Credential-harvesting malware compromised 84 versions of 42 TanStack npm packages on 2026-05-11 via chained GitHub Actions exploitation. Attackers combined pull_request_target misconfiguration, Actions cache poisoning, and OIDC token memory extraction to publish malicious code under the legitimate TanStack identity. Installing any affected version executes a 2.3 MB obfuscated payload that exfiltrates AWS/GCP/Kubernetes credentials, npm tokens, GitHub secrets, SSH keys, and HashiCorp Vault tokens over encrypted Session/Oxen messenger infrastructure. The payload propagates by republishing victim-maintained packages with identical injection. Socket.dev and the TanStack team confirmed the incident via GHSA-g7cv-rxg3-hmpx. No EPSS or CISA KEV data available for this recent supply-chain attack. CVSS 9.6 reflects the cross-scope credential theft impact (S:C/C:H/I:H), though exploitation requires user-initiated package installation (UI:R).
Unauthenticated remote code execution in Dalfox REST API server mode (versions ≤2.12.0) allows network attackers to execute arbitrary OS commands by injecting shell payloads via the `found-action` parameter in POST /scan requests. The server binds to 0.0.0.0:6664 by default with no API key enforcement unless explicitly configured, and deserializes attacker-controlled JSON directly into execution-control options without sanitization. Attackers trivially guarantee exploitation by hosting a reflective XSS endpoint to trigger the injected command. Fixed in version 2.13.0. CVSS 10.0 (AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H). EPSS data not available; no CISA KEV listing at time of analysis. Public exploit code exists (detailed proof-of-concept published in GitHub advisory GHSA-v25v-m36w-jp4h).
Remote code execution in the Windows Netlogon service (across Windows Server 2012 through 2025) allows an unauthenticated attacker to run arbitrary code over the network by triggering a stack-based buffer overflow (CWE-121). With a CVSS 3.1 vector of AV:N/AC:L/PR:N/UI:N and total technical impact, this is a critical, pre-authentication server-side flaw. Microsoft has released patches; publicly available exploit code exists, though EPSS remains low (0.09%) and there is no confirmed active exploitation at time of analysis.
Remote code execution in the Exim mail transfer agent (versions before 4.99.3) built against certain GnuTLS configurations arises from a use-after-free in the BDAT/CHUNKING body-parsing path. An unauthenticated network attacker who sends a TLS close_notify mid-body during a CHUNKING transfer followed by a trailing cleartext byte on the same TCP connection can corrupt the heap and, per the advisory, execute arbitrary code. Publicly available exploit code exists, but there is no CISA KEV listing and EPSS is very low (0.06%, 18th percentile), so this is best treated as a high-impact issue with no confirmed active exploitation yet.
Unauthenticated host file system access in WGDashboard before 4.3.2 lets remote attackers reach protected functionality and read arbitrary server files without logging in. The root cause is a broken authentication gate combined with an unauthenticated file-download endpoint that accepted attacker-controlled file paths. No public exploit identified at time of analysis, but the fixing commit is public and the flaw is trivially reachable, while EPSS remains low at 0.12%.
Path traversal sandbox escape in Pulpy (versions prior to 0.1.1), a cross-platform desktop packager for web apps, lets any packaged web application read and write arbitrary files in the user's home directory via the injected pulpy.fs JavaScript API. The validateFsPath() sandbox relies on an incomplete blocklist, so a malicious or compromised packaged app can exfiltrate secrets such as ~/.ssh/id_rsa, ~/.aws/credentials, and macOS keychains. Publicly available exploit code exists (Exploit-DB 52616), though EPSS remains very low at 0.04%.
Plaintext TOTP secret exposure in sealed-env enterprise mode allows remote unauthenticated attackers to extract operator authentication credentials from base64-decoded JWS tokens. Versions 0.1.0-alpha.1 through 0.1.0-alpha.3 embed literal TOTP secrets in every minted unseal token's JWS payload without encryption, enabling credential harvesting from CI logs, container environments, monitoring tools, and log aggregators. Fixed in version 0.1.0-alpha.4. CVSS 9.1 (Critical) with network vector and no authentication required. No CISA KEV listing or public exploit code identified at time of analysis, but exploitation requires only base64 decoding of observable tokens.
Remote code execution in Microsoft SQL Server (2016 SP3 through 2025) lets an authenticated attacker with low-privilege database access run arbitrary code on the server via external control of a file name or path (CWE-73). The CVSS 3.1 base score is 8.8 with high impact to confidentiality, integrity, and availability; publicly available exploit code exists, though the flaw is not on CISA KEV and EPSS remains very low (0.06%, 20th percentile), indicating minimal observed exploitation to date. Microsoft has released patched builds across all affected branches.
Authenticated remote code execution in Wing FTP Server 8.1.2 lets a domain administrator inject arbitrary Lua code by abusing the session serialization routine, which writes session values into Lua source without escaping closing delimiters. When the poisoned session is later reloaded through loadfile(), the attacker's Lua executes on the server. Publicly available exploit code exists (Exploit-DB 52589, VulnCheck advisory), but there is no evidence of active exploitation and EPSS is very low (0.09%).
Remote denial of service in ASP.NET Core (.NET 8.0, 9.0, and 10.0) allows an unauthenticated attacker to send crafted network requests that trigger a loop with an unreachable exit condition (CWE-835), pinning CPU and rendering the application unresponsive. Publicly available exploit code exists, though there is no public exploit identified as being used in active attacks; EPSS is very low (0.04%, 12th percentile), consistent with the CISA SSVC 'Exploitation: none' assessment. Microsoft has released fixed runtime packages, and the flaw is availability-only with no confidentiality or integrity impact.
Pocket ID OIDC provider fails to validate user authorization state during refresh token exchange, allowing revoked, disabled, or unauthorized users to obtain fresh access tokens indefinitely. Affects all versions prior to 2.6.0. Publicly available exploit code exists via GitHub security advisory GHSA-w6p7-2fxx-4f44. Attack requires low privileges and user interaction (CVSS 8.5) but enables persistent unauthorized access even after administrative revocation actions. Fixed in version 2.6.0.
Local privilege elevation in Microsoft .NET (Windows Desktop/WPF runtime) for versions 8.0, 9.0, and 10.0 allows an attacker to run code at a higher privilege level after a targeted user opens crafted input. The root cause combines improper input validation (CWE-20) with a heap-based buffer overflow (CWE-122), yielding total technical impact per CISA SSVC. Publicly available exploit code exists, but the vulnerability is not in CISA KEV and EPSS is low (0.11%, 30th percentile), indicating no observed widespread exploitation to date.
Local privilege escalation in Microsoft .NET Framework (versions 3.5 through 10.0) and Visual Studio 2017 occurs through heap-based buffer overflow exploitation requiring user interaction with a malicious file. Attackers without initial privileges can achieve high-level code execution and data access by convincing a user to open a specially crafted document or application. Microsoft has released patches across all affected .NET versions per MSRC advisory, indicating this is a vendor-confirmed issue requiring immediate remediation for systems where users process untrusted .NET content.
Stored cross-site scripting (XSS) in Siemens SIMATIC S7-1500 and ET 200SP controller families allows authenticated attackers with high privileges to inject malicious scripts via Technology Object (TO) names when downloading TIA Portal projects. The scripts execute when authorized users access the Motion Control Diagnostics web interface page, enabling session hijacking, credential theft, or privileged actions performed under the victim's context. This affects over 100 product variants across industrial automation controllers, software controllers, and open controllers. No active exploitation confirmed (not in CISA KEV). EPSS score not provided in dataset. CVSS 4.0 score of 9.3 reflects high impact across confidentiality, integrity, and availability in both vulnerable and subsequent systems.
Stored cross-site scripting in Siemens SIMATIC S7-1500 controller family web interface allows authenticated high-privilege attackers to inject malicious code via crafted PLC/station names in TIA project files. When users with appropriate rights later access the communication parameters page, injected scripts execute in their session context with high impact to confidentiality, integrity, and availability across system boundaries (CVSS 9.3, CVSS:4.0 S:H). No public exploit identified at time of analysis, but CVSS vector indicates low attack complexity (AC:L) once attacker gains privileged project upload access.
Command injection in Siemens RUGGEDCOM ROX industrial router series allows high-privileged authenticated remote attackers to execute arbitrary commands with root privileges on the underlying operating system. Affects all MX5000/MX5000RE/RX1400/RX1500/RX1501/RX1510/RX1511/RX1512/RX1524/RX1536/RX5000 models running firmware versions below V2.17.1. The vulnerability exists in the Scheduler functionality of the Web UI due to improper input sanitization (CWE-78). CVSS v4.0 score of 8.9 reflects high impact across confidentiality, integrity, and availability with network attack vector but requires high-privilege authentication. No public exploit identified at time of analysis, and EPSS data not available for this recently published CVE.
Remote code execution in Microsoft Dynamics 365 (on-premises) versions 9.0 and 9.1 (prior to 9.1.44.15) lets an authenticated attacker inject and execute arbitrary code across the network with a scope change beyond the vulnerable component. Rated CVSS 9.9, the flaw yields total confidentiality, integrity, and availability impact, but there is no public exploit identified at time of analysis and EPSS exploitation probability is low (0.07%, 21st percentile). Microsoft has released a fixed build (9.1.44.15).
Privilege escalation in Microsoft Azure Logic Apps allows an authenticated attacker with low-level privileges to elevate their permissions over the network via improper access control (CWE-284). The CVSS 9.9 score is driven by a scope change (S:C), meaning a successful attack breaches the security boundary of the vulnerable component and impacts other resources with full confidentiality, integrity, and availability compromise. No public exploit identified at time of analysis, and EPSS probability is low (0.06%), consistent with CISA SSVC rating exploitation as 'none'.
Remote code execution in Microsoft Windows DNS lets an unauthenticated attacker corrupt heap memory over the network and run arbitrary code on affected Windows 11 (22H3 through 26H1) and Windows Server 2022/2025 systems. The flaw is a CWE-122 heap-based buffer overflow rated CVSS 9.8, reported by Microsoft with a patch available; there is no public exploit identified at time of analysis, and the EPSS score is very low (0.07%), so this is a high-severity but not-yet-exploited issue. CISA SSVC flags it as automatable with total technical impact, meaning successful exploitation fully compromises the host.
Remote unauthenticated attackers can exhaust resources and bypass authentication controls in Siemens SIMATIC CN 4100 versions before V5.0, enabling denial of service conditions and unauthorized actions that compromise system availability and integrity. The vulnerability stems from improper connection validation (CWE-306), allowing network-based exploitation without any user interaction or privileges. Siemens has released V5.0 to address this flaw, documented in security advisory SSA-032379.
Siemens SIMATIC CN 4100 versions prior to V5.0 can be rendered unavailable through TCP SYN flood attacks, allowing remote unauthenticated attackers to exhaust system resources and cause complete service disruption. The CVSS 4.0 score of 8.7 reflects the high availability impact (VA:H) combined with network-accessible attack vector requiring no privileges or user interaction. No active exploitation (CISA KEV) or public exploit code has been identified at time of analysis, though SYN flood techniques are well-documented and trivial to execute.
Denial of service in Siemens industrial networking equipment allows remote unauthenticated attackers to crash affected devices via specially crafted IPv4 packets, requiring manual restart for recovery. This vulnerability affects over 200 Siemens industrial automation products including SCALANCE switches/routers, SIMATIC PLCs, SINAMICS drives, and RUGGEDCOM devices. CVSS 4.0 score of 8.7 reflects high availability impact (VA:H) with network-accessible attack vector requiring low complexity and no privileges (AV:N/AC:L/PR:N). No public exploit code or CISA KEV listing identified at time of analysis, though the straightforward network-based attack and widespread product exposure warrant priority patching for operational technology environments where uptime is critical.
Hardcoded cryptographic keys in Siemens Teamcenter PLM software enable remote attackers to bypass authentication and gain unauthorized access to confidential product lifecycle management data. The vulnerability affects multiple Teamcenter versions (V2312, V2406, V2412, V2506, V2512) and is remotely exploitable without authentication (CVSS:4.0 AV:N/AC:L/PR:N). While no active exploitation or public POC has been identified at time of analysis, the straightforward nature of hardcoded credential extraction (AC:L) combined with the criticality of Teamcenter as an enterprise PLM platform housing intellectual property makes this a high-priority remediation target for manufacturing and engineering organizations.
Stored cross-site scripting (XSS) in Siemens Teamcenter allows authenticated attackers with low privileges to inject malicious JavaScript that executes in other users' browser sessions, enabling session hijacking, credential theft, or unauthorized actions within the product lifecycle management platform. Affects Teamcenter versions V2312 through V2512 with vendor patches released for all branches. CVSS v4.0 scores 8.5 (High) due to network attack vector with low complexity, though exploitation requires user interaction and authenticated access. No public exploit code or CISA KEV listing identified at time of analysis.
Information disclosure in SUBNET Solutions PowerSYSTEM Center (2020, 2024, and 2026) lets an authenticated user with only limited permissions call the REST API device-account-export endpoint to retrieve sensitive data that should be restricted to administrators. The flaw is a broken authorization (CWE-863) check, so a low-privilege operator can escalate their effective data access without any administrative role. There is no public exploit identified at time of analysis, and EPSS is very low (0.02%, 6th percentile), consistent with CISA's SSVC rating of 'Exploitation: none.'
Spoofing vulnerability in Microsoft Azure Entra ID (formerly Azure Active Directory) enables remote unauthenticated attackers to obtain sensitive authentication information via network-based attacks requiring user interaction. The vulnerability affects Microsoft Enterprise Security Token Service (ESTS), the authentication backbone of Azure Entra ID, with scope change indicating potential cross-domain impact. Microsoft has released a patch per MSRC advisory. CVSS 9.3 (Critical) reflects network accessibility, low complexity, and high confidentiality/integrity impact with changed scope.
Local privilege escalation in Microsoft Windows Hyper-V (Windows 11 22H3/23H2 and Windows Server 2022) stems from a use-after-free memory corruption flaw that lets an attacker executing code in a guest context elevate privileges with a scope-changing impact on the host. Microsoft has released fixed builds and rates this critical (CVSS 9.3, total technical impact per CISA SSVC), but there is no public exploit identified at time of analysis and EPSS probability is very low (0.04%). The scope change (S:C) combined with the Hyper-V attack surface is consistent with a guest-to-host virtualization boundary escape.
Schneider Electric EcoStruxure Panel Server can revert credentials to insecure default values under rare circumstances, allowing remote unauthenticated attackers to gain unauthorized access using known factory credentials. This CWE-1188 vulnerability enables complete confidential information disclosure (CVSS 8.2 High). Exploitation requires specific timing conditions (AT:P - Attack Timing: Present) to catch the window when credentials reset. EPSS data not available; no CISA KEV listing or public POC identified at time of analysis, suggesting targeted rather than widespread exploitation risk.
Authentication bypass in Microsoft's SAML SSO plugins for Atlassian Jira (versions 1.0.0 through 1.3.2) and Confluence (versions 1.0.0 through 7.3.x) allows a remote, unauthenticated attacker to impersonate other users and elevate privileges over the network due to an incorrectly implemented authentication algorithm. Rated CVSS 9.1 with full confidentiality and integrity impact, the flaw is deemed automatable with total technical impact by CISA's SSVC framework, though EPSS remains low (0.12%) and no public exploit has been identified at time of analysis. A vendor patch is available.
Remote code execution in Microsoft Dynamics 365 (on-premises) version 9.1 (and 9.0 builds prior to 9.1.44.15) allows an authenticated, high-privileged attacker to run arbitrary code over the network because the application executes with unnecessary privileges (CWE-250). Microsoft has released a patch, and there is no public exploit identified at time of analysis; EPSS is very low (0.07%) and CISA SSVC records no known exploitation, so this is a high-CVSS but not currently exploited issue. The scope-changed CVSS 9.1 reflects that successful code execution can impact resources beyond the vulnerable component's original security authority.
Authentication bypass in Microsoft Azure SDK for Java allows remote unauthenticated attackers to circumvent security controls over the network without user interaction. The vulnerability exposes confidentiality and integrity of Azure services to unauthorized access, with confirmed vendor patch available. CVSS 9.1 reflects critical network-based exploitation against default configurations, though no active exploitation (CISA KEV) or public POC has been identified at time of analysis.
Pre-authentication remote code execution in ChurchCRM before 7.3.2 allows unauthenticated attackers to inject and execute arbitrary code through the setup wizard's unsanitized DB_PASSWORD field. This vulnerability exists because the patch for CVE-2026-39337 was incomplete, leaving the original CWE-94 code injection fully exploitable; publicly available exploit code exists (SSVC exploitation status: poc), and CISA's SSVC framework rates it as automatable with total technical impact. Despite the maximum CVSS score of 10.0, EPSS estimates only a 0.27% 30-day exploitation probability (51st percentile), indicating it is not yet observed in widespread automated campaigns.
Arbitrary code execution in Adobe Connect (versions 2025.9.15, 2025.8.157 and earlier) stems from insecure deserialization of untrusted data (CWE-502) and runs in the context of the current user. Exploitation requires the victim to visit a maliciously crafted URL or interact with a compromised web page, so it is user-interaction gated rather than fully automatic. There is no public exploit identified at time of analysis and it is not on the CISA KEV list; EPSS is modest at 1.50% (81st percentile), consistent with SSVC scoring exploitation as 'none' despite a 'total' technical impact.
Remote code execution in Microsoft SharePoint (Enterprise Server 2016, Server 2019, and Subscription Edition) allows an authenticated attacker to run arbitrary code on the server by submitting maliciously crafted serialized data (CWE-502). An authorized user holding only low-level site privileges can escalate to full server compromise over the network with no user interaction. No public exploit identified at time of analysis, and CISA SSVC records exploitation status as 'none'; EPSS sits at 0.50% (66th percentile), indicating modest near-term exploitation probability despite the high CVSS 8.8 score.
Remote code execution in Microsoft SharePoint Server (Enterprise Server 2016, Server 2019, and Subscription Edition) allows an authenticated attacker (PR:L) to run arbitrary code over the network by submitting maliciously crafted serialized data that the server deserializes without validation. Microsoft has released a patch and no public exploit identified at time of analysis; EPSS is modest at 0.50% (66th percentile) and CISA SSVC rates exploitation as 'none' but technical impact as 'total'. The bug follows the well-known SharePoint deserialization RCE pattern (CWE-502), making it a high-value target for post-authentication lateral movement and full server compromise.
Remote code execution in Microsoft SharePoint Server (2016, 2019, and Subscription Edition) allows an authenticated attacker to run arbitrary code across the network by submitting maliciously crafted serialized data (CWE-502). Any user holding low-privilege authenticated access to the SharePoint web application can achieve total compromise of the server (C:H/I:H/A:H). There is no public exploit identified at time of analysis, and the EPSS probability is modest at 0.50% (66th percentile), but the impact is total and a vendor patch is already available.
Remote code execution in Microsoft SharePoint (Enterprise Server 2016, Server 2019, and Server Subscription Edition) allows an authenticated attacker to run arbitrary code over the network by supplying maliciously crafted serialized data (CWE-502). Any user holding valid low-privilege credentials on the SharePoint farm can achieve total compromise of confidentiality, integrity, and availability. There is no public exploit identified at time of analysis, and CISA SSVC records exploitation status as none; EPSS is a modest 0.50% (66th percentile).
Stack-based buffer overflow in the arduino-esp32 WebServer multipart form parser (all releases prior to 3.3.8) lets a remote attacker crash the device and potentially achieve code execution by sending an oversized multipart/form-data boundary value. The parser sizes a stack Variable Length Array from the attacker-controlled Content-Type boundary field with no length cap, so a boundary longer than ~8000 characters overruns the 8192-byte loopTask stack. Publicly available proof-of-concept exploit code exists (SSVC: poc); it is not listed in CISA KEV and EPSS remains low at 0.20%.
Remote code execution in TinyZero's HDFS utilities allows unauthenticated attackers to execute arbitrary OS commands via crafted file paths passed through the Hydra configuration framework. The vulnerability stems from unsanitized user input directly interpolated into os.system() shell commands within the _copy() function, affecting all deployments through commit 6652a63c57fa. No active exploitation confirmed at time of analysis, but EPSS score of 0.14% (33rd percentile) suggests below-average likelihood despite CVSS:9.8 critical rating. The attack requires network access to the TinyZero training process and ability to control path parameters via configuration.
Remote code execution in Horovod distributed training framework (versions through 0.28.1) allows unauthenticated network attackers to execute arbitrary code on worker nodes by injecting malicious pickle payloads into the KVStore HTTP server. The vulnerability combines unauthenticated write access to the KVStore coordination server with unsafe deserialization using cloudpickle.loads(), enabling trivial exploitation against any reachable Horovod cluster. EPSS score of 0.12% (31st percentile) suggests low widespread exploitation probability despite critical CVSS 9.8 rating, and no active exploitation confirmed (not in CISA KEV). Public exploit development is highly feasible given the straightforward attack path and publicly documented details.
Authentication bypass in SillyTavern (< 1.18.0) lets any network client that can reach the server port directly inject the `Remote-User` (Authelia) or `X-Authentik-Username` (Authentik) SSO headers and log in as any user - including administrators - without a password, because the app never verifies the headers came from the trusted reverse proxy. The flaw is only exploitable when `sso.autheliaAuth` or `sso.authentikAuth` is enabled in config.yaml (both default to false), and is aided by an unauthenticated `/api/users/list` endpoint that leaks admin handles. There is no public exploit identified at time of analysis, though the advisory includes a working PoC, and EPSS is low at 0.07%.
Security-feature bypass in GitHub Copilot and Visual Studio (affecting Visual Studio Code builds before 1.119.1) lets a remote, unauthenticated attacker inject unneutralized special elements into a downstream component (CWE-74) to defeat a built-in protection, provided a developer is lured into a triggering action (UI:R). With CVSS 8.8 and full confidentiality/integrity/availability impact, successful exploitation could subvert Copilot/editor security controls; however there is no public exploit identified at time of analysis, EPSS is low (0.06%), and CISA SSVC records exploitation status as none.
Remote code execution in Microsoft Data Formulator versions prior to 0.7 lets an unauthorized attacker run arbitrary code by inducing a user to interact with malicious content (CWE-94 code injection). The flaw is network-reachable and requires no authentication, but does require user interaction (CVSS 3.1 8.8, AV:N/UI:R). A vendor patch is available; there is no public exploit identified at time of analysis and EPSS probability is very low (0.06%).
Arbitrary code execution in Ludwig framework ≤0.10.4 occurs when attackers supply malicious pickle files to the predict() method, which deserializes untrusted data without validation using pandas.read_pickle(). Remote unauthenticated attackers can achieve full system compromise by exploiting the automatic file format detection mechanism that processes .pkl files through Python's unsafe pickle module. EPSS score of 0.06% (19th percentile) suggests low current exploitation likelihood despite the critical CVSS 9.8 rating, though no public exploit code or active exploitation has been identified at time of analysis.
Arbitrary code execution via torch-checkpoint-shrink.py script in ml-engineering project allows remote attackers to execute malicious Python code by providing crafted PyTorch checkpoint files. The vulnerability stems from insecure deserialization where torch.load() processes .pt files without the weights_only=True safeguard, enabling pickle-based arbitrary object instantiation. Despite a critical CVSS 9.8 score, EPSS probability is low (0.06%, 19th percentile) and no public exploit or active exploitation is confirmed, suggesting limited real-world targeting to date. SSVC assessment indicates total technical impact with automatable exploitation potential, making this a priority for organizations using ml-engineering scripts in production environments.
Remote code execution in Adversarial Robustness Toolbox (ART) through version 1.20.1 allows unauthenticated network attackers to execute arbitrary Python code by uploading malicious PyTorch model files to pipeline-accessible object storage locations. The vulnerability stems from unsafe use of torch.load() without the weights_only=True parameter in the Kubeflow component's model loading process, enabling Pickle deserialization of arbitrary objects. With CVSS 9.8 (AV:N/AC:L/PR:N/UI:N) but only 0.06% EPSS exploitation probability (19th percentile), this represents a critical-severity issue with low observed real-world targeting, likely due to the specialized nature of ML robustness evaluation deployments. No active exploitation confirmed (not in CISA KEV) and no public exploit code identified at time of analysis.
Privilege elevation in Microsoft Visual Studio Code (all releases from 1.0.0 up to but not including 1.119.1) stems from a session fixation weakness (CWE-384) that a remote, unauthenticated attacker can leverage to hijack a victim's authenticated session after the victim is lured into interacting with an attacker-supplied session identifier. The flaw carries a CVSS 8.8 rating with total confidentiality, integrity, and availability impact, but currently has no public exploit identified at time of analysis and a low EPSS score of 0.06% (18th percentile). Microsoft self-reported the issue and has shipped a fix, while CISA's SSVC framework records exploitation as 'none' and non-automatable.
Remote code execution in Guardrails AI through version 0.6.7 occurs when installing validator packages via the Hub mechanism. The guardrails hub install command dynamically executes post-installation scripts from Hub manifests without validating the script path or content, allowing attackers who publish malicious packages to achieve arbitrary code execution on victim systems during package installation. With CVSS 9.8 (AV:N/AC:L/PR:N/UI:N) but only 0.06% EPSS (18th percentile), this represents a supply chain attack requiring user-initiated installation rather than widespread automated exploitation. No active exploitation confirmed (not in CISA KEV), and patch availability not confirmed from available data.
Remote code execution in Adversarial Robustness Toolbox (ART) versions through 1.20.1 allows unauthenticated network attackers to execute arbitrary Python code via unsafe eval() usage in the Kubeflow robustness evaluation component. The vulnerability accepts unsanitized user input for LossFn and Optimizer parameters in PyTorch model evaluations, enabling complete system compromise. With CVSS 9.8 but only 0.06% EPSS score (18th percentile), this represents a severe theoretical risk that has not yet manifested in widespread exploitation. No public exploit code identified at time of analysis, and the vulnerability requires specific deployment of ART's Kubeflow integration component.
Remote code execution in Cognee v0.4.0 and earlier allows unauthenticated attackers to execute arbitrary Python code via the notebook cell execution API endpoint. The vulnerability stems from unsafe use of Python's exec() function without sandboxing or validation, enabling complete system compromise with server process privileges. While not actively exploited (not in KEV), the vulnerability is automatable with total technical impact per SSVC framework, though EPSS indicates low exploitation probability at 0.06%.