Arbitrary JavaScript execution in the mathjs (npm) expression parser lets an attacker who can submit expressions escape the math sandbox and run code in the host JavaScript runtime. It affects mathjs versions >= 13.1.1 and < 15.2.0 and stems from an unsafe object/array property setter that failed to block dangerous properties. Publicly available exploit code exists (a proof-of-concept gist), but there is no evidence of active exploitation; EPSS is only 0.05% (16th percentile), consistent with the CISA SSVC exploitation status of 'none'.
Unauthenticated information disclosure (IDOR) in the My Calendar WordPress plugin (versions ≤ 3.7.6) lets remote attackers read calendar events-including private and hidden ones-from any sub-site on a WordPress Multisite network by injecting an arbitrary site ID into the unauthenticated mcjs_action AJAX endpoint. On Single Site installations the same request triggers an uncaught fatal error, giving an unauthenticated denial-of-service vector. Publicly available exploit code exists, though EPSS remains very low (0.01%) and the flaw is not in CISA KEV. Fixed in version 3.7.7.
Unauthenticated remote code execution in the goodoneuz/pay-uz Laravel package (all versions through 2.2.24) lets remote attackers overwrite executable PHP payment-hook files through the /payment/api/editable/update endpoint, which is registered via Route::any() with no authentication middleware. User-supplied content is written verbatim with file_put_contents() into PHP files that are later loaded via require() during normal payment processing, so any web-facing deployment can be fully compromised without credentials. CVSS is 10.0 and SSVC rates technical impact as total and the attack as automatable; there is no public exploit identified at time of analysis and EPSS is a modest 0.89% (75th percentile).
Authenticated remote command execution in FlowiseAI Flowise (npm packages flowise and flowise-components) versions <= 3.0.13 allows any logged-in user to run arbitrary OS commands by configuring a Custom MCP stdio server. The MCP adapter's allow-list of 'safe' binaries (e.g. npx) can be abused with code-execution arguments that bypass the validateCommandInjection and validateArgsForLocalFileAccess sanitizers, yielding total host compromise. Publicly available exploit code exists (SSVC exploitation: poc), though EPSS is low at 0.07%; it is not listed in CISA KEV.
Unauthenticated remote attackers can escalate to administrator privileges on WordPress sites running Riaxe Product Customizer plugin ≤2.1.2. The plugin exposes an AJAX endpoint ('install-imprint') without authentication checks that allows arbitrary WordPress option manipulation, enabling attackers to create administrator accounts by modifying registration settings. CVSS 9.8 (Critical) reflects the complete site compromise potential. EPSS data not provided but exploitation requires only HTTP access to any vulnerable WordPress installation with this plugin active-no special conditions beyond plugin presence.
SQL injection in SourceCodester Vehicle Parking Area Management System v1.0 lets remote unauthenticated attackers manipulate backend database queries through the /parking/manage_park.php endpoint. Public proof-of-concept exploit code exists (per the mt-0505 CVE report and VulDB entry), and the flaw is automatable with total technical impact, though EPSS remains very low at 0.03% (9th percentile). No public evidence of active exploitation has been identified.
SQL injection in SourceCodester Simple Music Cloud Community System v1.0's /music/edit_music.php endpoint allows unauthenticated remote attackers to execute arbitrary SQL commands with full database access. The vulnerability carries a critical CVSS 9.8 score with network-accessible, low-complexity exploitation requiring no privileges or user interaction. EPSS probability is notably low at 0.01% (2nd percentile), suggesting limited real-world exploitation interest despite the critical severity rating. Proof-of-concept code is publicly available via GitHub, lowering the barrier for opportunistic attacks against exposed instances.
SQL injection in SourceCodester Simple Music Cloud Community System v1.0 allows unauthenticated remote attackers to execute arbitrary SQL commands via the /music/view_genre.php endpoint, enabling complete database compromise including data theft, modification, and potential server takeover. CVSS 9.8 (Critical) with network-exploitable attack vector requiring no authentication or user interaction. EPSS score of 0.01% (2nd percentile) indicates low observed exploitation probability despite critical severity. Publicly available proof-of-concept exists (GitHub repository), lowering exploitation barrier for opportunistic attackers.
Integer underflow in Microsoft QUIC's ACK frame parser enables remote unauthenticated privilege escalation. The vulnerability (CWE-191: integer wrap-around) affects Microsoft's native QUIC library implementations (both OpenSSL and SChannel variants) distributed via NuGet packages. With CVSS 9.8 (AV:N/AC:L/PR:N/UI:N) and vendor-confirmed patch available (commit 1e6e999b), this represents a critical network-exposed flaw in QUIC protocol implementations. No active exploitation confirmed (not in CISA KEV) and public exploit code status unknown at time of analysis, but the straightforward attack vector (network-accessible protocol parsing) and authentication bypass capability warrant immediate patching priority for systems using Microsoft QUIC libraries.
OS Command Injection in HGiga iSherlock-base and iSherlock-audit versions 4.5 and 5.5 allows remote unauthenticated attackers to execute arbitrary operating system commands on the server with full system privileges. All four product variants (iSherlock-base-4.5, iSherlock-audit-4.5, iSherlock-base-5.5, iSherlock-audit-5.5) are affected in versions below build 476 (base) and 261 (audit). Vendor-released patch available per Taiwan CERT (TWCERT) advisory. CVSS 4.0 score of 10.0 reflects maximum severity with network attack vector, no authentication required, and high impact to all CIA triad properties including scope change. No public exploit identified at time of analysis.
SQL injection in SourceCodester Simple Music Cloud Community System v1.0 lets remote attackers manipulate backend database queries through the /music/view_user.php endpoint. Per the CVSS vector (PR:N), no authentication is required, and publicly available exploit code exists (SSVC exploitation: poc), though it is not listed in CISA KEV and carries a very low EPSS probability (0.03%, 8th percentile). Exploitation can expose or tamper with stored user and application data.
Heap buffer overflow in Creolabs Gravity scripting language before 0.9.6 enables remote code execution when applications evaluate untrusted scripts containing many string literals at global scope. The vulnerability stems from insufficient bounds checking in gravity_fiber_reassign(), allowing heap metadata corruption. VulnCheck disclosed this issue with a vendor-released patch (commit 18b9195) available. CVSS 9.3 reflects the critical network-accessible, unauthenticated attack vector. No active exploitation (CISA KEV) or public POC identified at time of analysis, but technical details in GitHub issue #437 could facilitate exploit development.
Remote unauthenticated code execution in Openfind MailGates (5.0-6.0) and MailAudit (5.0-6.0) via stack-based buffer overflow allows complete system compromise. Attackers can send crafted network requests to exploit CWE-121 buffer overflow conditions without authentication, achieving arbitrary code execution with high impact to confidentiality, integrity, and availability. Vendor patches available (MailGates 6.1.10.054, 5.2.10.099; MailAudit 6.1.10.054, 5.2.10.099). CVSS 9.3 with network attack vector (AV:N), low complexity (AC:L), and no privileges required (PR:N) creates critical exposure for internet-facing mail security appliances. EPSS data unavailable; no confirmed active exploitation or public POC identified at time of analysis.
WinMatrix agent escalates privileges to SYSTEM without authentication, enabling authenticated local users to execute arbitrary code with full administrative control on both the local machine and all networked hosts where the agent is deployed. This environmental spread capability (CVSS scope change: H) transforms a local vulnerability into an enterprise-wide threat. Taiwan CERT issued advisories in January 2026 for versions 3.5.13 through 3.5.26.15. No public exploit identified at time of analysis, but CVSS 9.3 reflects catastrophic potential impact given the agent's privileged access and network propagation capability. EPSS data not available for new 2026 CVE.
Lua sandbox escape in Luanti 5.x (formerly Minetest) game engine allows malicious mod code to break out of LuaJIT security restrictions and execute arbitrary code on the host system. Affects all Luanti 5.0.0 through 5.15.1 when compiled with LuaJIT instead of standard Lua. Attackers with ability to distribute crafted mods can achieve complete system compromise with scope change (S:C in CVSS), escalating from sandboxed mod execution to full host access. No authentication required but local access needed (AV:L). Patch available in version 5.15.2 via two upstream commits. EPSS data not available; no confirmed active exploitation or public POC at time of analysis.
Unauthenticated remote denial-of-service in ChilliCream's Hot Chocolate GraphQL server for .NET (HotChocolate.Language package) lets any client crash the entire worker process by sending a small, deeply nested GraphQL document. Because the recursive-descent Utf8GraphQLParser has no depth limit, ~40 KB of nested selection sets, object/list values, or list types triggers an uncatchable StackOverflowException that terminates the process, dropping all in-flight HTTP requests, IHostedService background tasks, and WebSocket subscriptions on that worker. No public exploit identified at time of analysis and EPSS is low (0.09%), but the attack is trivial (AC:L, PR:N) and CISA SSVC flags it as automatable, so it is a practical availability threat to internet-facing GraphQL endpoints.
Remote code execution in Digital Knowledge KnowledgeDeliver (all versions prior to February 24, 2026) via malicious ViewState deserialization. A hard-coded ASP.NET machineKey allows unauthenticated remote attackers to bypass ViewState validation and execute arbitrary code on the server. Mandiant reported this critical deserialization vulnerability. EPSS score of 0.08% (24th percentile) suggests low observed exploitation activity, though no public exploit is confirmed at time of analysis. CVSS vector indicates network-accessible attack requiring no privileges or user interaction, but the 7.5 score reflects only Confidentiality impact-real-world RCE capability makes this significantly more severe than the partial CVSS rating suggests.
SQL injection in SourceCodester Payroll Management and Information System v1.0 lets remote unauthenticated attackers manipulate backend database queries through the /payroll/view_employee.php endpoint. The flaw exposes confidential employee and payroll data and permits tampering with stored records. Publicly available exploit code exists (community vulnerability report), though EPSS remains very low (0.03%, 10th percentile) and it is not listed in CISA KEV.
Authentication bypass in @fastify/middie 9.3.1 and earlier lets unauthenticated attackers reach protected routes because inherited middleware is not registered directly on child plugin engine instances. When a Fastify app defines auth/authz middleware in a parent scope and mounts child plugins through middie, those child-scope routes silently skip the parent's checks, exposing them to anyone. Rated CVSS 9.1; SSVC lists a public PoC and total technical impact, though EPSS is only 0.03% and there is no evidence of active exploitation.
Stored cross-site scripting escalating to remote code execution in SiYuan (b3log) versions 3.6.3 and below, where Mermaid diagrams render with securityLevel 'loose' and inject SVG via innerHTML, letting javascript: URLs in Mermaid code blocks survive into the DOM. On Electron desktop builds — created with nodeIntegration enabled and contextIsolation disabled — a victim opening a note with a malicious Mermaid block and clicking the diagram node triggers arbitrary code execution. Publicly available exploit code exists (SSVC 'poc'); not listed in CISA KEV, and EPSS is very low at 0.05%.
In products of the MSE6 product-family by Festo a remote authenticated, low privileged attacker could use functions of undocumented test mode which could lead to a complete loss of confidentiality,. Rated high severity (CVSS 8.8), this vulnerability is remotely exploitable, low attack complexity. No vendor patch available.
Local file inclusion in Livemesh Addons for Elementor (WordPress plugin) ≤9.0 allows authenticated attackers with Contributor-level privileges to include and execute arbitrary PHP files via recursive directory traversal bypass in widget template parameters. The vulnerability requires Elementor plugin installation and either admin interaction (social engineering) or direct Contributor access. CVSS 8.8 reflects high impact (RCE potential) but limited by authentication requirement. No active exploitation confirmed (not in CISA KEV), but publicly available exploit code exists (Wordfence disclosure with technical details and code references).
Arbitrary file write and deletion in the go-acme/lego ACME client (versions before 4.34.0) allows a malicious or compromised ACME server to escape the configured webroot via path traversal. When lego solves an HTTP-01 challenge with the webroot provider, it writes the key authorization to a path built by directly concatenating the server-supplied challenge token; a token containing '../' sequences (e.g. '../../../../tmp/pwned') redirects the write - and the corresponding CleanUp() delete - to any location writable by the lego process, up to remote code execution if lego runs as root. Publicly available exploit code exists (SSVC 'poc'), but the EPSS score is very low (0.05%, 14th percentile), consistent with the requirement that the victim first point lego at an attacker-controlled ACME server.
Cross-Site Request Forgery (CSRF) in Career Section WordPress plugin versions ≤1.6 enables unauthenticated attackers to delete arbitrary server files through social engineering. Attackers trick authenticated WordPress administrators into clicking malicious links that exploit missing nonce validation in the delete action handler, leading to path traversal and unrestricted file deletion. CVSS 8.8 (High) with network attack vector but requires user interaction (UI:R). EPSS and KEV data not provided; public exploit code status unknown. Wordfence advisory and upstream patch commit available.
Privilege escalation in AcyMailing WordPress plugin (versions 9.11.0-10.8.1) allows authenticated Subscriber-level users to gain administrator access through a multi-stage attack chain. Attackers exploit a missing capability check in the wp_ajax_acymailing_router AJAX handler to access admin-only configuration controllers, enable autologin features, inject malicious cms_id values into newsletter subscribers, and authenticate as any WordPress user including administrators. EPSS data not available; no confirmed active exploitation (CISA KEV absent), but the low attack complexity (AC:L) and detailed public code references increase exploitation risk for installations with subscriber registration enabled.
Remote command injection in OpenHarness gateway handler allows authenticated remote chat users to execute administrative commands like /permissions full_auto without authorization, escalating privileges to modify security controls of running instances. Vulnerability exploits insufficient command validation in chat interface. Fixed in commit dd1d235. CVSS 8.7 (High) with network attack vector and low complexity. EPSS data unavailable; not listed in CISA KEV. VulnCheck advisory and GitHub patch available.
Denial of service in mcp-framework (npm) versions <= 0.2.21 lets a remote unauthenticated attacker crash any server using the HttpStreamTransport by sending a single oversized HTTP POST to /mcp, exhausting memory. The readRequestBody() function concatenates request-body chunks into a string with no cap, and although a 4MB maxMessageSize config value exists it is never enforced. Publicly available exploit code exists (SSVC 'poc'), a vendor patch shipped in 0.2.22, and EPSS is low (0.14%, 34th percentile).
Privilege escalation in Microchip IStaX before 2026.03 allows an authenticated low-privileged web user to derive the shared per-device cookie secret directly from their own webstax_auth session cookie and mint a forged cookie granting full administrative access. Because the secret is common to the device rather than per-user and can be reconstructed from an already-issued cookie, any account holder can self-elevate to admin. There is no public exploit identified at time of analysis, and EPSS is very low (0.04%), but the technical impact is total.
Authenticated SQL injection in DataEase (open-source BI/data-visualization platform) versions 2.10.20 and below lets an attacker with valid datasource credentials achieve full read and write access to a connected database via the /de2api/datasetData/previewSql endpoint. User-supplied SQL is wrapped in a subquery without enforcing a single SELECT, and a JDBC blocklist bypass re-enables allowMultiQueries=true, allowing stacked statements such as UPDATE. Publicly available exploit code exists (SSVC: PoC), but EPSS is only 0.03% and it is not in CISA KEV, so no active exploitation is confirmed. Fixed in 2.10.21.
SQL injection in DataEase, the open-source data visualization and analytics platform, allows an authenticated user to execute arbitrary SQL through the API datasource save workflow in all versions up to and including 2.10.20. The flaw stems from the Base64-encoded 'deTableName' value being spliced into a DDL statement via unsanitized string replacement, letting an attacker break out of identifier quoting to run error-based injection that leaks backend database details such as the MySQL version. SSVC rates the technical impact as total and automatable, though publicly available exploit code exists only as a proof of concept and EPSS remains very low (0.03%); the issue is fixed in 2.10.21.
SQL injection in DataEase, an open-source data visualization and analytics platform, lets an authenticated user inject arbitrary SQL via the sort parameter of the datasetData enum-value endpoint, achieving time-based blind SQL injection against the backing database. All versions 2.10.20 and below are affected, with a total-impact potential to read, alter, or disrupt data. There is no public exploit identified at time of analysis, EPSS is very low (0.03%), and it is not listed in CISA KEV, but the technical impact is rated total.
SQL injection in DataEase, an open-source data visualization and analytics platform, allows an authenticated user to inject arbitrary SQL through the orderDirection sorting parameter on dataset endpoints (/de2api/datasetData/enumValueDs and /de2api/datasetTree/exportDataset) in versions 2.10.20 and below. Because the Order2SQLObj class places the raw value directly into the ORDER BY clause via StringTemplate with no whitelist, an attacker can perform time-based blind data extraction and denial of service. Publicly available exploit code exists (SSVC exploitation: poc); it is not in CISA KEV and the EPSS probability is very low (0.03%). Fixed in version 2.10.21.
SQL injection in DataEase's dataset export endpoint (POST /de2api/datasetTree/exportDataset) lets an authenticated attacker inject arbitrary SQL through the expressionTree parameter in versions 2.10.20 and below. User-controlled values in 'like' filter terms are concatenated unsanitized into SQL during filter-tree translation, enabling blind, time-based extraction of database contents. Publicly available exploit code exists (SSVC: poc); EPSS is low (0.03%, 8th percentile) and the flaw is not in CISA KEV, so no active exploitation is confirmed. Fixed in 2.10.21.
CRLF Injection in Openfind MailGates/MailAudit allows remote unauthenticated attackers to read arbitrary system files via HTTP header manipulation. Affects MailGates/MailAudit versions 5.0-6.0 (prior to 5.2.10.099 and 6.1.10.054 respectively). CVSS 8.7 with network vector, low complexity, and no authentication required indicates critical real-world risk. Taiwan CERT advisory published; no CISA KEV listing or public exploit code identified at time of analysis, suggesting early disclosure phase.
Remote denial of service in moby/spdystream (Go SPDY/3 library) versions <= v0.5.0 allows an unauthenticated remote peer to crash any service that accepts SPDY connections. By sending a small number of malformed control frames whose attacker-controlled 32-bit counts and lengths are used as unbounded allocation sizes, an attacker forces the process to allocate gigabytes of memory and hit an out-of-memory crash. There is no public exploit identified at time of analysis and EPSS is very low (0.01%), but the flaw is trivially automatable and reaches container-platform components (advisory titled 'DOS on CRI'), so patched v0.5.1 should be adopted where SPDY is exposed.
SQL injection in DataEase, the open-source data visualization and analytics platform, lets an authenticated attacker execute arbitrary SQL commands via the /datasource/getTableField endpoint in versions 2.10.20 and earlier. The getTableFiledSql method in CalciteProvider.java concatenates the tableName parameter into a query using String.format without parameterization, and the existing table-name validation is bypassed by registering an API datasource with a malicious deTableName that getTables returns as legitimate. Publicly available exploit code exists per the SSVC assessment, though EPSS remains very low (0.03%, 8th percentile) and it is not in CISA KEV; it is one of a cluster of SQLi issues patched together in 2.10.21.
SQL injection in DataEase (versions ≤ 2.10.20) lets an authenticated user run arbitrary SQL against the backing database through the API datasource update endpoint. The attacker-controlled deTableName field is dropped into a CREATE TABLE statement without escaping, yielding error-based injection that can extract database contents. A proof-of-concept is indicated by CISA SSVC, though EPSS remains very low (0.03%) and it is not on CISA KEV; fixed in 2.10.21.
Authentication bypass in Cloud Foundry UAA allows remote unauthenticated attackers to obtain access tokens for arbitrary users when SAML 2.0 bearer assertions are enabled, leading to unauthorized access to all UAA-protected systems. Affects UAA versions 77.30.0 through 78.7.0 and CF Deployment versions 48.7.0 through 54.14.0. The vulnerability stems from acceptance of unsigned and unencrypted SAML 2.0 assertions (CWE-290: Authentication Bypass by Spoofing), enabling complete authentication mechanism bypass with network access and low attack complexity (CVSS 8.6, AV:N/AC:L/PR:N/UI:N). No public exploit identified at time of analysis, but the technical details disclosed in the vendor blog create high weaponization potential.
Arbitrary server-side file deletion in SiYuan (b3log/SiYuan) versions 3.6.3 and prior lets a low-privileged actor with access to the /api/av/removeUnusedAttributeView endpoint - such as a Reader-role or publish context - supply an id parameter containing ../ sequences to escape the storage/av directory and delete any .json file on the server, including conf/conf.json and workspace metadata. The GitHub Security Advisory (GHSA-vw86-c94w-v3x4) includes step-by-step reproduction, so publicly available exploit code exists, but there is no public exploit identified as weaponized and SSVC exploitation status is 'none'; EPSS is low at 0.06%. Fixed in version 3.6.4.
Sandbox escape leading to arbitrary local code execution affects Snowflake Cortex Code CLI (an AI coding agent) in all versions prior to 1.0.25. Because the CLI improperly validates bash commands, an attacker who plants specially crafted content in untrusted material-such as a malicious repository-can cause the agent to run subsequent commands outside its sandbox and execute arbitrary code on the user's device without consent. No public exploit identified at time of analysis; EPSS is low (0.07%) and exploitation is non-deterministic and model-dependent, so real-world reliability is limited.
Arbitrary file read in DataEase (open-source data visualization and analytics platform) versions 2.10.20 and below lets an authenticated attacker exfiltrate sensitive files from the server host by abusing the MySQL datasource configuration. Because the Mysql class exposes an auto-generated public setter for its JDBC security blocklist (via Lombok @Data), a crafted JSON datasource payload replaces the blocklist with an empty list before validation, enabling dangerous JDBC parameters like allowLoadLocalInfile=true. There is no public exploit identified at time of analysis and EPSS is very low (0.04%), but the technical chain is well-documented in the vendor GHSA advisory.
Vision Helpdesk before 5.7.0 (patched in 5.6.10) allows attackers to read user profiles via modified serialized cookie data to vis_client_id. Rated medium severity (CVSS 4.3), this vulnerability is remotely exploitable, low attack complexity. Public exploit code available.
Authentication bypass in Zohocorp ManageEngine Log360 (builds 13000 through 13013) lets remote unauthenticated attackers reach certain protected actions because of an improper request-filter configuration, exposing high-confidentiality data with limited integrity impact (CVSS 8.2). No public exploit is identified at time of analysis and EPSS is low (0.14%), but CISA's SSVC flags the flaw as automatable, so opportunistic scanning is plausible once details spread. Log360 is an enterprise SIEM/log-management console, making any auth bypass on it a high-value target for attackers seeking to blind defenders or harvest aggregated log data.
Remote code execution in sagredo qmail before 2026.04.07 arises because the notlshosts_auto TLS-fallback feature in qmail-remote.c builds a /bin/touch shell command with the remote partner's FQDN and runs it through popen(), enabling OS command injection when that FQDN carries shell metacharacters. An operator-controlled remote mail server (or attacker able to control the MX/hostname a victim delivers to) can inject commands that execute as the qmail-remote (qmailr) user. Publicly available exploit code exists per the SSVC proof-of-concept rating, but it is not listed in CISA KEV and EPSS is very low (0.08%), indicating no evidence of widespread exploitation.
SQL injection in Zohocorp ManageEngine PAM360 (before build 8531) and Password Manager Pro (builds 8600 through 13230) lets an authenticated attacker inject arbitrary SQL through the query report module, enabling extraction and modification of privileged-vault data. Because these products are enterprise password/privileged-access managers, a successful injection can expose stored credentials and secrets for an entire environment. There is no public exploit identified at time of analysis, EPSS risk is very low (0.02%), and a vendor patch is available.
Logic error in Luanti 5 (formerly Minetest) game engine before 5.15.2 allows malicious mods to gain unauthorized access to security-restricted APIs by intercepting mod environment setup. When any mod is designated as trusted (via secure.trusted_mods or secure.http_mods), a specially crafted mod can exploit the environment initialization sequence to receive the insecure environment or HTTP API access intended only for trusted mods. CVSS 8.1 reflects local attack vector with high complexity but no authentication required and scope change with high confidentiality/integrity/availability impact. GitHub security advisory and two fix commits confirm patch availability. No CISA KEV listing or public exploit code identified at time of analysis.
Receiver-side use-after-free in rsync 3.0.1 through 3.4.1 lets a malicious peer corrupt memory on a host running rsync with the -X (--xattrs) option. The receive_xattr function trusts an attacker-supplied length value during a qsort operation, which a malicious sender/server can abuse to crash the receiver or potentially achieve code execution; Linux hosts are vulnerable in many common configurations and non-Linux platforms more broadly. No public exploit identified at time of analysis, EPSS is very low (0.01%), and SSVC scores exploitation as none, indicating a real-but-not-urgent memory-safety bug rather than a mass-exploited threat.
Arbitrary code execution in Eaton Intelligent Power Protector (IPP) software via insecure library loading allows local authenticated attackers with low privileges to execute code with elevated integrity impact across security boundaries. Attack complexity is high, requiring the attacker to have access to the software package installation files. EPSS data unavailable; no CISA KEV listing or public POC identified at time of analysis. Eaton has released a patched version available through their download center.
Arbitrary file write in OCaml opam before 2.5.1 allows a malicious package to escape its intended installation scope: a destination filepath in a package's .install field can contain ../ sequences to reach parent directories and drop files outside the target prefix. Exploitation requires a user to install a crafted opam package, so it is a supply-chain/local vector rather than a remote one. Tracked as OSEC-2026-03 and fixed in opam 2.5.1; no public exploit identified at time of analysis and EPSS is negligible (0.01%).
Authenticated low-privileged users in wger can modify installation-wide gym configuration via /config/gym-config/edit due to missing permission enforcement, enabling vertical privilege escalation. The GymConfigUpdateView declares 'config.change_gymconfig' permission but inherits WgerFormMixin instead of WgerPermissionMixin, causing the permission check to never execute. Exploiting this allows attackers to manipulate default gym assignments affecting all users, with GymConfig.save() automatically reassigning user profiles and creating gym configurations tenant-wide. CVSS 7.6 (High) with network attack vector, low complexity, and low privileges required. No active exploitation (KEV) or public POC identified at time of analysis, though GitHub advisory provides detailed reproduction steps.