Incorrect authorization of the import target namespace in GitLab CE/EE lets an authenticated user wielding a fine-grained Personal Access Token that is scoped only to user-boundary import permissions push project imports into group namespaces that fall outside the token's intended scope (CWE-863). Only installations running versions from 18.9 up to 19.2.8, 19.3.x before 19.3.5, and 19.4.x before 19.4.2 are affected, and the issue is limited to integrity (I:L) with no confidentiality or availability impact per CVSS:3.1/AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:L/A:N - a low-severity finding (base score 3.1) that does not warrant emergency prioritization. Publicly available exploit code exists (a HackerOne report and the corresponding GitLab work item are referenced), there is no confirmed actively exploited (CISA KEV) status, and exploitation additionally requires a valid scoped token plus favorable conditions, so the practical exposure is narrow.
An out-of-bounds write in GPAC's MPEG-2 transport stream AIT parser (versions up to and including 26.07.0) allows remote attackers to corrupt memory by supplying a crafted stream whose AIT section declares more application descriptors than the fixed application_descriptors_id array can hold. Exploitation requires user interaction - someone or an automated workflow must feed the malicious stream into GPAC's M2TS/AIT parsing path - but no authentication or privileges are needed (CVSS:4.0 base 5.3; assessed CVSS:3.1 AV:N/AC:L/PR:N/UI:R, C:L/I:L/A:L). Publicly available exploit code exists (a PoC archive is linked from the VulDB entry and the upstream issue), though there is no evidence of confirmed active exploitation (not in CISA KEV); the practical risk is concentrated in services that automatically ingest untrusted media, such as transcoding pipelines and upload processors.
Privilege escalation in code-projects Online Reviewer System 1.0 allows an authenticated low-privilege user to raise their own account privileges by tampering with the usertype_id parameter submitted to the User Profile Update Handler. Exploitation is remote, requires a valid pre-existing account (CVSS PR:L) with network reachability to the admin/manage path, and requires no user interaction or non-default configuration; impact is limited to integrity and availability, with no confidentiality loss (VC:N). Publicly available exploit code exists (a proof-of-concept published on GitHub as an 'Online Reviewer-System-Privilege-Escalation'), and the VulDB-derived CVSS 4.0 base of 5.3 with an E:P exploit maturity flag indicates a functioning, easy-to-reuse attack; there is no CISA KEV confirmation of active exploitation, and the issue remains low-to-moderate and low priority despite the working public exploit.
Loading a malformed or tampered sherpa-onnx offline transducer model pack (versions up to and including 1.13.8) can trigger a buffer overflow in OfflineTransducerGreedySearchDecoder::Decode, because the decoder trusts an unvalidated vocab_size value taken from model metadata to index decoder buffers. The issue is exploited by supplying a model pack whose metadata vocab_size is inconsistent with the joiner's actual output dimension; the attack requires the victim to load that attacker-supplied model, so the CVSS 4.0 vector carries UI:P and our independent assessment rates it CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:L/I:L/A:L with only partial confidentiality/integrity/availability impact. Publicly available exploit code exists via the upstream GitHub issue, but there is no evidence of confirmed active exploitation (not in CISA KEV), and the vendor notes that the bug is not reachable when only official model packs with metadata consistent with the joiner are used, making real-world risk LOW despite the 'remote buffer overflow' framing and the low EPSS-style base score of 2.1.
Deserialization of untrusted data in TensorLayer up to 2.2.4 allows local code execution when a victim loads a malicious HDF5 model file via load_hdf5_graph(), requiring user interaction (UI:R) but no privileges (PR:N). Despite high potential impact (C:H/I:H/A:H per our assessed CVSS 3.1 vector), the official CVSS 4.0 score is only 2.1 (low) because it assumes a network vector, and operational priority is LOW since the flaw is not reachable in unattended network services. A publicly available exploit exists; the vendor has not responded to the issue report, and no patch is available.
Code injection in TinyLLaVA Factory (TinyLLaVA_Factory) allows a remote attacker to execute arbitrary code with the privileges of the process that loads a model configuration, by supplying a crafted llm_model_name_or_path value that is handled by TinyLlavaConfig._load_text_config in tinyllava/model/configuration_tinyllava.py. Exploitation is not fully autonomous: user interaction is required (CVSS 4.0 UI:P / assessed CVSS 3.1 UI:R) because an operator must point the loader at the hostile path or model reference - an untouched default deployment is not affected - while authentication is not required per the vector (PR:N). Publicly available exploit code exists (the CVE reference set includes a public upstream issue/exploit at GitHub issue #212 and is flagged as 'exploit'), but there is no confirmation of active exploitation via CISA KEV, and because the project uses a rolling release with no response to the report, no vendor-patched version can be identified; the independent assessment (CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H, base 8.8) rates confidentiality, integrity and availability impact as High, well above the 2.1 score published by the reporting source.
Improper privilege management in cal.com and its cal.diy fork up to version 6.2.0 lets an authenticated low-privilege user delete credential objects belonging to teams they are not authorized to manage. By calling the credentials.delete tRPC procedure with a teamId outside their scope, an attacker can remove scheduling/calendar integration credentials, causing a limited integrity and availability impact on the targeted team; no data is disclosed and no victim interaction is required. Publicly available exploit code exists (a proof-of-concept gist is referenced), and the vendor did not respond to the disclosure, so no fixed version has been identified; the assessed CVSS 3.1 vector is 6.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:L, making this a low-priority issue that still requires a valid account to abuse.
SQL injection in the itsourcecode Online Cake Ordering System 1.0 lets a remote, authenticated low-privilege user subvert database queries through the ID parameter of /addtransacdetail.php, which the PHP code interpolates into a SQL statement without sanitization, producing low confidentiality, integrity, and availability impact (independent assessment: CVSS 3.1 AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:L). Publicly available exploit code exists and the reported E:P metric reflects proof-of-concept maturity, but the CVE is not in CISA KEV, so active exploitation is not confirmed; the principal limiting factors are the low-privilege authentication requirement and the practically non-existent real-world deployment of this sample PHP application.
Path traversal (CWE-22) in the Document Intake component of Polaris Office App 9.9.19 on Android lets a local, low-privileged actor write files outside the application's designated temporary directory by supplying a crafted _display_name value to com.infraware.service.m.n.a. There is no remote or network vector - exploitation requires local access to the device plus the low-privilege ability to invoke the Document Intake function (CVSS 3.1 AV:L/PR:L/UI:N), and the assessed impact is limited to low integrity and low availability with no confidentiality disclosure. A public exploit has been released (publicly available exploit code exists), but the risk is low: the vendor shipped a fix in version 9.9.22 that adds path validation, and the issue is not listed as confirmed actively exploited (CISA KEV) at time of analysis.
Information disclosure in HBAI-Ltd Toonflow-app 1.1.8 and earlier allows a locally positioned, low-privilege authenticated user to expose sensitive application data via the exportData endpoint, with publicly available exploit code and a proof-of-concept exploit-maturity rating (CVSS 4.0 E:P); there is no confirmed active exploitation (no CISA KEV listing). Because the vector is local-only (AV:L) and requires an existing low-privilege session (PR:L), the confidentiality impact is limited (VC:L) with no integrity or availability effect, so real-world risk is low-to-moderate and mainly relevant to multi-user or shared-desktop hosts still running the legacy 1.1.x branch. The vendor disputes the report's applicability, stating that 2.0.x is a substantial architectural refactor in which the referenced code paths no longer exist, and that upgrading to 2.0.0 (current release 2.0.4) mitigates the issue.
Code injection in ModelTC LightX2V through version 0.5.0 allows a local attacker to execute arbitrary code when a victim loads an attacker-controlled tokenizer/model via the cosmos3 runner's AutoTokenizer.from_pretrained call, by supplying a crafted auto_map value in the model configuration. Exploitation requires the user to load untrusted model content and runs only with the local user's existing privileges; there is no remote network attack surface. Publicly available exploit code exists, the project has not responded or released a patch, and the assessed real-world risk remains low because the attack is local and depends on the victim choosing to load a malicious model.
An out-of-bounds read in the GGUF model parser of Tiiny-AI PowerInfer (gguf_init_from_file in ggml.c) lets a local attacker who can supply a crafted GGUF file trigger the parser to read past the end of a buffer via a manipulated data_offset value, resulting in memory disclosure or a crash rather than code execution. Only local users with low privileges (CVSS PR:L, AV:L, no user interaction) are able to reach the vulnerable code path, and the attacker must control or substitute the model file that PowerInfer loads; the issue is rated only 1.9 under CVSS 4.0, and our independent assessment aligns with a Low/C:L/I:N/A:L profile on CVSS 3.1. Publicly available exploit code exists (the reported issue includes a proof-of-concept), but there is no confirmed active exploitation (not listed in CISA KEV), and no accepted upstream patch - a pull request addressing the flaw is still awaiting acceptance.
Command injection in Toonflow-app 1.1.8 and earlier allows a local, authenticated low-privileged user to execute arbitrary commands by manipulating the 'path' argument of the Open Folder endpoint (src/routes/setting/fileManagement/openFolder.ts), which is passed without sanitization to child_process.exec. Publicly available exploit code exists, but the attack requires local access and the injected commands run at the privileges of the user already running the application, limiting impact. The vendor-released patch is version 2.0.0; the vendor further states that the affected code paths were removed in the architectural refactor leading to the current 2.0.4 release, so up-to-date installations are not exposed.
An out-of-bounds read in Tiiny-AI PowerInfer's CUDA backend lets a local, authenticated user crash or partially disclose memory by supplying a crafted model or matrix input to the FFN GPU-split loader path (create_striped_mat_to_gpu in smallthinker/ggml/src/ggml-cuda/ggml-cuda.cu). Exploitation requires an existing low-privileged local account and a CUDA-capable deployment that actually exercises the GPU-split FFN loading path; there is no remote vector and impact is confined to the local process (CWE-125, CVSS 4.0 base 1.9, assessed 3.1 AV:L/AC:L/PR:L/UI:N/S:U/C:L/I:N/A:L). Publicly available exploit code exists (referenced via the project's GitHub issue tracker), but there is no confirmed active exploitation (no CISA KEV listing), and the overall risk is low. PowerInfer uses a rolling release model, so no affected or fixed version numbers are disclosed, and the proposed fix pull request awaits acceptance.
Deserialization of attacker-influenced checkpoint files in torchdistill through 1.1.5 can lead to arbitrary code execution in the context of the process that invokes load_ckpt, but exploitation requires local access and low privileges and depends on the application loading an untrusted src_ckpt file. Publicly available exploit code exists, and there is no indication of confirmed active exploitation (CISA KEV). Overall priority is low-to-moderate because default workflows that load only trusted, self-generated checkpoints are not exposed.
A missing authorization check in Fortinet FortiSIEM 7.5.0 lets an attacker with an already-authenticated, high-privilege account send specially crafted HTTP requests to invoke functionality they should not be authorized to reach, potentially resulting in unauthorized code or command execution. Exploitation requires an existing PR:H-level account (CVSS:3.1/AV:N/AC:L/PR:H/UI:N/S:U/C:N/I:L/A:L, base 3.8), so the realistic threat is a malicious insider or an attacker who has already compromised a privileged FortiSIEM account rather than an unauthenticated outsider; confidentiality is unaffected and only low-level integrity and availability impact is scored. There is no confirmed active exploitation (not in CISA KEV) and no public exploit identified at time of analysis, keeping this a low-priority item, though its command-execution potential warrants prompt patching where privileged-account hygiene is weak.
Header validation bypass in Fastify 5.12.2 through 5.12.5 can permit unauthenticated remote attackers to smuggle a forbidden header past route schema validation when an application uses a propertyNames subschema containing enum or const header names written in canonical casing. The flaw is low impact, with CVSS 3.7, and is limited to request integrity because the affected schemas are uncommon, default Fastify configurations are not vulnerable, and no public exploit identified at time of analysis. Upgrading to Fastify 5.12.6 or later resolves the issue by correctly lowercasing the nested header-name restrictions.
Username enumeration in BusyBox httpd lets unauthenticated remote attackers distinguish valid from invalid account names when the daemon is built with non-default compile options (for example, without shadow password support) or operates with uncommon password-file states and is configured to use system password authentication. The weakness is an observable discrepancy (CWE-203) in how authentication failures are reported or timed for existing versus non-existing users; default BusyBox builds with shadow password support are explicitly not affected, so exposure is narrow and the impact is confined to information disclosure with no credential compromise, data modification, or denial of service. The assessed CVSS 3.1 base score is 3.7 (AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:N), reflecting high attack complexity and the need to reliably observe failure behavior; no public exploit identified at time of analysis and no evidence of confirmed active exploitation (CISA KEV).
Indico versions 3.3.3 through 3.3.13 mis-handle authorization in the legacy contribution-export path, allowing an authenticated low-privileged user to cause speaker and author email addresses plus contribution access-control entries to be written into an API cache entry that can subsequently be retrieved with fewer privileges. Exploiting this requires a valid low-privileged session and successful social engineering of a privileged user into triggering a crafted legacy contribution export, so it is not remotely exploitable without authentication and not a full privilege escalation. There is a vendor-released patch in Indico 3.3.14, and no public exploit code or confirmed active exploitation has been identified at time of analysis; risk is genuinely low.
Server-side request forgery in GitLab CE/EE (all versions from 10.3 before 19.2.8, 19.3 before 19.3.5, and 19.4 before 19.4.2) allows an authenticated user with repository import rights to reach internal network resources by winning a time-of-check to time-of-use race condition in the URL validation performed during import operations. The base score is low (CVSS 3.1 3.5, AV:N/AC:H/PR:L/UI:N/S:C/C:L/I:N/A:N): the attacker must repeatedly race the validation and fetch windows, so the flaw does not trigger reliably and yields only limited confidentiality impact with no integrity or availability effect. There is no public exploit identified at time of analysis and no confirmed active exploitation; the primary risk is limited internal-network reconnaissance or access to internal services from a GitLab instance that can route to them.
Low-privileged attackers positioned on a network adjacent to an SAP Cloud Connector can extract limited information about internal systems by sending specially crafted requests that slip past a flawed authorization check. Exploitation carries a dual prerequisite - adjacency to the connector (AV:A, so not reachable from the public internet) plus valid low-privilege credentials (PR:L) - and the resulting impact is confined to low confidentiality loss with no effect on integrity or availability, which is why this is assessed as a genuinely low-priority issue despite being a real vulnerability at CVSS 3.5. No public exploit code exists and no active exploitation has been confirmed (not in CISA KEV); no EPSS data was supplied with this report.
Out-of-bounds memory reads in the openUBMC device-management process can be induced by a malicious or compromised device on the internal management bus that returns an undersized PLDM-over-MCTP or NCSI-over-MCTP response. Only environments where an attacker already controls a PLDM endpoint (hitting the base, non-vendor PLDM control class) or a Huawei NIC speaking NCSI-over-MCTP are affected; the bug is not reachable over the network or from a normal host OS, since the CVSS 3.1 vector requires physical/bus access (AV:P, base score 3.5) but no authentication (PR:N). An undersized length wraps an unsigned subtraction, causing the parser to read well past the response buffer, yielding limited information disclosure and a low availability impact; no public exploit code has been identified at time of analysis, and the issue is not in CISA KEV.
A local attacker can cause a denial of service in device-mapper-persistent-data utilities by supplying crafted device-mapper metadata containing an excessively large block index, which triggers an integer overflow in the synchronous block reading logic and results in a panic that terminates the tool. The attack requires the attacker to place malicious metadata on the host and a victim to invoke the affected utility against it (UI:R), yielding only a low-impact availability loss (CVSS 3.3). No public exploit code or confirmed active exploitation has been identified at time of analysis.
Cross-origin information leakage in Google Chrome's Media component affects desktop builds prior to 155.0.8059.79, allowing an unauthenticated remote attacker to observe a discrepancy (CWE-203) and infer small amounts of data belonging to another origin. Exploitation requires the victim to be lured into loading a crafted HTML page in a vulnerable browser, and the HIGH attack complexity / LOW confidentiality impact combination means only low-volume disclosure is possible with no integrity or availability effect. No confirmed active exploitation and no public exploit code were identified at time of analysis, and EPSS is very low at 0.17% (6th percentile); a vendor-released fix is available in Chrome 155.0.8059.79.
H3C HDM Family-A G3/G5 BMCs allow an authenticated Redfish account with the 'firmware-update' privilege (or administrator) to replace the RSA trust anchor used for firmware signature verification, enabling a malicious firmware image to pass validation. This affects builds carrying the Redfish/DRE update tree, observed on G5 HDM 6.14.01, while HDM2 uses a different update path and is not affected; no public exploit has been identified at time of analysis. The independent assessment rates this as a low immediate-risk, defense-in-depth trust-model flaw given the high privilege requirement and the need to craft a signed malicious image, though successful exploitation would yield high confidentiality, integrity, and availability impact.
An authenticated user holding maintainer-role permissions on GitLab CE/EE can remove certain package registry metadata that package protection rules are supposed to reserve for higher-privileged roles, due to improper authorization checks in the package registry. The issue affects all versions from 17.6 before 19.2.8, 19.3 before 19.3.5, and 19.4 before 19.4.2, and manifests only on instances where package protection rules are actually configured; the assessed impact is integrity-only (CVSS 3.1 base score 2.7), with no confidentiality or availability loss, and there is no public exploit identified at time of analysis. Vendor-released patches are available in GitLab 19.2.8, 19.3.5, and 19.4.2.
Crafted ISO-2022 encoded text passed to the m17n-lib decoder can trigger an uninitialized stack memory read, potentially leaking small amounts of process memory if the application then returns or displays the decoded output to the attacker. The issue is rated low severity (CVSS 2.5, AV:L/AC:H/UI:R, unauthenticated per vector PR:N) and affects multilanguage library builds shipped across Red Hat Enterprise Linux 6 through 10, though any application embedding m17n-lib's ISO-2022 handling is in scope. No public exploit identified at time of analysis; risk is gated by the fact that the deploying application must both decode attacker-supplied ISO-2022 input and reflect the decoded result back, which is uncommon in typical installations.
Code injection in Linketic CityGaussian through version 2.0 allows a local, low-privileged attacker who can influence the cfg_args value processed by parse_cfg_args in internal/utils/common.py to execute code in a victim's context once that operator loads the tampered configuration. There is no remote attack surface and no public exploit code identified at time of analysis, and the maintainers have not responded to the reported issue (#206), so no vendor patch is available; our assessment rates this as a low-priority, low-real-risk flaw because exploitation requires local access plus passive user interaction and is confined to low confidentiality, integrity, and availability impact.
Code injection in RAE, through its Config-Driven Model Instantiation component, allows a local attacker with low-level privileges to cause arbitrary Python object resolution and potentially code execution when a user loads or runs a crafted model configuration. Exploitation is constrained to local access with required user interaction and only low confidentiality, integrity, and availability impact, so despite the RCE and code-injection tags this is assessed as a low-priority, low-real-risk issue. No public exploit code has been identified at time of analysis, and no vendor-released patch has been identified.
Server-side request forgery in Kavita 0.9.0 through 0.9.1.4 allows an authenticated user with permission to use the CBL import feature to submit a URL to POST /api/Cbl/upload-cbl-file and, by winning a DNS-rebinding race or following an HTTPS redirect, bypass the SSRF blocklist so the server probes internal ports or services blindly. Exploitation requires authentication and a time-of-check/time-of-use precondition, is limited to low confidentiality impact, and no public exploit identified at time of analysis; KEV and EPSS data were not provided. This is a confirmed CWE-918 SSRF but a low-severity, low-priority issue per the independent assessment.
An insecure direct object reference in RuoYi-Cloud-Plus through 6.0.0 lets any authenticated user delete another user's social login binding by calling DELETE /auth/unlock/{socialId}, causing the victim's OAuth token row in sys_social to be removed and effectively blocking their social login. Exploitation requires a valid session on the application (CVSS PR:L) plus knowledge or enumeration of the victim's socialId primary key, so this is a low-severity, low-priority issue rather than a remote pre-auth compromise. No public exploit identified at time of analysis, and the flaw is not listed as confirmed actively exploited (CISA KEV); the assessed CVSS:3.1 vector is AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:L, consistent with limited integrity and availability impact on a single user's binding.
RuoYi-Vue-Plus through 6.0.0 exposes an insecure direct object reference in its social-login unbind endpoint, letting an authenticated user delete another user's OAuth binding (GitHub, WeChat, or Gitee) - along with the stored OAuth tokens - simply by supplying the victim's socialId to DELETE /auth/unlock/{socialId}. The attacker must hold a valid account on the instance (CVSS vector PR:L) and must know or guess the target's binding identifier, and the impact is confined to denial of the victim's social-login path: no confidentiality loss, no credential or password impact, and no account takeover, which is why our independent assessment scores this CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:L (2.3). There is no CISA KEV entry or EPSS signal in the provided data and no confirmed public exploit code, though a public technical writeup of the issue is referenced; a low-priority, hardening-class fix rather than an emergency.
Race condition in the coupon redemption flow of macrozheng mall through version 1.0.3 allows an authenticated low-privilege user to issue parallel requests to /order/generateOrder reusing the same couponId and win a narrow time-of-check/time-of-use window, redeeming a single coupon more than once. The impact is confined to integrity - coupon double-redemption - with no confidentiality or availability effect, no code execution, and no unauthenticated path; the assessed vector is AV:N/AC:H/PR:L/UI:N/S:U/C:N/I:L/A:N, matching the vendor-supplied CVSS 4.0 base score of 2.3. Exploitation is difficult due to the required request timing, no public exploit code has been identified, and there is no confirmed active exploitation (not in CISA KEV), making this a low-priority finding.
Crafted CLI commands fed to the newcli daemon on Fortinet FortiOS - and on the related FortiProxy, FortiSwitch, FortiPAM and FortiSASE platforms that share the code base - trigger a double free that crashes the daemon, giving a local, highly privileged authenticated administrator a low-impact denial of service. Exploitation requires an interactive, already-authenticated administrative CLI session, with no remote, unauthenticated or default-configuration path; no public exploit has been identified at time of analysis. At CVSS 2.3 (CVSS:3.1/AV:L/AC:L/PR:H/UI:N/S:U/C:N/I:N/A:L), this is a low-priority, low-real-risk issue despite being a named CVE.
Information disclosure in libsndfile 1.2.2 and earlier allows remote attackers to trigger use of an uninitialized variable when a crafted WAV file is parsed through wavlike_subchunk_parse in src/wavlike.c, exposing a limited amount of memory contents (CVSS 3.1: 5.3-equivalent vector rated C:L, I:N, A:N; CVSS 4.0 base score 2.1). Affected are applications that link libsndfile and accept attacker-supplied WAV input up to and including version 1.2.2; no authentication is required but the victim must passively process or open the malicious file (UI:P), so there is no zero-interaction remote trigger. Publicly available exploit code exists and was disclosed alongside a project issue report that upstream has not yet responded to, but the issue is not confirmed as actively exploited (not in CISA KEV), and the low-severity, read-only impact keeps overall risk low despite the public PoC.
Server-side request forgery in conductor-oss Conductor 3.31.2 allows an authenticated low-privileged user to force the server to issue outbound HTTP requests to attacker-chosen destinations by supplying a crafted specUrl when defining an API tool, since ListApiToolsTask.start processes that argument without validating or restricting the target address. Successful abuse can expose internal-only services, loopback interfaces, and cloud instance metadata reachable from the Conductor host, with the assessed CVSS:3.1 vector (AV:N/AC:L/PR:L/UI:N/S:N/C:L/I:L/A:N) reflecting limited confidentiality and integrity impact and no availability effect. Publicly available exploit code exists and the flaw is disclosed via VulDB; there is no indication of confirmed active exploitation (not in CISA KEV), and exploitation requires a valid low-privileged account that can create or modify API tool definitions plus a network position where internal targets are actually reachable.
Open redirect in PublicCMS through 5.202608.a lets unauthenticated remote attackers send users to arbitrary external sites by crafting the returnUrl parameter processed by SafeConfigComponent.getSafeUrl in the pay endpoint. Publicly available exploit code exists, and exploitation requires a victim to follow an attacker-supplied link (social engineering), with impact limited to browser redirection. The vendor was contacted but did not respond, and the independent assessment rates this as a low-priority issue despite remote reachability.
Stored cross-site scripting in Rapyd Admin before 9.20.0 allows any authenticated low-privilege user to persist JavaScript through a x-rpd::rich-text Livewire-bound property, which then executes in an administrator's browser session when that record's form is opened in the admin panel. Exploitation requires a valid low-privilege account plus a subsequent administrator interaction, which substantially limits real-world risk, and the vendor scores the flaw at CVSS 4.0 base 2.1 while an independent assessment rates it around CVSS 3.1 6.0 (AV:N/AC:L/PR:L/UI:R/S:C/C:L/I:L/A:N); no public exploit code or confirmed active exploitation was identified at time of analysis. A vendor-released patch exists in 9.20.0 (fix commit acd6ee85), which sanitises rich-text output through HTMLPurifier.
Session hijacking on NETGEAR EX3700 and EX6120 WiFi extenders allows a network-adjacent attacker who already holds valid WiFi/network credentials to predict valid session identifiers and take over an administrator's live web management session (CWE-335). Rated Medium by the vendor with a CVSS 4.0 score of 4.5, and assessed as CVSS 3.1 AV:A/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:N - successful hijack grants full confidentiality and integrity impact over the portal but not availability. The attack is not internet-exploitable, requires a concurrent authenticated admin session and stays limited to these two extender models, and there is no public exploit identified at time of analysis; NETGEAR has published patched firmware for both products.
Server-side request forgery (CWE-918) in the MCP Service of conductor-oss Conductor 3.22.0 lets an authenticated low-privilege user who can invoke the listTools operation supply a crafted mcpServer argument and cause the Conductor server to issue requests to attacker-chosen destinations. The impact is scored low (CVSS 4.0 base 2.1; independent assessment CVSS 3.1 AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:L) because authentication is required, the vulnerable path is only reachable when the MCP integration is enabled and exposed, and there is no scope change to downstream systems. Publicly available exploit code exists for this issue, but there is no confirmation of active exploitation, and no EPSS or CISA KEV data was provided.
Server-side request forgery in Fit2Cloud MeterSphere up to 3.6.9-lts allows an authenticated low-privilege user to make the server issue outbound HTTP requests to an attacker-chosen URL by manipulating the swaggerUrl argument handled by Swagger3ParserApiDefinition.testUrlTimeout in the Swagger3 API-definition import component. Publicly available exploit code exists and the vendor was contacted but did not respond, so no fix is confirmed; the flaw is network-reachable (AV:N) and needs no user interaction, but it requires valid credentials plus access to the API-import feature, which keeps it a moderate priority rather than an urgent one. Real-world impact depends on what the MeterSphere host can reach internally - internal services, admin interfaces, or cloud metadata endpoints - and the assessed impact is limited to low confidentiality exposure (C:L, I:N, A:N).