Privilege escalation in SurrealDB before v3.1.0 lets unauthenticated attackers hijack the privileges of concurrently-active authenticated sessions via a TOCTOU race condition on the HTTP /rpc endpoint — the primary interface used by all official SurrealDB SDKs. Because concurrent requests share mutable authentication state, an unauthenticated request that races in before the state is cleared executes as the authenticated user, potentially escalating to root. No public exploit is identified at time of analysis, but the flaw is credited by VulnCheck and carries a high CVSS 4.0 score of 9.2.
Privilege escalation and information disclosure in Best Practical's Request Tracker (RT) 6.0.0-6.0.2 and versions before 5.0.10 lets a privileged but non-administrative user harvest other users' authentication credentials - including those of administrators - through the REST 2.0 API and replay them against RT's RSS/iCal feed endpoints to read data as those victims. Because the same credential-exposing request also rotates the tokens, exploitation is destructive: it silently invalidates every previously distributed feed URL across the instance. No public exploit has been identified at time of analysis; the flaw is fixed in RT 5.0.10 and 6.0.3.
Arbitrary file write in pulpcore's FilesystemExport feature allows an authenticated administrator to escape the intended export directory via directory-traversal sequences embedded in a user-controlled relative_path. Because the exported file content is also attacker-supplied (an uploaded artifact), the flaw lets a privileged user plant or overwrite arbitrary files anywhere the Pulp service user can write, enabling service compromise or broader host takeover. Rated CVSS 9.0 and reported by Red Hat, it affects pulpcore as embedded in Ansible Automation Platform 2, Satellite 6, and Red Hat Update Infrastructure 4/5; there is no public exploit identified at time of analysis.
TLS hostname verification bypass in the lettre Rust mailer library (0.10.1 through 0.11.21) exposes users of the boring-tls feature to man-in-the-middle attacks; an inverted boolean caused the default strict configuration to silently disable hostname checking. An on-path attacker holding any chain-valid certificate for any domain can transparently intercept SMTP submission, harvesting PLAIN/LOGIN credentials and full message contents. There is no public exploit identified at time of analysis, but the one-line root cause is visible in the public fix commit, making the flaw trivially understandable; native-tls and rustls backends are not affected.
Authenticated remote code execution in Wazuh's Cluster Distributed API (DAPI) lets any peer holding the shared cluster key coerce the master node into deserializing an attacker-controlled callable (CWE-502) and running it under a forged RBAC context, yielding full manager compromise. Rated CVSS 9.1 and affecting all cluster deployments from 4.0.0 up to the 4.14.5 fix, it combines an overly broad deserialization allowlist with client-controlled authorization to bypass permission checks entirely. SSVC records a proof-of-concept, so publicly available exploit code exists, though EPSS remains low at 0.19% (9th percentile), consistent with a signal that requires possession of the cluster key rather than opportunistic mass exploitation.
Improper access control in the Page Builder CK extension for Joomla (versions 1.0.0 through 3.6.1) allows remote unauthenticated attackers to enumerate and view frontend page list views that should be restricted, effectively bypassing intended authentication on that view. The CVSS 9.1 rating reflects high confidentiality and integrity exposure over the network with no privileges or user interaction, though CISA's SSVC assessment rates the technical impact as only 'partial' and no public exploit has been identified. EPSS is very low (0.13%, 3rd percentile), indicating no observed exploitation activity to date.
Cleartext SFTP credential disclosure in VSee Clinic 7.1.26 and VSee Clinic API 1.3.0 allows any unauthenticated remote attacker to harvest working SFTP logins by simply reading the HTTP responses of three exposed endpoints. The credentials appear whenever SFTP integration has been configured in the application, giving the attacker direct access to the backing SFTP server and whatever clinical or file data it holds. No public exploit or CISA KEV listing exists at time of analysis, but with a 9.0 CVSS and zero authentication required, exploitation is trivial once the endpoints are located.
Stored cross-site scripting in HeyForm's form builder (all versions prior to 3.0.0-rc.7) lets a low-privileged team member persist malicious JavaScript that fires in the browser of a team owner who later views the affected form, yielding full account takeover through privilege escalation. The flaw stems from insufficient sanitization of rich-text/HTML block content, allowing event-handler attributes and dangerous URL protocols to survive serialization. No public exploit has been identified at time of analysis, but the underlying commit (cc97d27) and GHSA advisory clearly document the injection vector, and the CVSS 9.0 rating reflects the high blast radius of hijacking an owner session.
Authenticated PHP code execution in ThemeXpert's Quix Page Builder Pro extension for Joomla lets a privileged builder user (holding core.create or core.edit permissions) embed raw PHP tags in element content that are subsequently executed when the server's view-cache layer include()s the cached template. Exploitation requires caching to be enabled, which is the default, and yields full remote code execution on the underlying web server. No public exploit identified at time of analysis, though the reporter is Joomla itself and the CVSS 4.0 base score is 8.9 (High).
Information disclosure in Elementor Website Builder (all versions before 4.1.4) allows authenticated WordPress users with Contributor-level access or above to read the title, body, and metadata of private posts, private pages, and drafts belonging to other users - including site administrators - via an improperly permission-checked REST endpoint. A publicly available proof-of-concept exploit exists (reported by WPScan), though EPSS places exploitation probability at 0.14% (4th percentile), indicating limited current in-the-wild activity despite public PoC. No active exploitation has been confirmed by CISA KEV; the vendor has released a patched version (4.1.4).
NTLM credential theft and denial-of-service in Scille Parsec (parsec-cloud before 3.3.3-rc.0) arises because the Windows client mounts organization workspaces using their unsanitized names, and a workspace name containing backslashes can evaluate to a UNC path. A low-privileged organization member who names a workspace as an attacker-controlled UNC path (e.g. \\attacker\share) forces victim clients to reach out over SMB, leaking their NetNTLM hash to the attacker or hanging the application if the path is unreachable. Publicly available exploit code has not been identified and it is not in CISA KEV, though a fixed vendor release exists.
Information disclosure in Google Chrome prior to 150.0.7871.128 allows a remote attacker to leak sensitive data after luring a user to a malicious web page, per the Chrome stable channel update. The CVSS 3.1 score of 8.8 reflects network reach with only user interaction required, and a vendor patch is available. No public exploit identified at time of analysis, and EPSS estimates a low ~0.29% probability of near-term exploitation.
Authenticated remote code execution in FreeScout help desk prior to version 1.8.224 allows any logged-in agent to upload a malicious .pht file that executes as PHP on the default Apache plus libapache2-mod-php stack, yielding arbitrary command execution as the www-data user. The flaw is a denylist bypass: the restricted-extension filter added by the earlier CVE-2025-48471 fix blocked phtml and phar but missed .pht (as well as .phtm and .phps), which Apache's default FilesMatch handler still runs as PHP. No public exploit or CISA KEV listing is identified at time of analysis, but the technique is well understood and low-complexity given a valid agent account.
Use-after-free memory corruption in Google Chrome prior to 150.0.7871.128 allows a remote attacker to corrupt heap memory and potentially achieve arbitrary code execution when a victim visits a malicious web page. The flaw carries a CVSS 9.6 with a scope change, reflecting the risk of escaping the renderer sandbox, though exploitation requires user interaction. No public exploit has been identified at time of analysis, and the EPSS probability is low (0.27%, 19th percentile), consistent with a freshly patched browser bug that has not yet been weaponized.
Use-after-free memory corruption in Google Chrome versions prior to 150.0.7871.128 allows a remote attacker to corrupt heap memory and potentially disclose sensitive information or execute code after a user visits a malicious web page. The flaw was reported through the Ubuntu vendor channel and fixed in Chrome's Stable desktop channel; no public exploit identified at time of analysis, and its EPSS exploitation probability is low (0.22%). Successful exploitation requires user interaction (visiting attacker-controlled content) but no authentication.
Signature-verification bypass in Network-AI before 5.13.4 lets unauthenticated remote attackers forge APS delegation payloads and mint signed permission-grant tokens for privileged scopes, including SHELL_EXEC. The root cause is a default local verifier in APSAdapter that treats any non-empty string as a valid cryptographic signature, so no real key material or authentication is needed. No public exploit identified at time of analysis, and the flaw is not listed in CISA KEV, but the trivial exploitation logic and network reachability make it a high-priority fix.
Deadlock in the Linux kernel Bluetooth L2CAP subsystem during connection teardown can be triggered by an adjacent attacker via crafted Bluetooth traffic, leading to a denial of service. The vulnerability arises from an incorrect mutex acquisition order in l2cap_conn_del() and process_pending_rx(), as confirmed by a provided proof-of-concept and static analysis. While the official CVSS score indicates full compromise, the exploit only demonstrates a deadlock; no public exploitation outside of the PoC has been identified, and EPSS probability is low (0.16%).
Shortcode injection in the Reviews Feed WordPress plugin before 2.6.5 allows unauthenticated network attackers to execute arbitrary WordPress shortcodes on any site page that renders the affected feed. The attack is indirect: an adversary plants a shortcode payload inside a review on a connected third-party review platform (e.g., a Google Business profile linked to the plugin); when the plugin ingests and renders that review through its dynamic block without neutralizing shortcode syntax, WordPress's native shortcode processor executes the embedded macro. A publicly available proof-of-concept exists per WPScan; no active exploitation has been confirmed in CISA KEV, though the public POC meaningfully elevates practical risk above what the moderate CVSS score of 4.8 suggests.
Privilege escalation in HCL Commerce (versions 7 through 9.1.19) allows an authenticated low-privileged user to elevate their rights and perform unauthorized administrative operations, disclose other users' personal data, and cause denial of service. The flaw is network-reachable with low attack complexity and requires only existing low-level credentials, yielding high confidentiality, integrity, and availability impact (CVSS 8.8). No public exploit identified at time of analysis; the issue was self-reported by HCL and is not listed in CISA KEV.
Arbitrary code execution in ProFTPD's mod_sftp module (versions before 1.3.9c and 1.3.10rc3) allows authenticated low-privilege SFTP users to run code in the server process by abusing SFTP packet reassembly in fxp.c. The flaw stems from an incorrectly conditioned reallocation of the 16 KB reassembly buffer that corrupts memory-pool freelist metadata, ultimately hijacking pr_fsio_stat() to reach system(). Reported by VulnCheck with a vendor patch available; no public exploit identified at time of analysis, though the advisory describes a complete exploitation primitive.
Privilege escalation in Extreme Networks ExtremeXOS (Switch Engine) EXOS shell allows a low-privileged CLI user to abuse the built-in mv, cp, and rm utilities to escape their authorized filesystem scope and tamper with security-critical files. Because these utilities follow symbolic links without safely canonicalizing paths, an attacker can plant a symlink to a privileged location and, under certain conditions, escalate to root and persistently modify the device software stack. Reported by Extreme Networks with a CVSS 4.0 score of 8.7 (High); no public exploit identified at time of analysis, and it is not listed in CISA KEV.
Denial of service in SurrealDB before 3.1.0 lets an unauthenticated remote attacker crash the database with a single WebSocket message by sending a deeply nested JSON payload to the /rpc endpoint. The value/JSON parser ignores the configured recursion depth limit for nested braces, brackets, and parentheses, so uncontrolled recursion (CWE-674) exhausts server memory and terminates the process. This is an incomplete-fix regression of GHSA-6r8p-hpg7-825g, which patched the same bug class in the expression parser but missed the value/JSON code path; no public exploit identified at time of analysis, though the trigger is trivially reproducible from the advisory.
Denial of service in SurrealDB before 3.1.0 lets an unauthenticated remote attacker crash the entire database process with a single WebSocket message to the /rpc endpoint. Sending an RPC 'use' call that sets a database (db) without first selecting a namespace triggers a Rust panic on an .expect() call; because surrealdb-core is compiled with panic = 'abort', the panic aborts the whole process rather than the connection. The vulnerability (CVSS 4.0 8.7, VulnCheck/GHSA-wjjj-24cx-f28g) requires no credentials, token, or --allow-guests flag, and no public exploit has been identified at time of analysis.
Session hijacking in SurrealDB before 3.1.0 lets unauthenticated attackers enumerate attached session UUIDs through the HTTP /rpc 'sessions' method and then impersonate any authenticated session by supplying an arbitrary 'session' field that the /rpc handler never ownership-checks. Because impersonation inherits the victim session's authorization, an anonymous caller can read, write, and delete data and escalate to that session's level, up to root. There is no public exploit identified at time of analysis, but the vendor GHSA advisory documents the mechanism in full detail, making reproduction straightforward.
Arbitrary file disclosure in ThemeXpert's Quix Page Builder Pro extension for Joomla allows unauthenticated frontend users to read files outside the intended directory by supplying traversal sequences to a Form element on a published page. The flaw (CWE-22) carries a CVSS 4.0 base score of 8.7 driven by high confidentiality impact with no authentication or user interaction. There is no public exploit identified at time of analysis and it is not listed in CISA KEV, but the network-reachable, unauthenticated nature makes it a strong candidate for opportunistic scanning.
Privilege escalation in Extreme Networks ExtremeXOS (EXOS / Switch Engine) lets an attacker reach the privileged debug-mode function because its challenge-response authorization derives the challenge from an insufficiently random source (CWE-338), making the expected response predictable under certain conditions. An attacker holding a valid low-privilege account (remote) or physical serial console access (local) can bypass debug-mode authorization and, depending on device model and firmware, escalate to root and persistently modify the device software stack. There is no public exploit identified at time of analysis and the CVE is not listed in CISA KEV; the vendor (Extreme Networks) self-reported the issue.
Authenticated OS command injection in the system_certificates view lets a high-privileged remote attacker inject arbitrary operating-system commands, yielding total loss of confidentiality, integrity, and availability on the affected device. The flaw was coordinated through CERT@VDE (advisories VDE-2026-044 and VDE-2026-058), indicating an operational-technology / industrial device with a web-based management interface. There is no public exploit identified at time of analysis and it is not listed in CISA KEV; the CVSS 4.0 base score is 8.6.
Improper access control in the Quix Page Builder Pro extension for Joomla (by ThemeXpert) lets authenticated users upload media files even when they lack the media-management permission normally required for that action. The flaw is a permission-enforcement gap on the media upload function rather than a full unauthenticated compromise; no public exploit identified at time of analysis and it is not listed in CISA KEV. The vendor-assigned CVSS 4.0 base score is 8.7 (High), though the supplied vector conflicts with the 'authenticated users' language in the description.
Insecure Direct Object Reference in VSee Clinic 7.1.26 (API 1.3.0) lets an authenticated user tamper with the 'remark' parameter on the /v1.3.0/api/files endpoint to enumerate, download, and delete files owned by other users of the telehealth platform. Because a single low-privileged account yields full read and destructive access to every user's files, the flaw combines confidentiality with integrity and availability loss. There is no public exploit identified at time of analysis and it is not listed in CISA KEV, but the technique is trivial and self-evident from the SRA Labs disclosure.
Unauthenticated information disclosure in FileCodeBox before 2.4 lets remote attackers defeat the IPRateLimit anti-bruteforce control by spoofing X-Real-IP and X-Forwarded-For headers, which the application trusted without verifying the request originated from a configured reverse proxy. By rotating a unique fake IP on every request, an attacker resets their throttling counter indefinitely and can brute-force the short share codes to download other users' files. CVSS 4.0 is 8.7 (high), the fix shipped in release V2.4, and no public exploit or CISA KEV listing was identified at time of analysis.
Missing authentication in HeyForm's POST /api/upload endpoint lets any anonymous internet user upload files up to 10 MB (PDF, Office docs, CSV, MP4, images, etc.) and receive a permanent, public URL hosted on the HeyForm domain. The endpoint is shared by authenticated form creators and unauthenticated submitters but performs no auth guard, no form-context binding, no openToken check, and no session validation, so every request is accepted anonymously in all deployments prior to 3.0.0-rc.9. No public exploit identified at time of analysis, and the flaw is not listed in CISA KEV; abuse is trivial but limited to unauthorized file hosting rather than server compromise.
Cross-tenant authorization bypass in SurrealDB before 3.2.0 lets any authenticated user reach custom API endpoints belonging to other namespaces and databases by manipulating the scope in the URL path. Because the server fails to validate namespace/database scope on custom API routes, an attacker holding valid credentials for one tenant can read sensitive data or trigger unintended operations in unrelated tenants. No public exploit identified at time of analysis, and it is not listed in CISA KEV, but the multi-tenant data-exposure impact is significant (CVSS 4.0 8.6).
Stored cross-site scripting in the Quix Page Builder Pro extension for Joomla (by ThemeXpert) lets an authenticated builder-level user persist malicious scripts - including unsanitised SVG payloads - that then execute in the browser of any visitor or administrator who renders the affected page. Because the payload fires against higher-privileged admins, it enables session/account takeover of the Joomla back end. No public exploit identified at time of analysis, and it is not listed in CISA KEV.
Message injection in FreeScout self-hosted help desk (all versions prior to 1.8.223) lets a remote unauthenticated attacker forge customer replies inside any existing support conversation. By emailing the helpdesk's public intake address with a crafted In-Reply-To header that references a target thread, the attacker's message renders in the agent UI as a legitimate customer reply, silently reopens the conversation, and overwrites last_reply_from with the attacker's identity. No credentials, tokens, or prior access are required (CVSS 8.6); no public exploit has been identified at time of analysis and it is not listed in CISA KEV.
Argument injection in FreeRDP before 3.28.0 lets a malicious .rdp connection file smuggle arbitrary command-line options into the client, because any line beginning with a forward slash is passed straight to the CLI parser rather than treated as an RDP setting. An attacker who convinces a victim to open a crafted RDP file can invoke options such as /rdp2tcp (command execution), /cert:ignore (certificate-validation bypass), or /drive (local filesystem exposure) the moment the file is opened. There is no public exploit identified at time of analysis, but the fix is published (3.28.0) alongside a coordinated VulnCheck and GitHub security advisory.
Arbitrary file read in SurrealDB before 3.1.5 lets an authenticated database user holding the EDITOR or OWNER role disclose files readable by the SurrealDB process by pointing the DEFINE ANALYZER mapper() filter at an arbitrary path and harvesting the content from query error messages. The flaw is exploitable in default deployments because the SURREAL_FILE_ALLOWLIST is empty by default and an empty allowlist enforced no restriction. There is no public exploit identified at time of analysis and it is not on CISA KEV, but the GHSA advisory documents the technique in enough detail to reproduce.
XML injection in @astrojs/rss (all versions from 1.0.0 through 4.0.18) allows an attacker who can supply content to the `source.title` or `enclosure.type` RSS feed item fields to break out of XML string templates and inject arbitrary XML elements into the generated feed. The root issue is that both fields are interpolated directly into raw XML strings before `fast-xml-parser` processes them, with no escaping applied, so XML metacharacters are treated as markup rather than text. A public proof-of-concept is confirmed in the GitHub Security Advisory; no CISA KEV listing is indicated, and EPSS data was not provided in the available intelligence.
HTML injection in the Kirki WordPress plugin before 6.0.12 allows unauthenticated attackers to embed arbitrary HTML into password-reset emails delivered to registered site users, enabling targeted phishing from a trusted site domain. The flaw affects all Kirki versions below 6.0.12 and requires no authentication to trigger, though successful exploitation depends on the target user interacting with the malicious email. A publicly available POC exists per WPScan; the vulnerability is not currently listed in CISA KEV, and the 0.10% EPSS score (1st percentile) indicates minimal observed automated exploitation.
Indirect prompt injection in AgenticMail's inbound-mail bridge lets any external, unauthenticated email hijack the operator's Claude Code session, which resumes with permissionMode 'bypassPermissions' and full tooling (Bash/Write/Edit/WebFetch plus the agenticmail MCP toolbelt) under the operator's OAuth identity. Affected are @agenticmail/core < 0.9.43, @agenticmail/claudecode < 0.2.39, @agenticmail/codex < 0.1.33, and @agenticmail/openclaw < 0.5.71, where the bridge-wake handler embeds attacker-controlled from/subject/preview verbatim into the agent prompt with no sender verification - despite a sibling handler correctly gating on isOperatorReplySender. There is no public exploit identified at time of analysis and it is not on CISA KEV, but the GitHub Security Advisory (GHSA-fq4x-789w-jg5h) documents the exact vulnerable code paths, and EPSS was not supplied.
Brute-force protection bypass in Glance (glanceapp/glance) 0.8.5 lets remote unauthenticated attackers defeat the per-IP failed-login lockout by rotating the leftmost value of the X-Forwarded-For header, but only when the server 'proxied' option is enabled. Because the authentication handler trusted the client-supplied leftmost XFF value instead of the trusted reverse proxy's rightmost value, each guess appears to come from a new IP and the lockout counter never trips, enabling unlimited credential guessing against the login endpoint. No public exploit identified at time of analysis and it is not in CISA KEV, though the upstream fix (PR #1033) and VulnCheck advisory make the technique trivially reproducible.
Predictable session ID generation in the Dancer2 Perl web framework (all versions through 2.1.0) allows remote attackers to guess valid session identifiers and hijack user sessions. When both the Math::Random::ISAAC::XS and Crypt::URandom CSPRNG modules are absent, Dancer2::Core::Role::SessionFactory::generate_id silently falls back to a SHA-1 hash of low-entropy, largely deterministic inputs (built-in rand(), module path, a counter, PID, an object memory address, and a rand-shuffled string). There is no public exploit identified at time of analysis, KEV lists no active exploitation, and EPSS is low (0.14%), but the flaw is rated CVSS 8.2 and CISA SSVC marks it automatable with partial technical impact.
Heap and stack memory disclosure in rsync versions prior to 3.4.3 lets a malicious sender leak process memory from a receiver by overflowing an unchecked 32-bit signed counter in the compressed-token decoder. The leaked contents can include environment variables, passwords, and heap/stack/library pointers, undermining ASLR and enabling follow-on exploitation. No public exploit is identified at time of analysis, there is no CISA KEV listing, and EPSS was not provided, so real-world exploitation should be treated as unproven but plausible.
Privilege escalation via broken access control in NextCRM (open-source CRM by pdovhomilja) before version 0.12.0 lets any authenticated low-privilege user (member or viewer) call the activateUser and deactivateUser Next.js Server Actions, which never verify the caller holds the admin role. An attacker can therefore toggle any account's active state - including deactivating the main administrator to lock legitimate admins out, or reactivating disabled accounts. No public exploit identified at time of analysis, but the fix is confirmed in release v0.12.0 and the root cause is well documented in the vendor security advisory.
Broken object-level authorization in dataCycle-CORE (versions up to and including 25.07.3) lets any authenticated API user with a valid access token impersonate another user by passing a `user_email` parameter to the collection API, causing permissions to be evaluated as the victim and exposing that user's collections. In the V4 code path the same controller further exposes `add_item` and `remove_item` routes lacking object-level `authorize!` checks, enabling a likely cross-user modification path once a collection ID is known. No public exploit is identified at time of analysis, and the flaw is fixed in version 26.06.08.
Server-Side Request Forgery in Apache Syncope lets a low-privileged authenticated user abuse the Connectors and Resources connectivity check to coerce the server into issuing arbitrary outbound requests. Affected builds span the 3.0.x, 4.0.x, and 4.1.x lines up to 3.0.16, 4.0.6, and 4.1.1 respectively, with high confidentiality and integrity impact per the CVSS 8.1 rating. There is no public exploit identified at time of analysis, and CISA SSVC scores exploitation as 'none,' but the total technical impact and low privilege bar make this a meaningful internal-network exposure.
Sensitive data exfiltration in Gitleaks before 8.30.1 lets an attacker who controls the report template read arbitrary host environment variables through non-hermetic Sprig functions (env, expandenv, getHostByName) and smuggle them out via attacker-triggered DNS lookups. Because Gitleaks runs in CI/CD with broad access to credentials, tokens, and API keys - including the very secrets it just discovered during a scan - a malicious template turns the scanner into an exfiltration channel. A detailed public write-up describing the technique and the fix exists; there is no CISA KEV listing, so this reflects a demonstrated technique rather than confirmed active exploitation.
Path traversal in Menyoo (MenyooSP), a popular GTA V single-player trainer/mod, before commit 729aa48 lets a local user supply crafted file names (containing traversal sequences) to the Spooner, VehicleSpawner, WeaponOptions, and PedComponentChanger save/create-folder/rename functions, writing or renaming files outside the intended data directories. Because the resulting XML/config files can be planted in attacker-chosen locations, this escalates to arbitrary code execution in the victim's context. There is no public exploit identified at time of analysis and it is not in CISA KEV; the fix (added IsSafePath() input validation) is publicly visible in PR #550.
Arbitrary code execution in Adobe DNG SDK version 1.7.1 (build 2536) and earlier arises from a stack-based buffer overflow triggered when parsing a maliciously crafted image file, allowing an attacker to run code in the context of the current user. Applications that embed the DNG SDK to decode Digital Negative (DNG) raw camera files are affected, and successful exploitation requires the victim to open the malicious file. No public exploit identified at time of analysis, and the flaw is not listed in CISA KEV.
We need to analyze CVE-2026-64191. The provided data includes description, CVSS, EPSS, patch status, CPE, affected versions, references. I need to synthesize a comprehensive analysis, return JSON with specific fields. The vulnerability is in Linux kernel, specifically i2c-stub driver. It's a development/test tool, not built by default. Local user with access to /dev/i2c-* can cause out-of-bounds read/write via I2C_SMBUS ioctl with crafted block transfer. The fix rejects invalid lengths. Now, let's fill each field carefully. product_name: it's "Linux kernel" or more specifically "i2c-stub driver"? The instructions say "1-3 words ONLY. The specific product or service name affected." The vulnerability is in the Linux kernel's i2c-stub driver. But product name should be "Linux Kernel" because it's a driver within the kernel. However, maybe "i2c-stub driver" is too specific; "Linux Kernel" is the product. CPE is for linux:linux. The EUVD lists Linux various versions. So product_name = "Linux Kernel". I'll say "Linux Kernel" (2 words). summary: 2-3 sentence plain-language synthesis. Must not start with "A vulnerability", etc. Start with impact verb and product. Explain what, who, what attacker can do. Use all data: it's a local out-of-bounds access in i2c-stub test driver, allowing local user to read/write beyond buffer, potentially leading to privilege escalation or information disclosure. Actively exploited? No KEV. EPSS is 0.17% very low. POC likely exists because description includes a reproducer with KASAN output. But no explicit "POC=YES". The description says "A local user with access to /dev/i2c-* can issue an I2C_SMBUS ioctl ... causing ... out-of-bounds". That's a POC description. So "publicly available exploit code exists". But note that the driver must be loaded with chip_addr parameter. So it's not widely deployed. Summary: "Local privilege escalation in Linux Kernel's i2c-stub test driver allows any local user with access to /dev/i2c-* devices to trigger
Use-after-free in the Linux kernel's Qualcomm rmnet networking driver allows a local attacker with low privileges to escalate to kernel-level code execution. The flaw occurs when the rmnet endpoint is freed while still being referenced by concurrent receive path handlers, leading to memory corruption. No active exploitation or public exploit code has been identified, but the high CVSS score reflects a significant local privilege escalation risk.