with dataset-creation privileges (PR:L) and the victim must actually load the affected settings view (UI:P), which keeps the practical impact narrow; the CVSS 4.0 score is 2.0 and EPSS is only 0.25% (17th percentile), so the vulnerability is tracked as monitor priority. No vendor patch exists at time of analysis - Argilla was notified through GitHub issue #5875 and has not responded - and a public PoC reference exists; it is not verified as working exploit code.
Cross-site scripting in the patient registration workflow of subhajitkhan's Online Clinic Management System (patientregistration.php, pat_name parameter) lets an authenticated user inject markup or script that executes in another user's browser when the affected page is rendered, yielding low integrity impact with no confidentiality or availability loss per the assessed CVSS:3.1/AV:N/AC:L/PR:L/UI:R/S:C/C:N/I:L/A:N vector. The project does not use versioning, so all snapshots through commit e9ee77a8827a1446220fa07ee693dc4d9a29a578 are treated as affected and no fixed release has been published; the maintainer was notified through GitHub issue #12 and has not responded. Risk is low in practice (EPSS 0.25%, 17th percentile, priority label 'monitor', no CISA KEV listing and no confirmed active exploitation), and the publicly referenced proof-of-concept is an issue-tracker reference rather than verified working exploit code.
Stored cross-site scripting in FFXIV Teamcraft up to and including 11.4.31 lets comment content be rendered as trusted HTML in the desktop/Electron renderer, so an attacker who can post comments can execute script in the context of another user who later views them. Exploitation requires a low-privileged authenticated account to plant the payload and a victim to open the affected comment view, which keeps realistic impact at medium severity (CVSS 4.0 base 5.1, CWE-79); the network vector alone overstates the practical risk. No public working exploit code is confirmed - only a public proof-of-concept reference exists - and the issue is not listed as actively exploited; upgrading to 11.4.32 is the recommended fix and the vendor patch commit is public.
Cross-site scripting in Bdtask Garage 24 Garage Management Software 1.0 lets an authenticated user inject script through the first_name or last_name fields handled by /edit_profile, executing in a victim's browser when the affected profile view is loaded. Exploitation requires an authenticated account (PR:L) and user interaction (UI:P); a public PoC reference exists but is not verified as working exploit code, and no CISA KEV entry or vendor-released patch was identified. This is a low-to-moderate priority issue that does not warrant emergency handling.
An out-of-bounds read in libimobiledevice's AFC file handler can disclose a small amount of adjacent memory to a hostile or spoofed AFC peer. Applications linked against libimobiledevice 1.0 through 1.4.0 that call afc_file_open() are affected; the CVSS:4.0 vector rates confidentiality impact as low with no integrity or availability effect, and requires low privileges plus passive user interaction. A public write-up reference exists but is not verified as working exploit code, no CISA KEV entry or other active-exploitation signal is present, and priority is assessed as moderate and non-urgent.
Stored cross-site scripting in Argilla's annotation text field component (argilla-frontend/components/features/annotation/container/fields/text-field/TextField.vue) allows a user who can write annotation text to persist markup that executes in another user's browser session when the poisoned record is viewed. The assessed vector requires an authenticated principal with privileges to submit or edit annotation text plus interaction from a second viewer, with impact limited to low integrity and no confidentiality or availability loss; all Argilla branches from 2.0 through 2.8.0 are listed as affected. A public proof-of-concept reference exists but is not verified as working exploit code, EPSS is 0.21% (11th percentile), and the vulnerability is not in CISA KEV - priority guidance is to monitor rather than act urgently.
Stored cross-site scripting in Argilla 2.0 through 2.8.0 allows a privileged, authenticated user to inject a crafted value into the 'template' argument of a custom annotation field, which is rendered by argilla-frontend/components/features/annotation/container/fields/sandbox/Sandbox.vue and executes in the browser of a second user who later views that field. The assessed CVSS:3.1 vector (AV:N/AC:L/PR:H/UI:R/S:C/C:L/I:L/A:N) confines the outcome to limited integrity impact in the victim's browser session, with no confidentiality or availability impact, and the high-privilege prerequisite plus the need for a victim to open the crafted annotation are the main limiting factors. This is a low-priority, low-risk issue: EPSS is 0.21% (11th percentile), there is no CISA KEV entry, the priority decision is 'monitor', and a public PoC reference exists but is not verified as working exploit code. No vendor fix has been identified; the project was notified via issue #5876 and has not responded.
Missing authentication on the local HTTP service in 777genius agent-teams-ai through 2.17.10 allows a locally positioned attacker with existing low privileges to access service endpoints without credentials, yielding low-impact confidentiality disclosure. The service is started by doStart in HttpServer.ts and is reachable only with local access; remote exploitation is not possible per the assessed vector, and integrity and availability impacts are absent. A public PoC reference exists but is not verified as working exploit code, and no vendor patch or response is identified; priority is monitor.
Cross-site scripting in the Virtual Server page of the Edimax BR-6208AC web management interface (firmware 1.32) lets an attacker who already holds an authenticated administrator session inject script that executes when a victim subsequently loads the manipulated page. Impact is confined to low integrity in the victim's browser, with no confidentiality or availability effect, and the CVSS 4.0 base score is 1.9 (independent assessment: CVSS 3.1 AV:N/AC:L/PR:H/UI:R/S:U/C:N/I:L/A:N). The vendor was contacted and did not respond, no vendor-released fix is identified, and a public PoC reference exists; it is not verified as working exploit code, and priority signals rate this as monitor-level.
Stored cross-site scripting in gougucms affects releases up to and including 4.08.18, specifically the Add function in app/admin/controller/Level.php where the Title argument of a User Level record is not sanitized before being stored and later rendered. Exploitation requires authenticated access at administrator privilege level (PR:H) to the back-office panel, plus a second user viewing the affected admin page for the script to run (UI:P), so this is not anonymously reachable despite the network vector; impact is confined to the admin interface. A public PoC reference exists (GitHub write-up linked from the VulDB entry); it is not verified as working exploit code, there is no CISA KEV listing, and the vendor did not respond to the disclosure, so no fixed release is identified.
Improper initialization in the WiFi Configuration component of the Yoosee GM8135S Camera (firmware 21.00.01.36) lets an attacker positioned on the same local or WiFi network manipulate the wpa_supplicantX.conf file, with the impact assessed as limited to integrity (no confidentiality or availability loss). Exploitation is rated as high complexity and difficult by both the CVSS vectors and the vendor, no authentication or user interaction is required, and the mechanism that actually triggers the improper initialization is not described in the available sources. A public writeup reference exists (not verified as working exploit code), the EPSS-style signal and vuln.today decision label are both low ('monitor'), and this is a low-priority issue on every available signal; the vendor was contacted but did not respond, and no fixed version is listed.
Routinator writes a newly retrieved trust anchor (TA) certificate fetched over HTTPS into its cache even when that certificate fails validation, so a failed update can overwrite the previously cached, known-good TA and leave the validator working from bad trust anchor material. Only deployments whose TAL points at an https:// TA URI are exposed, and the attacker must already hold the privileged position over that retrieval channel implied by PR:H - impact is limited to partial integrity and availability loss for RPKI validation (I:L/A:L, CVSS 3.8, C:N). No public exploit code and no confirmed active exploitation were identified at time of analysis, and the priority signal is monitor.
Mattermost 11.7.x through 11.7.10 allows an existing team admin to assign built-in team-member roles through the update-team-member-roles REST API beyond what the configured permission scheme permits, giving the member elevated team permissions. The flaw is an incorrect-authorization issue (CWE-863) with a CVSS score of 3.8; exploitation requires authenticated team-admin access (PR:H) and is realistically an insider or compromised-admin risk. No public exploit has been identified at time of analysis, and the independent assessment and vuln.today priority label rate this as monitor/low priority rather than a widespread threat.
Remote attackers can cause a limited integrity impact on WordPress sites running WooCommerce Photo Reviews through version 1.2.31 by exploiting a race condition, though the high-complexity timing window makes practical exploitation difficult. The CVSS 3.1 vector indicates network reachability with no authentication or user interaction required, but only integrity is affected and confidentiality and availability are unchanged. The priority label is monitor, and no public exploit identified at time of analysis; the issue does not warrant emergency action.
Cross-session AES key confusion in the Zephyr RTOS Espressif ESP32 hardware crypto driver (drivers/crypto/crypto_esp32_aes.c, shipped in Zephyr 4.3.0 through 4.4.2) causes cipher operations to run under another session's key and direction instead of the key the caller installed. The driver advertises up to four independent cipher sessions, but the ESP32 AES peripheral has a single chip-global key/mode register set that aes_begin_session() programmed once, outside the aes_lock mutex, so a second session opened with a different key silently redirects every subsequent ECB, CBC or CTR operation from the first session, and the threaded pattern of begin-then-operate on separate sessions also races. Impact is cryptographic rather than memory corruption: plaintext the application believed protected under its own key may be encrypted or decrypted under a key the attacker knows or supplied, and encrypt/decrypt direction can be inverted; unauthenticated remote exploitation is constrained to custom applications on ESP32 hardware that hold two or more concurrent sessions with different keys, no stock Zephyr subsystem does so, and no public exploit has been identified at time of analysis (vendor fix merged, projected release 4.5.0).
Predictable IPv6 interface identifiers in Zephyr RTOS 3.7.0 through 4.4.2 arise because the stack seeds the long-lived secret key behind RFC 7217 stable IIDs (gen_stable_iid() in subsys/net/ip/ipv6.c) and RFC 8981 temporary IIDs (gen_temporary_iid() in subsys/net/ip/ipv6_pe.c) with sys_rand_get(), an API documented as not cryptographically secure. On boards whose entropy driver selects CONFIG_ENTROPY_NEEDS_PRNG, that API resolves to the shared, invertible xoshiro128++ generator, and on entropy-device backends it silently degrades to k_cycle_get_32() samples on driver failure, so an attacker who passively observes on-wire values drawn from the same stream (ephemeral source ports, DNS query IDs, CoAP message IDs and tokens, DHCPv6 transaction IDs, WebSocket masking keys) can reconstruct the key and then compute a device's stable address for any prefix or link its temporary addresses across networks. Exploitation requires the default-off experimental options CONFIG_NET_IPV6_IID_STABLE or CONFIG_NET_IPV6_PE plus the affected PRNG-backed platform class, and no public exploit has been identified at time of analysis; impact is limited to address predictability and tracking (confidentiality of address-lifetime privacy only), with no memory-safety, integrity or availability effect. No public exploit identified at time of analysis, and the item is not tracked as confirmed actively exploited (CISA KEV).
Authenticated channel members on Mattermost 11.7.x through 11.7.10 can abuse the scheduled-post creation and publication paths to publish messages that render as trusted system events because reserved system post types are not validated. The attacker must already have an account and membership in the target channel, and the deception depends on other users viewing and trusting the forged system-style message; impact is limited to message integrity, with no confidentiality or availability effect. The CVSS vector includes PR:L and UI:R, and the score is 3.5; no public exploit was identified at time of analysis, and the priority label is monitor.
Internal-connection filtering is not applied to the OpenID Connect discovery fetch in Mattermost 11.7.x through 11.7.10, so a System Administrator who controls the configured discovery endpoint URL can learn whether internal addresses and ports are reachable from the server, with the fetch itself triggered by an unauthenticated login request. The assigned CVSS 3.1 score is 2.7 (LOW) with PR:H, reflecting that the actor must already hold System Administrator privileges and that impact is limited to low-confidentiality reachability inference with no integrity or availability effect. No public exploit has been identified at time of analysis, and the reported decision signal is to monitor; the primary remediation is the Mattermost release addressing advisory MMSA-2026-00774.
Reflected cross-site scripting in Asterisk's phone-provisioning HTTP error pages (res_phoneprov) lets an unauthenticated attacker who can reach the web server reflect attacker-controlled text into a response and, if an authorized AMI user clicks the crafted link, read that user's AMI session cookie and impersonate them over the HTTP manager interface. Fixed in Asterisk 20.20.1, 21.12.3, 22.10.1, 23.4.1 and Certified Asterisk 20.7-cert11 and 22.8-cert3; the CVSS 4.0 score is 2.3 and the vector requires AT:P plus UI:P, so exploitation needs a deliberately configured stack (HTTP server, AMI, AMI-over-HTTP and res_phoneprov all enabled, all off by default) and normally a non-default bind address, since the HTTP server listens on 127.0.0.1 by default. No public exploit code or active exploitation has been identified at time of analysis, and priority is assessed as monitor.
HtmlSanitizer's SanitizeDom(string) API leaves attributes untouched on the <html> and <body> wrapper elements it parses into, so markup such as a second <body onload=alert(1)> tag can place a live event handler on the returned document. The issue affects releases before 9.2.1039 and 10.0 prereleases before 10.0.1037-beta, and only matters where the caller serializes more than dom.Body.ChildNodes - for example dom.ToHtml() or the body/html outer HTML - and where a victim then renders that output. Severity signals are low (CVSS 4.0 score 2.3; independent assessment CVSS 3.1 AV:N/AC:H/PR:N/UI:R/S:C/C:L/I:L/A:N), no public exploit or CISA KEV entry is associated with the flaw at time of analysis, and the vendor has released fixed versions.
boneIO Black firmware before 1.6.0 places no throttle on authentication attempts, so an unauthenticated attacker who can reach the device over an adjacent network can send unlimited login guesses to the web panel endpoint POST /api/login or against SSH, while weak passwords were also permitted by the absence of complexity rules. Compromise is not automatic: it depends on a targeted account using a weak or guessable password and on the attacker sustaining a large number of attempts against a service with no lockout, and the resulting impact is bounded by the privileges of whichever account is cracked. CERT-PL reported the issue and version 1.6.0 introduces per-source and per-account rate limiting on web logins, an 8-character minimum for panel passwords and a three-attempt SSH limit with a shortened login grace time; no public exploit code was identified at time of analysis and the priority signal is monitor.
Authenticated members on MacCMS v10 can redeem one recharge card multiple times by racing concurrent flag=card requests through the user/buy endpoint, inflating their points balance to purchase VIP upgrades and paid content. The defect is a race condition in Card::useData() where the non-locking card read can complete before the card is marked used, but success is timing- and concurrency-dependent and requires a valid member account plus at least one legitimate unredeemed card. A public GitHub issue reference exists but is not verified as working exploit code, and the priority label is monitor.
Open redirect in tornado.web.RedirectHandler (Tornado prior to 6.6) lets crafted request paths shift the scheme or host of the redirect target, so a link on the application's own domain can send victims to an attacker-controlled site for phishing. Only applications whose RedirectHandler URL template places a regex capture-group substitution where it can influence the scheme or host are affected - substitutions at the start of the path, immediately after a fixed host, or at the very start of the template - while static templates and substitutions after a fixed host plus a slash are unaffected, and the attacker needs the victim to click the crafted link (CVSS 4.0 base 2.1, assessed 3.1 vector AV:N/AC:L/PR:N/UI:R/S:C/C:N/I:L/A:N). Impact is limited to social-engineering-grade redirection with no code execution or data exposure; EPSS is 0.46% (40th percentile), there is no CISA KEV listing, no public exploit identified at time of analysis, and the vendor has released Tornado 6.6.
Out-of-bounds write in SCP: Containment Breach Multiplayer's Skynet++ virtual machine allows remote attackers to corrupt memory when a victim running versions up to 1.295 joins or interacts with a crafted multiplayer session. The flaw is in the BlitzBanks function, is fixed in version 1.3R, and publicly available exploit code exists; no CISA KEV confirmation is present in the supplied data. The CVSS 4.0 record scores it 2.1 with UI:P, while the independent CVSS 3.1 assessment is 8.8 because successful exploitation can fully compromise confidentiality, integrity, and availability on the client.
Unsafe deserialization (CWE-502) in the Workshop Scenario Loader of Ndemic Creations' Plague Inc Evolved affects releases up to and including 1.24.0.3, allowing an attacker who can supply a crafted scenario to have it deserialized when the victim loads it. No authentication is required per the CVSS vector (PR:N), but exploitation depends on user interaction (UI:P, passive): the victim must open or subscribe to attacker-supplied Workshop content, so the attacker cannot trigger the flaw unprompted. Impact is bounded to Low confidentiality, integrity and availability per both the vendor-style CVSS 4.0 score of 5.3 and the independent CVSS 3.1 assessment (C:L/I:L/A:L); the CVSS 4.0 Exploit Maturity value of Proof-of-Concept indicates a public PoC reference exists (a third-party blog post is cited), though it is not verified as working exploit code, and there is no CISA KEV listing. Upgrading to 1.24.0.4 is sufficient to fix the issue.
Improper access control in the Local API Server component of OpenBB (openbb_platform/core/openbb_core/app/model/api_settings.py) allows remote attackers to reach API behaviour they should not be able to influence, resulting in limited confidentiality disclosure. The issue affects versions up to and including 4.7.2 and 5.0.0, and exploitation requires the Local API Server to be deployed and running plus user interaction - consistent with luring a user into triggering a browser-based or cross-origin request against the locally running server; library or CLI-only users are not exposed. The disclosure states that exploit code has been published and may be used, but there is no KEV confirmation of active in-the-wild exploitation, EPSS scoring is not available in the supplied data, the impact is confined to low confidentiality with no integrity or availability effect, and the vuln.today decision label is 'monitor'; the vendor was contacted and did not respond.
Remote SQL injection in itsourcecode Online Book Store 1.0 through the bookisbn argument of /book.php can let a low-privilege authenticated attacker manipulate database queries, with low confidentiality, integrity, and availability impact. The issue carries CVSS 2.1 (CVSS:4.0) and an EPSS score of 0.25% (16th percentile), and the vuln.today decision label is monitor. SSVC exploitation status is poc and a public PoC reference exists; it is not verified as working exploit code.
Denial of service in Xpdf 4.06 and earlier occurs when the viewer processes an attacker-supplied PDF containing a JPEG 2000 image whose progression order field is malformed, causing an infinite loop and a hung process. The attack is local and file-based and is unauthenticated per the CVSS vector (PR:N), but practical exploitation depends on the victim opening the crafted PDF in Xpdf; impact is limited to availability, with no code execution, data disclosure, or integrity loss. No public exploit has been identified at time of analysis, EPSS is 0.11% (1st percentile), and the vuln.today decision is monitor.
HTTP request splitting in the Erlang hackney HTTP client (2.0.0 through 4.9.0) allows a caller-supplied proxy_auth_fun return value to inject additional header lines into the CONNECT request hackney sends to an HTTP proxy. The issue is only reachable in a non-default setup: hackney must be used over an HTTP CONNECT proxy, a proxy_auth_fun callback must be configured, the proxy must answer with a 407, and the callback must build its returned binary from untrusted data such as the proxy's challenge headers instead of static credentials, so an attacker must control or intercept the proxy response to influence the value. Impact is limited to request integrity against the proxy, no public exploit has been identified at time of analysis, and real-world priority is low; the vendor-released fix is 4.9.0.
Heap-based buffer overflow in libimobiledevice's Web Inspector client (webinspector_receive_with_timeout in src/webinspector.c) affects releases through 1.4.0 when the library parses a crafted or oversized message from a Web Inspector peer. The assessed impact is bounded: no confidentiality loss, low integrity and low availability, with exploitation requiring a low-privileged session (PR:L) and a peer that is attacker-controlled, spoofed, or compromised, and the precise overflow trigger is not described in the available sources. The CNA description states the exploit is now public and a writeup is linked; the CVSS 4.0 vector carries exploit maturity E:P (proof-of-concept), there is no KEV confirmation, and the vuln.today decision label is 'monitor', so real-world priority is low.
Denial of service in Qt's Network module: a null pointer dereference in QNetworkAccessManager lets a malicious or compromised HTTP/2 server crash Qt-based client applications that connect to it, affecting Qt 6.8.0-6.8.9 and 6.9.0-6.11.2. The victim client must initiate the connection and the attacker must control or compromise the server and orchestrate a specific HTTP/2 frame sequence; there is no code execution, data disclosure, or integrity impact. A vendor patch is available, and no public exploit was identified at time of analysis.
Heap-based buffer overflow in libimobiledevice's AFC handling (src/afc.c) affects versions through 1.4.0 and can be triggered when the library parses data from an attacker-controlled AFC peer, such as a malicious iOS device or a rogue AFC service. All impact ratings are Low and the priority decision is monitor; the description states exploit code is publicly available and a third-party writeup is referenced, but no KEV or EPSS signal is supplied. Affected deployments must actively use AFC and accept data from an untrusted peer, and the precise trigger is not described in the available sources.
OS command injection in continuedev Continue 2.0.0 is reachable through MCPManagerSingleton.setConnections in core/config/ConfigHandler.ts when the MCP Server Transport is initialized, allowing an attacker who controls an MCP server definition to run operating system commands in the context of the Continue process. Exploitation is gated by user interaction: a Continue user must add or connect to a hostile MCP server configuration, so a default installation is not remotely reachable without that trust decision. VulDB - the sole source for this entry - states that exploit details have been disclosed publicly and may be used, and that the vendor was contacted but did not respond; active exploitation is not confirmed by CISA KEV and no EPSS signal was supplied. Reported severity is modest (CVSS 4.0: 5.3; our assessment CVSS 3.1: AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H), and the priority label is this_month rather than an emergency.
Missing authorization in the Datafeedr Product Sets plugin for WordPress (versions up to and including 1.3.24) allows any logged-in user to delete saved searches they do not own. The AJAX handler dfrps_ajax_delete_saved_search in functions/ajax.php performs no capability check (CWE-862), so an attacker only needs a low-privilege account on the target site-no administrative role is required-to trigger unauthorized state changes. Impact is limited to low integrity and availability (no confidentiality exposure), reflected in the 2.1 CVSS v4.0 score and the assessed CVSS 3.1 vector of AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:L; the vulnerability is fixed in 1.3.25, the disclosure states public exploit code exists, and the current priority decision is monitor rather than emergency.
Code injection in the NNP Converter utility shipped with Sony Neural Network Libraries (NNabla) through version 1.39.0 allows an attacker who supplies a crafted model file to have Python code executed through eval() when a victim converts that file. The unsafe sink is func_set_nnabla_version_decorate in python/src/nnabla/utils/converter/utils.py, which passes the model's attacker-controlled 'function.type' value into eval(); the CVSS 4.0 vector requires user interaction (UI:P), so nothing happens until a person or an automated pipeline actively loads and converts the malicious model. The CNA reports that exploit code has been released publicly and that the vendor did not respond to disclosure; the library is end-of-life with no fixed version, and the overall priority is 'monitor' because exploitation depends on someone processing untrusted input rather than on a remotely reachable service.