Oracle Commerce Guided Search / Oracle Commerce Experience Manager 11.4.0 (Forge component) can be compromised by an unauthenticated attacker who already sits on the same physical network segment as the host, yielding full read access to application data, limited data manipulation, and a repeatable denial-of-service condition. Oracle rates the flaw CVSS 7.1 (AV:A/AC:H/PR:N/UI:N/S:U/C:H/I:L/A:H), reflecting that it is reachable only from an adjacent segment and is difficult to trigger reliably. No public exploit code has been identified at time of analysis and the issue is not listed in CISA KEV; Oracle has published a fix in its Critical Security Patch Update.
Cross-site request forgery in the J2Store e-commerce extension for Joomla (versions 1.0.0–3.3.2, 4.0.0–4.0.22 and 4.1.0–4.1.7) allows an attacker to silently overwrite a shopper's billing or shipping address while their checkout session is active, potentially redirecting a paid order to an attacker-controlled address. The cart, checkout and myprofile controllers omit CSRF token validation, so each forged request executes only with the victim's own session privileges — there is no cross-account data access. No public exploit code or active exploitation has been identified at time of analysis.
Server-side request forgery in mcp-searxng before 1.7.1 lets an unauthenticated network client coerce the MCP server into fetching internal-only resources. The web_url_read tool validates only the literal hostname string before handing the URL to undici's fetch, so an attacker-controlled hostname that DNS-resolves to a loopback, private, link-local, or cloud-metadata address passes the lexical check and is connected to by the OS resolver. Successful abuse can read internal HTTP services, leak credentials or service tokens, and enumerate reachable internal hosts; publicly available exploit code exists and EPSS remains low (0.14%, 3rd percentile), so this is a real but conditional risk concentrated in deployments that expose the default HTTP transport without the MCP_HTTP_HARDEN authorization layer.
Oracle Siebel CRM Deployment (Server Infrastructure component), versions 17.0 through 26.7, allows an attacker who already holds a low-privileged logon on the host where the deployment component runs to read, create, modify, and delete all data accessible to that deployment. The vulnerability is locally exploitable with low complexity and no user interaction, but it does not grant entry to the host itself — it escalates what an existing low-privileged account can reach across the Siebel deployment layer, producing high confidentiality and integrity impact (CVSS 3.1 base 7.1) with no availability impact. No public exploit code has been identified at time of analysis, the flaw is not listed in CISA KEV, and no EPSS score, CWE mapping, or confirmed fix version was supplied with the source data.
Oracle E-Business Suite Applications Manager 12.2.3 through 12.2.15 lets a low-privileged attacker who already holds a logon on the host running Applications Manager read and tamper with critical data through the Command Line - RapidClone component. The flaw is a local-only, low-complexity authorization weakness (CVSS 7.1, AV:L/PR:L/UI:N) that yields high confidentiality and integrity impact with no availability impact, and no scoped privilege escalation beyond the component. No public exploit code has been identified at time of analysis, and no EPSS or CISA KEV signal was supplied with this record.
Local low-privileged users who can log on to a host running Oracle Commerce Guided Search / Oracle Commerce Experience Manager 11.4.0 can abuse the Forge component to create, delete or modify critical search data and to hang or repeatedly crash the service, producing high integrity and availability impact with no confidentiality exposure. The issue carries a CVSS 3.1 base score of 7.1 (AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:H/A:H), so it is not remotely exploitable and requires a pre-existing authenticated foothold on the server, which keeps real-world risk genuinely moderate rather than urgent. There is no CISA KEV entry and no public exploit identified at time of analysis, and exploitation is not confirmed actively exploited.
Rsbuild versions before 2.0.9 contain a command injection flaw in the developer-server browser-launcher on macOS. The openBrowser() helper in packages/core/src/server/open.ts passes a URL through encodeURI() and then interpolates it into a string handed to child_process.exec(), and because encodeURI() leaves characters such as $, (, ), and ; untouched, a crafted URL is evaluated by /bin/sh and can execute arbitrary operating-system commands with the privileges of the developer running the dev server. The issue was fixed in the 2.0.9 release (upstream PR 7789), so a vendor-released patch is available; no CISA KEV listing or public exploit code was present in the supplied intelligence, and the attack requires local conditions on the developer machine.
Stack-based buffer overflow in the A2DP profile of the BlueZ Linux Bluetooth stack allows a network-adjacent attacker to run arbitrary code as root on a target once a malicious Bluetooth audio device has been paired with it. The flaw lies in stream endpoint (SEP) handling, where attacker-supplied capability strings are copied into fixed-size stack buffers without a length check (CWE-121), so oversized SEP data corrupts the stack in the privileged bluetoothd process. CVSS 7.1 (AV:A/AC:H/PR:N/UI:R) reflects the requirement for a paired device and radio proximity; no public exploit code and no CISA KEV listing were identified at time of analysis, and no EPSS score was supplied.
Command injection in the GitHub Actions workflow shipped with githubtoplanguages (versions before 1.1.4) lets anyone who can open an issue on a repository running the bundled discord-issue.yml workflow execute arbitrary shell commands on the Actions runner. The issue title is expanded by GitHub Actions directly into a Bash assignment, so a title containing $(...) command-substitution syntax is evaluated by the shell before the value is posted to Discord; a public proof-of-concept was demonstrated by the reporter in a fork. A successful injection can spoof trusted bot notifications and may expose the configured DISCORD_WEBHOOK secret or other workflow environment data, and remediation is available in version 1.1.4 / commit 6bf9c3a9cb66c937b9047ca266b3d02f2bb11027.
Any authenticated user of a Payload CMS instance running @jhb.software/payload-cloudinary-plugin (0.3.0 through 0.3.x, fixed in 0.4.0) can obtain a cryptographically valid Cloudinary HMAC-SHA1 upload signature for parameters of their choosing, because the client-upload signing endpoint POST /api/cloudinary-generate-signature forwards attacker-controlled 'paramsToSign' straight to cloudinary.utils.api_sign_request() with no key allowlist, folder policy, or timestamp check. With that signature plus the intentionally client-visible Cloudinary API key, an attacker can overwrite existing assets, flip upload visibility with type=private, register notification_url callbacks to attacker-chosen hosts, force cache invalidation, and write files outside the intended upload folder. The issue is only reachable when the plugin's clientUploads option is enabled, and no public exploit code or active-exploitation reporting was present in the supplied intelligence.
Cost-guardrail bypass in atomic-agents-stack 1.0.0 and earlier lets a caller run parallel helper/delegate batches with a model identifier that is absent from the hardcoded pricing table, causing the per-batch reservation to be calculated as $0.00 and silently skipped. Because the reservation is the only defense against the documented fan-out race — where every parallel worker reads the same pre-batch on-disk cost total and each individually passes the cap check — a single batch can blow past a configured daily_cap_usd budget that the operator believes is enforced. Exploitation requires cost_guardrails enabled with daily_cap_usd set and use of an unpriced model (self-hosted Ollama/vLLM, a new provider SKU, or a steered model argument); the vendor has released 1.1.0 with a fix.
Unauthenticated remote denial-of-service in HPE Networking EdgeConnect SD-WAN Gateways (formerly Silver Peak), affecting firmware branches 9.4.x up to 9.4.8.2, 9.5.x up to 9.5.8.1, 9.6.x up to 9.6.3.1, and 9.7.0.0. An attacker who can reach the vulnerable gateway service can exhaust resources (CWE-400, uncontrolled resource consumption), rendering certain services on the appliance unavailable and disrupting SD-WAN connectivity for sites behind the gateway. The CVSS 3.1 score is 7.0 (AV:N/AC:H/PR:N/UI:N), and no public exploit code or CISA KEV listing was identified at time of analysis.
Unauthenticated attackers with HTTP network reachability to the Forge component of Oracle Commerce Guided Search / Oracle Commerce Experience Manager 11.4.0 can read critical application data, make limited unauthorized modifications, and degrade service availability. Oracle rates exploitation as difficult (CVSS:3.1/AV:N/AC:H/PR:N/UI:N, base 7.0), so the flaw likely depends on timing, specific application state, or non-default conditions rather than a simple one-shot request. There is no public exploit code and no CISA KEV listing at time of analysis, and no EPSS score was included in the supplied intelligence, leaving real-world exploitation likelihood unquantified.
Unauthenticated HTTP requests against Oracle Helidon deployments that use the helidon-dbclient-mongodb component can let a remote attacker read or modify MongoDB-backed data and cause a partial denial of service in versions 3.0.0–3.2.20 and 4.0.0–4.5.4. Oracle rates this CVSS 3.1 base 7.0 with high attack complexity (AV:N/AC:H/PR:N/UI:N), meaning successful exploitation depends on conditions that are difficult for an attacker to arrange. No public exploit code or CISA KEV listing was provided at time of analysis, so this should be treated as a patch-priority issue rather than an emergency.
Oracle GraalVM (including GraalVM for JDK 17/21, GraalVM Enterprise Edition and GraalVM 25.x) contains a flaw in its Compiler component that a network-positioned attacker can reach over HTTP without authenticating. Successful exploitation can expose critical data, allow limited unauthorized modification of data, and cause a partial denial of service; Oracle characterizes the issue as difficult to exploit per the AC:H metric, implying attacker-controlled timing or memory-layout preconditions. No public exploit code has been identified at time of analysis, and the issue is not listed in CISA KEV; the CVSS 3.1 base score is 7.0 with a high confidentiality impact.
Denial of service with possible disclosure of memory contents in IBM MQ (all LTS and CD releases from 9.1.0.0 through 9.4.5.1 and 10.0.0.0) is reachable by a remote attacker who already holds a TLS client certificate the queue manager trusts. Improper validation of deeply nested certificate data during TLS certificate processing produces a stack-based buffer overflow (CWE-121) that can crash the queue manager and potentially expose fragments of process memory. No public exploit code has been identified at time of analysis and the issue is not listed in CISA KEV; IBM has released fixes via its support advisory.
Cross-site request forgery in the Netcore NR255-V wireless router's web management interface (firmware 1.5.130703) allows an attacker to silently rewrite WAN and LAN settings by luring a logged-in administrator to a malicious page, which then fires forged requests at the wan_config_set_cgi, wan_num_set_cgi, and lan_ip_change_cgi endpoints. The supplied CVSS 4.0 vector (7.0, AV:N/AC:L/AT:N/PR:N/UI:A) indicates the attacker does not need credentials of their own but must get an authenticated admin to actively interact, and that the resulting integrity and availability impacts are High while confidentiality is unaffected. No public exploit code was identified in the provided intelligence and the CVE is not listed in CISA KEV, so this is currently a disclosure-driven risk rather than a confirmed campaign.
Local privilege escalation in Oracle Business Intelligence Enterprise Edition 26.01.0.0.0 allows an attacker who already holds a low-privileged logon on the host running the platform to take full control of the BI EE deployment, with high confidentiality, integrity and availability impact across the Platform Security component. Oracle rates the flaw 7.0 (AV:L/AC:H/PR:L/UI:N), reflecting that exploitation is difficult and confined to the local host rather than reachable over the network. No public exploit code or evidence of active exploitation was identified in the supplied intelligence at time of analysis.
Local takeover of Oracle PeopleSoft Enterprise PeopleTools (Cube Manager component) is possible for a low-privileged user who can log on to the infrastructure hosting the product, affecting versions 8.61 through 8.63. Oracle scores this 7.0 (CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H), reflecting a difficult-to-exploit but fully compromising flaw that yields high confidentiality, integrity and availability impact on the PeopleTools instance itself. No public exploit code or CISA KEV listing was identified at time of analysis, so the immediate threat is limited to actors who already hold a valid low-privileged account and interactive access to the PeopleSoft application tier.
Local attackers who can log on to a host running Oracle Application Testing Suite 13.3.0.1 can achieve full takeover of the application, with high impact to confidentiality, integrity and availability. The flaw requires human interaction from someone other than the attacker and is rated high complexity, meaning a foothold on the machine plus a successful social-engineering or timing condition is needed rather than a simple remote request. There is no confirmed CISA KEV listing, no public exploit code identified, and no EPSS score supplied in the source data, so current risk is driven by the local-access precondition and Oracle's own assessment rather than evidence of active exploitation.
Local privilege escalation in Oracle Commerce Guided Search / Oracle Commerce Experience Manager 11.4.0 (Endeca Application Controller component) allows an attacker who already holds a low-privileged logon on the host to fully compromise the application, affecting confidentiality, integrity, and availability. Oracle rates it 7.0 with CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H, meaning the attack requires local access and high complexity rather than remote reachability. No public exploit code has been identified, the CVE is not listed in CISA KEV, and no EPSS score was provided in the source data.
Local privilege escalation in Oracle Hyperion Financial Management 11.2.26.0.000 lets an authenticated attacker who already holds a low-privileged logon on the host running the application seize full control of the HFM instance, with high confidentiality, integrity and availability impact. Oracle rates the attack as difficult (AC:H) and the vector is strictly local (AV:L) with no user interaction, so the exposure is limited to operators of the affected server rather than the network at large. No public exploit code has been identified at time of analysis, and the only supplied reference is Oracle's Critical Security Patch Update advisory.
Let me synthesize this. CVE-2026-76159: Duplicati for Windows before v2.4.0.0, incorrect permission assignment (CWE-732) in configuration loader, allows local low-priv attacker to escalate to SYSTEM via attacker-controlled preload.json. Product name: Duplicati. Tags include "Microsoft" which is likely a mislabeled tag — Duplicati is an independent open-source project (duplicati.com), not Microsoft. Though it runs on Windows... The tag "Microsoft" as vendor is wrong. Emit mislabeled_tag with target "Microsoft". Confidence high? Duplicati is an independent open-source backup tool, not Microsoft. Yes, high confidence. Missing tags? Maybe "windows" — since it's Windows-specific. Hmm, missing_tag target "Windows", suggested "windows". That's unambiguously implied ("Duplicati for Windows", MSI installers). Could emit. But careful — is "windows" a valid tag slug in their taxonomy? They said emit only when unambiguously implied. I'll include it with high confidence... Actually risk: their taxonomy may be "Microsoft Windows". Hmm. I'll include it maybe. Let's be a bit conservative but it seems clearly applicable. I'll include "windows" for missing_tag. Hmm, but if the slug doesn't exist it might attach a bad tag. The instructions say suggested is "lowercase hyphenated slug (e.g. target "WordPress", suggested "wordpress")". So "windows" is fine. Also "privilege-escalation" technique tag is already present as "Privilege Escalation". Fine. CVSS given is 4.0 base with E:X etc. Score 7.0. PR:L, AV:L, UI:P. Hmm, why UI:P? Probably because attacker needs to force the app to load the config... The preload.json is attacker-controlled — likely attacker with write access to a directory that gets loaded. Local low-priv attacker writes preload.json, then triggers service restart (UI:P passive — waiting for service restart/admin action). Anyway. My assessed CVSS 3.1: AV:L/AC:L/PR:L/UI:P? Hmm, UI:P isn't valid in 3.1. UI options: N, R. If user interaction required (someone must rest
An out-of-bounds write reachable through a race condition in google_mba_recv_msg() in google_mba_poll.c allows a locally installed low-privilege application on an affected Android device to corrupt kernel memory and escalate to system/kernel privileges. Rated CVSS 7.0 (AV:L/AC:H/PR:L/UI:N), the flaw requires no user interaction but does need the attacker to already run code on the device and to win a timing window, which keeps it materially below a remote, unauthenticated severity. No public exploit code has been identified and the issue is not listed in CISA KEV at time of analysis; no EPSS score was supplied in the source data.
Memory corruption in the ARM64 TLB-invalidation path of the Android kernel allows a local attacker holding low privileges to win a race condition and escalate to kernel/root privileges, with no user interaction and no additional execution privileges required. Affected builds are Google Android (CPE cpe:2.3:a:google:android:*:*:*:*:*:*:*:*), specifically ARM64 devices; Google's September 2026 Pixel security bulletin is the vendor reference for the fix. No EPSS score, CISA KEV listing, or public proof-of-concept was present in the supplied intelligence, so active exploitation is unconfirmed and no public exploit code has been identified at time of analysis.
Let me build the JSON. CVE-2026-58724: Android kernel use-after-free due to race condition, local EoP requiring System execution privileges, no UI. CVSS 7.0 AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H. Reported by Google Devices. CPE: google:android. Tags: Privilege Escalation, Google, Race Condition. Reference: Android security bulletin September 2026 Pixel. Note: "System execution privileges needed" — this is Android bulletin phrasing meaning the attacker needs System-level privileges already... Actually in Android bulletins, "System execution privileges needed" means the vulnerability requires the attacker to already have System privileges? Hmm, actually Android bulletin says "This could lead to local escalation of privilege with System execution privileges needed" — meaning the attacker needs System execution privileges? Historically it means exploitation requires the attacker to run code with System privileges already... Actually it means the vulnerability can lead to local escalation of privilege with System execution privileges needed — i.e., it requires System-level privileges to exploit. That means privilege escalation from System to... hmm, in Android, "System" is a UID (system uid 1000). So it escalates to kernel. Actually "with System execution privileges needed" means the attacker must already have System privileges. CVSS PR:L is consistent-ish. Actually PR:L. Hmm but if System privileges needed, PR:H might be more apt. Given CVSS says PR:L, I'll assess PR:L... Hmm. My own assessment: AV:L/AC:H/PR:H? Let me think. The Android phrasing "System execution privileges needed" — needed to exploit — implies significant privileges already (system UID, not just any app). That's PR:H arguably. But CVSS gives PR:L. I'll go with PR:H? Let's be careful — I should ground my assessment in the description. The description says "with System execution privileges needed" — this means the exploit requires system-level privileges. So PR:H is honest. But the vendor's own CVSS says
A race condition in the trusty_dputc() debug output path of generic-arm64-smcall.c in the Android kernel allows an out-of-bounds write that can be leveraged for local privilege escalation. The flaw is only reachable by a process already running with System execution privileges, and exploitation requires winning a timing window (AC:H, PR:L), with no user interaction. Impact is high across confidentiality, integrity and availability because the corrupted memory is kernel/secure-monitor adjacent, but no public exploit code or in-the-wild exploitation is known from the available intelligence.
A command injection flaw in the WSO2 Integrator MI extension for Visual Studio Code lets a crafted Micro Integrator project run arbitrary operating system commands when its unit test execution flow is triggered. Exploitation is not remote or silent: the developer must open the untrusted project, explicitly grant it workspace trust, and then run the project's unit tests, after which injected commands execute with the privileges of the VS Code user (CVSS 3.1 base 7.0, AV:L/AC:H/PR:N/UI:R). Impact is therefore bounded by the developer workstation's account privileges, and no public exploit code or active exploitation is indicated in the available intelligence.
Audit log injection in Apache ZooKeeper 3.9.0–3.9.5 and 3.8.0–3.8.6 allows attackers to forge entries in zookeeper_audit.log when audit logging is enabled. An unauthenticated attacker can embed tab characters in a digest authentication username to inject arbitrary tab-separated key=value fields, and a client permitted to call setACL can embed tabs in a digest ACL id to make a successful setAcl event appear to record forged operations such as operation=delete and znode=/forged. The vulnerability corrupts forensic evidence and can defeat SIEM alerting, but does not permit direct modification of ZooKeeper data; no public exploit code or CISA KEV listing was identified at time of analysis, and vendor-released fixes exist in 3.9.6 and 3.8.7.