Denial Of Service
Monthly
Heap exhaustion in the http4s Ember backend (0.23.x before 0.23.35 and 1.0.0 milestones before 1.0.0-M47) allows a single unauthenticated HTTP/2 connection to open an unbounded number of peer-created streams because SETTINGS_MAX_CONCURRENT_STREAMS is never enforced, and each stream retains per-stream state until the JVM heap is exhausted. The same unchecked allocation is reachable on the client side through server-initiated PUSH_PROMISE frames, since enablePush is likewise not honored. Rated CVSS 7.5 (AV:N/AC:L/PR:N/UI:N, availability-only impact) and fixed in 0.23.35 and 1.0.0-M47 via a Semaphore-gated stream handler; no active exploitation is confirmed (not in CISA KEV) and no public exploit code was identified at time of analysis.
A crafted WebSocket frame can pin a worker thread and exhaust memory in the http4s Scala HTTP library, causing denial of service on servers built with the Ember backend. The flaw lies in WebSocket FrameTranscoder.bodyLength, which rejects extended payload lengths above Integer.MAX_VALUE but accepts negative 64-bit lengths; a remote client that completes a WebSocket handshake can send such a frame, and the decoder then returns an empty frame without advancing its input, so the decode loop spins forever while an ArrayBuffer grows without bound. Versions prior to 0.23.35 and 1.0.0-M47 are affected, and both fix releases are available; no public exploit tooling was identified at time of analysis, though the trigger frame is easily reconstructed from the public regression test added in the fix commit.
Heap memory exhaustion in the Ember HTTP/2 backend of http4s allows an unauthenticated remote peer to crash a server built with withHttp2 by sending PING, SETTINGS, or DATA frames faster than the write loop can drain the outbound frame queue. Because outbound frames are serialized through a single unbounded queue (cats.effect.std.Queue.unbounded) prior to versions 0.23.35 and 1.0.0-M47, acknowledgements and WINDOW_UPDATE responses accumulate without limit when the peer simply stops reading, ultimately triggering an OutOfMemoryError. The same unbounded-queue behavior lets an ember-client be attacked by a hostile HTTP/2 server; CVSS rates it 7.5 with availability-only impact, and no public exploit code or active exploitation is identified in the supplied data.
Unbounded WebSocket message buffering in http4s lets any remote client that completes a WebSocket handshake against an http4s-blaze-server or http4s-ember-server endpoint exhaust server memory, causing a denial of service. The shared decoder accepts declared frame lengths up to Int.MaxValue and defragments continuation frames without any size limit, so oversized frames or a long stream of fragments are accumulated in memory until the JVM heap is exhausted. The flaw is addressed in http4s 0.23.35 and 1.0.0-M47, which enforce a configurable 64 MiB default cap on both individual frames and defragmented messages; no public exploit code or CISA KEV entry is present in the supplied intelligence, and no EPSS score was provided.
Heap-exhaustion denial of service in the http4s DigestAuth server middleware: an inverted stale-nonce comparison in NonceKeeper causes fresh nonces to be discarded while eviction halts at the first stale entry, so the nonce map grows without bound. Any application that protects at least one route with DigestAuth and runs a version prior to 0.23.35 or 1.0.0-M47 is affected, and an unauthenticated remote attacker can drive repeated 401 authentication challenges until the JVM throws OutOfMemoryError and the service collapses. CVSS 7.5 (network, no privileges, no user interaction, availability-only) makes this a straightforward resource-exhaustion attack; no public exploit code or CISA KEV entry was identified at time of analysis, but exploitation requires nothing more than a loop of unauthenticated HTTP requests.
{ "product_name":
Unbounded memory consumption in the web_url_read tool of mcp-searxng before 1.7.1 lets a remote unauthenticated client exhaust the MCP server process by pointing it at an HTTP endpoint that omits (or understates) the Content-Length header. The 5 MiB URL_READ_MAX_CONTENT_LENGTH_BYTES guard is implemented as a HEAD preflight only, so when the header is absent the check short-circuits to false and the full response body is loaded via response.text() and then converted to Markdown, driving both memory and CPU usage toward exhaustion. Exploit code and a Docker-based proof of concept are published in the vendor's GHSA advisory, and the flaw is fixed in 1.7.1; CISA KEV listing and EPSS scoring are not present in the supplied data.
A heap buffer overflow in the IBM MQ client (CWE-787) can be triggered when the client processes crafted messages delivered by a malicious or compromised queue manager, or by an attacker positioned to man-in-the-middle the client-to-manager channel. Successful exploitation can crash the client for a denial-of-service condition or, more seriously, allow arbitrary code execution in the context of the client process. IBM has released a vendor fix via support bulletin 7284943; there is no public exploit code or CISA KEV entry identified at time of analysis, and the elevated complexity (the attacker must control the server side of the connection) keeps real-world risk below its 8.1 CVSS base score for most deployments.
IBM Verify Identity Access (formerly Security Verify Access) can be driven into a denial-of-service state by a remote, unauthenticated attacker who sends crafted requests that the product fails to validate for resource consumption. Only availability is impacted (CVSS 7.5, C:N/I:N/A:H); confidentiality and integrity are untouched. No public exploit code and no CISA KEV entry were identified in the provided intelligence, and the vendor has published a fix; the practical risk is therefore tied to exposure of the affected endpoints rather than to confirmed active exploitation.
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.
Unauthenticated remote denial of service in IBM Verify Identity Access (the current name of IBM Security Verify Access) and its container editions, where crafted incoming requests reach a handler that fails to validate resource consumption, leading to uncontrolled recursion (CWE-674) and service exhaustion. CVSS 7.5 (AV:N/AC:L/PR:N/UI:N, availability-only impact) means any network-reachable instance can be knocked offline without credentials, but confidentiality and integrity are unaffected. IBM has published a vendor fix; no public exploit code or CISA KEV entry was present in the supplied intelligence, and no EPSS score was provided.
IBM MQ message queueing middleware fails to properly restrict XML external entity resolution while processing MQRFH2 message headers, letting an authenticated queue user read server-side files or resources and drive the broker and its Java clients into a denial-of-service state. Affected releases span the 9.1.x, 9.2.x, 9.3.x and 9.4.x LTS/CD streams plus 10.0.0.0 Classes for Java, so any environment running the JMS/XMS or Java messaging client against a current broker is in scope. IBM has published a vendor fix and there is no public exploit code or CISA KEV entry identified at time of analysis, but the confidentiality and availability impact are both rated High (CVSS 8.1, AV:N/AC:L/PR:L/UI:N).
We need to produce a JSON object synthesizing data about CVE-2026-12667. Let me parse the data: - CVE ID: CVE-2026-12667 - Description: IBM MQ 9.1.0.0 through 9.1.0.37 LTS, 9.2.0.0 through 9.2.0.43 LTS, 9.3.0.0 through 9.3.0.41 LTS, 9.3.0.0 through 9.3.5.1 CD, 9.4.0.0 through 9.4.0.25 LTS, 9.4.0.0 through 9.4.5.1 CD, and 10.0.0.0 could allow an authenticated attacker to read files from a vulnerable .NET client or cause limited denial of service due to improper handling of XML external entities in RFH2 folder parsing. - CVSS Score: 7.1 - CVSS Vector: CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:L - CWE: CWE-611 (Improper Restriction of XML External Entity Reference) - Patch: Available from vendor - Reported by: ibm - CPE: cpe:2.3:a:ibm:mq:*:*:*:*:*:*:*:* - Tags: Denial Of Service, IBM, XXE, Mq - References: [cveorg] (patch) https://www.ibm.com/support/pages/node/7284936 No EPSS, no KEV, no POC. So exploitation status: no public exploit identified at time of analysis. Product name: IBM MQ (1-3 words). "IBM MQ". Summary: Must not copy/paraphrase description, original synthesis. Start with impact verb and product. Something like: "Authenticated XML external entity (XXE) injection in IBM MQ's .NET client RFH2 folder parsing allows a remote authenticated attacker to read arbitrary files from the client host or trigger a limited denial of service; CVSS 7.1 with high confidentiality impact and low availability impact. IBM has released patches for all affected LTS and CD release streams (9.1 through 10.0), and no public exploit code or active exploitation has been observed." Wait: "read files from a vulnerable .NET client" — so the impact is on the .NET client that parses RFH2 folders. The attacker sends a crafted message with an RFH2 folder containing XML with external entities; when the .NET client parses it, the XXE is triggered. So the attack vector: AV:N, PR:L (authenticated attacker — presumably requires the ability to inject messages into a queue). Impact: file re
Authenticated denial of service in Atlassian Confluence Data Center allows a low-privileged logged-in user to make the application unavailable to its intended users, with no confidentiality or integrity impact. The flaw was introduced across a broad swath of releases (8.9.0 through 10.2.0) and Atlassian rates it High with a CVSS 4.0 score of 7.1. Fixed builds are 9.2.24 and later on the 9.2 branch and 10.2.17 and later on the 10.2 branch; no public exploit code or confirmed active exploitation was identified at time of analysis.
Memory amplification denial of service in the Tornado Python web framework before 6.5.8 lets unauthenticated remote attackers exhaust server memory with a single crafted multipart/form-data request containing tens of thousands of parts — roughly 600KB of body is enough to build a 100,000-element transient list. Tornado's parse_multipart_form_data splits the body on the boundary before it enforces the max_parts limit, so the safety check fires only after the expensive allocation has already happened. Publicly available exploit code exists (a gist with poc.py), patch status is 'available from vendor' with the fix shipped in 6.5.8, and no CISA KEV listing or EPSS score accompanied 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.
Authenticated users of self-hosted Vikunja instances before 2.6.0 can crash the server by uploading a highly compressed archive through the data-import feature, which is expanded without any size or expansion-ratio limit and consumes tens of gigabytes of memory and disk. The issue is a classic decompression-bomb denial of service (CWE-400) rated 7.1 under CVSS 4.0 (AV:N/AC:L/PR:L/UI:N, availability-only impact); any account able to reach the import endpoint is sufficient, no administrative rights are needed. A vendor fix is available in 2.6.0, and no public exploit code or CISA KEV listing was identified in the supplied intelligence.
Vikunja versions before 2.6.0 omit pixel-count decode limits on avatar and project-background image upload endpoints, so an authenticated user can upload tiny files whose decoded dimensions are enormous and drive unbounded CPU and memory consumption during image processing. Repeated or concurrent uploads can exhaust server resources and deny service to all users of the instance. The flaw is rated CVSS 7.1 with high availability impact only, no public exploit code has been identified at time of analysis, and a vendor fix is available in 2.6.0.
An authenticated user of Vikunja before 2.6.0 can exhaust the memory of migration worker processes by launching a Planka import job that points at a server they control, which advertises a large number of individually size-compliant attachments. Because the migrator enforces per-file size limits but no aggregate memory budget, the worker is consumed and the instance becomes unavailable to every user on it. The flaw is rated 7.1 (CVSS 4.0, availability-only impact), requires low privileges but no user interaction, and no public exploit code or CISA KEV entry is associated with it at time of analysis; the vendor has released a fix in 2.6.0.
Vikunja releases before 2.6.0 can be knocked offline by any authenticated user who uploads a deliberately oversized multipart CSV file to the POST /api/v2/migration/csv/migrate endpoint, because the importer parses rows without bounding cardinality. Each tiny record consumes heap, so a file containing millions of minimal rows drives the process to out-of-memory conditions and terminates the API service, causing a full denial of service for all tenants on that instance. The flaw is rated 7.1 (CVSS 4.0) with high availability impact and low privileges required; exploitability is high and automatable, though no public exploit code and no CISA KEV listing were provided at time of analysis. Notes in the source metadata separately cite a vendor-available patch.
Memory exhaustion in the Vikunja task-filter endpoint lets any authenticated user crash the API service by submitting a filter query containing thousands of nested parentheses, which the recursive expression parser consumes without a depth limit. All deployments before version 2.6.0 are affected (EUVD lists 2.5.0 through 2.5.x), and a single low-privileged account is enough to repeatedly take the backend offline. A vendor advisory and a VulnCheck advisory are published, a patched release exists in 2.6.0, and no public exploit code or CISA KEV entry was identified at time of analysis, though the attack is trivial to reproduce by hand with plain HTTP requests.
FreeRDP clients before 3.31.0 terminate unexpectedly when connected to a malicious RDP server, because the URBDRC (USB redirection) control-transfer request path forwards an attacker-supplied OutputBufferSize of 65536 to the libusb backend without validating it, hitting a reachable assertion. The attacker must control the RDP server and convince a user to connect, so exploitation is network-facing but user-initiated. Impact is limited to availability (client process crash) rather than code execution or data exposure; no public exploit code or CISA KEV listing was identified at time of analysis.
Denial of service in FreeRDP clients before 3.31.0: a rogue or compromised RD Gateway can send a WebSocket Ping frame with a crafted 64-bit extended payload length that wraps an integer inside WinPR's Stream_EnsureRemainingCapacity, producing a double free that crashes the client while the connection is being established. Exploitation requires no credentials on the attacker's side and only a victim client that routes its session through that gateway over the WebSocket transport; the primary impact is loss of availability of the client session. Vendor-released patches are available in FreeRDP 3.31.0, and there is no public exploit code or active exploitation reported in the supplied intelligence.
Remote, unauthenticated attackers can crash any FreeRDP-based RDP server running a version prior to 3.31.0 by sending crafted RDP connection packets that supply zero or oversized DesktopWidth and DesktopHeight values during GCC capability negotiation. The malformed dimensions are used unchecked in multifragment update capability calculations, producing a division-by-zero (CWE-369) or assertion failure that terminates the server process. This is a pure availability impact (CVSS 4.0 base 8.2, VC:N/VI:N/VA:H); no publicly available exploit code has been identified and the flaw is not listed in CISA KEV at time of analysis.
FreeRDP clients before 3.31.0 crash when a malicious RDP server sends a crafted Surface Bits command advertising the NSCodec codec ID — the GDI surface-bits handler dereferences a NULL decoder pointer instead of rejecting the unsupported codec. The victim only has to initiate an RDP session to an attacker-controlled or compromised server; the outcome is a client-side denial of service (application crash), with no memory corruption, code execution, or data exposure reported. No public exploit code and no CISA KEV listing were identified at time of analysis.
FreeRDP versions before 3.31.0 contain a heap buffer overflow in the client-side negotiation routine nego_send_negotiation_request(), which writes the server-supplied LB_LOAD_BALANCE_INFO field into a fixed 512-byte buffer without checking its length. A malicious RDP server — or an attacker positioned as a man-in-the-middle — can trigger the overflow by replying to a client's connection attempt with a Server Redirection PDU carrying an oversized LB_LOAD_BALANCE_INFO value, corrupting heap memory before authentication even completes. The confirmed impact is denial of service or heap corruption (CVSS 7.1, availability-only); no public exploit code and no CISA KEV listing were identified at time of analysis.
We need answer with JSON only. Need include "kielce" in risk_assessment only. Need produce fields. Let's parse data. CVE-2026-91952 FreeRDP before 3.31.0 infinite loop DoS in pool_decode_rect when decoding AVC444 metablocks with more region rectangles than preallocated worker array size. Malicious RDP server sends crafted AVC444 graphics updates causing threaded decode path loop indefinitely, CPU, prevents client operation. CVSS 7.1 vector CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:P/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N. CWE-835 Loop with Unreachable Exit Condition ('Infinite Loop'). CPE freerdp:freerdp all versions? * means all? Affected versions before 3.31.0. Tags Denial Of Service, Freerdp. References advisory GHSA and VulnCheck. Need JSON fields as specified. product_name: "FreeRDP" (1 word maybe). summary original synthesis, not start with "A vulnerability" etc. Mention affected versions before 3.31.0, malicious RDP server, AVC444, infinite loop, CPU DoS, no KEV/POC? Input doesn't say POC or KEV. So "no public exploit identified at time of analysis" maybe in summary? Mention EPSS absent? Not provided. KEV not provided. So no active exploitation confirmed. CVSS 7.1. Do not say zero-day. Start with specific impact verb and product: "An infinite-loop denial of service in FreeRDP before 3.31.0 allows a malicious RDP server..." That starts with "An infinite-loop denial of service" — okay not forbidden? It says Do NOT start with 'A vulnerability', 'A problem', 'A flaw', or 'An issue'. "An infinite-loop..." okay. But maybe better start "Malicious RDP servers can freeze FreeRDP clients..." Let's craft. technical_context: explain RDP, FreeRDP, AVC444, H.264/AVC 4:4:4? AVC444 is a remote desktop graphics codec extension, metablocks, region rectangles, worker array preallocation. CWE-835 infinite loop due to no exit when rectangles exceed worker array size. Threaded decode path. Use CPE. risk_assessment: Must use "kielce" once lowercase. Assess CVSS 4.0 vector AV:N AC:L AT:N PR:N UI:
An out-of-bounds write in FreeRDP's urbdrc (USB redirection) client channel allows a malicious or compromised RDP server to crash the connecting client. When the client loads the urbdrc channel, a crafted 28-byte USB redirection message drives urb_send_current_frame_number_result() to write 4 bytes past a 16-byte heap buffer, which trips an assertion (CWE-617) and terminates the process on builds with verbose asserts enabled. The flaw affects FreeRDP 3.14.0 through 3.30.0 and is fixed in 3.31.0; impact is availability only, and no public exploit code or CISA KEV listing has been identified at time of analysis.
Denial of service in Crawl4AI before 0.9.3 lets untrusted clients exhaust disk, CPU and bandwidth on shared worker processes by choosing the PDF scraping strategy in POST crawl requests and pointing it at large remote PDFs that the server downloads without any size or page limits. The flaw is unauthenticated and remotely triggerable (CVSS 4.0 8.7, AV:N/AC:L/PR:N/UI:N) wherever the Crawl4AI HTTP/server mode is reachable by untrusted clients, while deployments that use Crawl4AI purely as an in-process library with operator-controlled URLs are not exposed. No CISA KEV listing, EPSS score or public exploit code was provided in the available intelligence, so active exploitation and exploit maturity are unconfirmed at time of analysis.
Agent-coderag versions up to and including 1.3.0 execute a repository-supplied Gradle wrapper during the default `agent-coderag sync` dependency-discovery flow, letting an attacker who plants a malicious `gradlew` in a Gradle project run arbitrary code with the victim's operating-system privileges. The flaw is a command-injection design error (CWE-78) in `code_rag/core/manager.py`, where a repository-local `gradlew` is unconditionally preferred over the system `gradle` binary and passed to `asyncio.create_subprocess_exec`; CVSS 3.1 base 8.6 with user interaction required (UI:R) because the victim must index the hostile repository. Upstream fixed the issue in 1.3.1 by using only the PATH-resolved system `gradle`; no public exploit code has been identified and the CVE is not in CISA KEV at time of analysis.
Resource exhaustion in Apple macOS allows a remote, unauthenticated attacker to deny service to legitimate users by feeding input that the operating system fails to bound correctly. Apple fixed the flaw with improved input validation in macOS Tahoe 26.7, macOS Sequoia 15.8, and macOS Golden Gate 27. It is rated 7.5 with an availability-only impact (CWE-400), and while no public exploit code or observed exploitation has been identified, the attack is network-reachable and fully automatable per CISA's SSVC assessment.
A use-after-free in Apple macOS kernel memory management allows a locally installed and executed app to escalate to kernel privileges and run arbitrary code with full system control. Affected releases are macOS before Sequoia 15.8, before Tahoe 26.7, and before Golden Gate 27; Apple shipped fixes in all three branches. The flaw is a local privilege escalation (CVSS 7.8, AV:L/PR:L) rather than a remote code execution bug, and there is no public exploit code or CISA KEV listing identified at time of analysis, so real-world exploitation currently depends on an attacker first getting a malicious app onto the target Mac.
Memory exhaustion in adnanh/webhook through 2.8.3 allows unauthenticated remote attackers to crash the service by streaming multi-gigabyte HTTP request bodies at any configured hook endpoint. The daemon reads the entire request body into memory before it evaluates trigger rules or validates the payload signature, so invalid or unsigned requests are just as effective as valid ones at forcing an out-of-memory kill. SSVC rates the issue as automatable and already exploited by proof-of-concept code, though there is no CISA KEV listing and no vendor-issued fix version identified in the available references.
A regular-expression denial-of-service (ReDoS) flaw in DotVVM's routing engine (DotvvmRoute.IsMatch / DotvvmRouteParser.RouteRegex) lets a remote requester send a long, near-matching URL path and force excessive regex backtracking, tying up a worker thread for roughly one second per request. Only applications that declare routes with multiple unconstrained parameters inside a single path segment (e.g. "edit/{a}-{b}-{c}/done") are exposed; fully constrained or slash-separated routes are unaffected. The issue is fixed in DotVVM 4.2.11, 4.3.15 and 5.0.0-preview09-final, which apply a one-second regex timeout and then permanently switch the route to the .NET non-backtracking engine (or return HTTP 503 on runtimes older than .NET 8). There is no CISA KEV entry and no public exploit code identified at time of analysis, and CISA's SSVC assessment rates Exploitation as none and Automatable as no.
We need to produce a JSON object with the specified fields. Let me synthesize. CVE-2026-25687: race condition in ZPA tunnel handler of Zscaler Client Connector (ZCC), heap corruption → DoS (client crash) and potentially RCE in ZCC process context. CVSS 8.1, vector AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H. CWE-366 (Race Condition within a Thread). Patch available from vendor. Reported by Zscaler. CPE: zscaler client_connector. Tags: RCE, Denial Of Service, Client Connector, Zscaler. SSVC: Exploitation none, Automatable no, Technical Impact total. Affected versions: 4.7 <4.7.0.350, 4.6 <4.6.0.486, 4.8 <4.8.0.267, 4.9 <4.9.0.412. Product name: "Zscaler Client Connector" (1-3 words? "Zscaler Client Connector" is 3 words). Good. Summary: Must not start with "A vulnerability" etc. Start with impact verb and product. Something like: "Heap corruption from a race condition in the Zscaler Client Connector (ZCC) ZPA tunnel handler allows remote unauthenticated attackers to crash the client and potentially execute arbitrary code within the ZCC process, affecting versions 4.6 through 4.9 prior to fixed builds. A vendor patch is available; no public exploit or active exploitation has been confirmed (SSVC exploitation: none)." Actually CVSS vector has PR:N, so unauthenticated remote. But note AV:N with AC:H. We can say remote unauthenticated. But the race condition is in the ZPA tunnel handler — likely reachable over network (ZPA tunnel traffic). So remote attackers. Technical context: CWE-366 race condition within a thread — concurrent access to shared state in ZPA tunnel handler. Zscaler Client Connector is endpoint agent that establishes tunnels to Zscaler cloud (ZPA = Zscaler Private Access). The ZPA tunnel handler processes tunnel traffic; race condition leads to heap corruption. CPE indicates all versions of zscaler:client_connector; EUVD gives specific version ranges. Risk assessment: CVSS 8.1 with AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H. High impact but high complexity (race
LangBot's debug WebSocket server binds to all interfaces on port 5401 and ships with an empty authentication key by default - and the official Docker image together with its docker-compose configuration publish that port to the host, placing every default deployment within reach of any network attacker. A remote unauthenticated attacker who reaches the /plugin/ws endpoint can register an arbitrary debug plugin, after which LangBot broadcasts the full plaintext context of every inbound IM message event to that plugin, enabling mass interception of private chats, forged bot replies, LLM pipeline poisoning via malicious tool registration, knowledge-base enumeration, and persistent denial of service by blocking legitimate plugin installations with the prod_mode flag. No patched version was available at time of publication; SSVC classifies exploitation as proof-of-concept-available and fully automatable.
Configuration injection in the Froxlor hosting control panel (all versions before 2.2.5) lets an authenticated customer embed literal newline characters in a subdomain redirect URL, which are written verbatim into nginx or Apache vhost configuration files when the privileged cron job rebuilds them. Because the injected lines become active web server directives affecting other hosted domains on the same server, a low-privileged tenant can corrupt configs, cause denial of service, or hijack HTTP responses across co-hosted sites. This is a CWE-93 CRLF/newline-injection flaw; there is no public exploit identified at time of analysis and it is not listed in CISA KEV.
Memory exhaustion in File Browser through 2.63.23 allows authenticated users with download permission to trigger unbounded server memory consumption via the subtitle conversion endpoint. The endpoint loads entire .srt, .ass, or .ssa files into memory without size constraints, and concurrent requests amplify impact to a full denial of service. A proof-of-concept exists per SSVC data, though no active exploitation has been confirmed and the vulnerability is not listed in the CISA KEV catalog.
Heap memory exhaustion in filebrowser through 2.63.23 allows any authenticated user to crash the server by sending arbitrarily large WebSocket messages to the /api/command endpoint. The server buffers incoming WebSocket frames before performing permission checks, meaning the DoS path is reachable regardless of whether the EnableExec setting is active or the user holds Execute permission - conditions administrators typically rely on to restrict command access. No public exploit code has been identified at time of analysis, and the vulnerability is not listed in CISA KEV.
Stack-based buffer overflow in the camera driver of Samsung Exynos mobile processors (1330, 1380, 1480, 1580, 1680, 2400, 2500, 2600) allows a local low-privileged attacker to crash the camera subsystem or corrupt adjacent kernel memory by sending a malformed message to the driver. The vendor's CVSS vector (S:C, I:H, A:H) indicates that impact extends beyond a simple denial of service to potential kernel-level integrity corruption, though the advisory text describes only DoS - this discrepancy suggests the stack overflow may be exploitable beyond a crash. No public exploit code or active exploitation (KEV) has been identified at time of analysis.
Null pointer dereference in Zephyr RTOS's MQTT-SN client subsystem allows a malicious or on-path network actor to crash any Zephyr device running an active MQTT-SN session by suppressing gateway keepalive responses. The defect is a missing assignment from the SYS_SLIST_PEEK_HEAD_CONTAINER macro in process_ping(): gw is never populated, remains NULL, and is subsequently dereferenced and passed to the memory slab allocator. The outcome is a kernel panic (with CONFIG_MEM_SLAB_POINTER_VALIDATE) or silent slab free-list corruption on targets where address 0 is writable. No public exploit has been identified at time of analysis, and no KEV listing exists; however, the trivial trigger condition and absence of MQTT-SN authentication make this straightforward for any party controlling or impersonating the client's gateway.
Unauthenticated remote attackers can exhaust the heap of the Apache Storm DRPC server by sending requests with arbitrarily many distinct function names, causing a permanent, unbounded memory leak that accumulates until the process is killed by OOM. Affected is org.apache.storm:storm-server versions 3.0.0 through 3.0.x; version 3.1.0 is the fix release. Because drpc.authorizer is unset in the default configuration, no credentials are required to trigger the condition, making this a zero-configuration DoS against any network-reachable DRPC endpoint. No public exploit code has been identified at time of analysis.
Local privilege escalation to root in Apache Storm's setuid-root `worker-launcher` binary (storm-core 3.0.0) allows a tenant user with code execution on a supervisor node to redirect root-owned `lchown` and `chmod` operations to arbitrary host files via a symlink race. The binary uses FTS to walk worker directories and then re-resolves pathnames at syscall time, creating a TOCTOU window exploitable by any topology user who can write to an intermediate directory component. Because worker relaunches and blob updates repeatedly re-trigger the walk, exploitation is low-cost and repeatable; this defeats the isolation boundary that `supervisor.run.worker.as.user` is designed to enforce. No public exploit is identified at time of analysis.
Stack buffer overflow in SIPp through 3.7.7 allows unauthenticated remote attackers to crash the process by sending a crafted SIP message whose To header tag parameter is 2049 bytes or longer. The get_peer_tag() function in sip_parser.cpp copies the tag value into a fixed-size static buffer without a bounds check, enabling the overflow and resulting in a process crash. No public exploit has been identified at time of analysis, but the trigger condition is trivial to reproduce given the precise byte threshold documented by VulnCheck, and an upstream fix has been committed.
Quadratic time complexity in Nodemailer's addressparser component, versions 9.1.0 through 10.0.4, allows remote unauthenticated attackers to block the Node.js event loop for multiple seconds by supplying crafted email addresses containing RFC 5322 comment-separated atoms. The root cause is a `slice(-1)` call on a growing string accumulator in `_handleAddress` that forces JavaScript engine string-flattening on every token, producing O(n²) parse time - empirical test data in the fix commit shows 1.5 MB of crafted input caused approximately 10 seconds of event-loop stall before the patch. No public exploit code has been identified at time of analysis, and the vulnerability is not listed in CISA KEV; however, the attack payload is trivially constructable by simple string repetition and requires no authentication.
Denial of service in UnrealIRCd's embedded webserver (versions 6.0.5-6.2.6) allows unauthenticated remote attackers to exhaust server memory by sending an HTTP request containing an unbounded number of headers, rendering the IRC daemon unresponsive. The attack surface exists only when a WebSocket or JSON-RPC listener is explicitly enabled, both of which are disabled in the default configuration. The issue is resolved in version 6.2.7 per the vendor advisory, and no public exploit code has been identified at time of analysis.
Uncontrolled memory allocation in vLLM's transcription endpoint before 0.28.0 allows authenticated API clients to crash the entire server process by submitting FLAC audio files with forged, inflated sample rate headers. Because vLLM commonly operates as a shared API server across multiple tenants, a single authenticated attacker can deny service to all concurrent users. VulnCheck reported this with a CVSS 4.0 score of 7.1; no public exploit code or CISA KEV listing has been identified at time of analysis.
Memory corruption in ArduinoBLE's ATT layer write request handler allows an unauthenticated BLE client within Bluetooth range to overflow the ATTClass global object's write buffer by sending a crafted ATT Write Request. Versions prior to 2.0.2 are affected exclusively when at least one BLE characteristic is configured with the BLEEncryption property. No public exploit has been identified at time of analysis, but vendor-released patch version 2.0.2 fully resolves the flaw.
Overlayfs in the Linux kernel exposes an unprivileged local DoS - and a theoretical privilege-escalation path - through a double `end_creating()` call in `ovl_create_real()` when a casefold consistency mismatch is detected during directory creation. The double call unlocks the parent directory's `i_rwsem` twice and calls `dput()` twice on the same dentry: the spurious unlock wedges every subsequent creation under that parent indefinitely, and the spurious `dput()` creates a reference-count imbalance that could underpin a use-after-free chain. The flaw is reachable without root in a user namespace on most modern Linux distributions. No public exploit has been identified at time of analysis and EPSS is 0.18%.
An infinite loop in the Linux kernel's device property subsystem causes system hangs when iterating firmware nodes (fwnodes) with a secondary fwnode containing more than one child. Affected are Linux kernel versions across multiple stable branches (5.10.x through 6.19.x pre-patch). An attacker or malformed firmware blob that produces such a node topology can cause a kernel thread to spin indefinitely, consuming a CPU core and ultimately triggering a watchdog reset or complete system hang. No active exploitation has been confirmed, and EPSS stands at 0.17%.
NFSD's seq-lock-like rq_status_counter protocol in nfsd_dispatch() is broken on three non-success exit paths - cache-hit (RC_REPLY), drop (RC_DROPIT/RQ_DROPME), and encode-error - leaving the counter permanently odd and exposing actively-mutating kernel fields to the lockless reader in nfsd_nl_rpc_status_get_dumpit(). A local low-privileged user on an NFS server who can poll NFSD RPC status via the netlink interface while requests traverse these paths can trigger out-of-bounds reads past the 8-element inline ops array embedded in the rqstp structure, disclosing kernel memory or causing a kernel panic. No public exploit exists (EPSS 0.19%), but the 6.12 LTS series appears unpatched, with fixes confined to 7.2.4 and 7.3-rc1.
NULL pointer dereference in Linux kernel NFSD's NFSv4.2 compound dispatch loop allows remote clients to crash NFS servers built with CONFIG_NFSD_V4_2_INTER_SSC. By inserting an operation such as SETATTR between a foreign PUTFH and SAVEFH within a crafted COMPOUND request, an attacker causes fh_want_write() to dereference a NULL fh_export pointer - killing the nfsd kthread and denying NFS service to all clients. No public exploit has been identified and EPSS stands at 0.22%, reflecting low current exploitation interest.
Persistent denial-of-service in the Linux kernel's NFSv4 server grace-period logic allows any reachable NFSv4 client to hold the server in grace indefinitely by sending CLAIM_PREVIOUS OPEN requests. The root cause is a clock-domain mismatch in clients_still_reclaiming(): the deadline is derived from nn->boot_time (CLOCK_REALTIME, ~1.7 billion seconds since epoch) but compared against ktime_get_boottime_seconds() (CLOCK_BOOTTIME, seconds since last boot - orders of magnitude smaller), making the comparison permanently false. As long as any client continues sending CLAIM_PREVIOUS, the grace period never ends, blocking all non-reclaim NFSv4 operations for every other connected client. No public exploit or active exploitation (CISA KEV) has been identified; EPSS is 0.17%.
POSIX ACL reference-count leaks in Linux kernel NFSD allow a remote attacker to exhaust kernel memory via malformed NFSv4 OPEN compound requests. When the NFS server decodes an OPEN compound carrying valid POSIX ACL createhow4 attributes but encounters a non-seqid-mutating error from claim4 decoding, the compound dispatcher short-circuits before nfsd4_open() executes, permanently leaking both op_pacl and op_dpacl references because OP_OPEN lacked an op_release handler; a secondary leak exists on the NFSv4.0 replay path where op_release was never called from nfsd4_encode_replay(). No public exploit code has been identified and EPSS is 0.20%, but repeated malformed requests against a network-reachable NFSv4 server could drive the kernel toward OOM-induced service failure.
Kernel crash in the Linux libceph client allows a malicious or compromised authenticated Ceph OSD to trigger a BUG_ON assertion in ceph_msg_data_next() by sending a crafted sparse-read reply. The flaw in net/ceph/osd_client.c:osd_sparse_read() validates total data length against summed extents but omits per-extent monotonicity, overlap, and bounds checks, enabling a far-forward extent offset to advance the message-data cursor past the request buffer. Impact is limited to availability - a targeted kernel panic on any Linux client mounting a RADOS-backed filesystem or block device (RBD/CephFS). No public exploit code has been identified at time of analysis, and EPSS is low at 0.20%.
Denial of service and stale-memory disclosure in the Linux kernel SMB/CIFS client (cifs.ko) arise from an integer-overflow in symlink_data()'s error-context parsing loop. A previous fix (commit 7d9a7f1f96cd) checked the post-ALIGN() length against the remaining buffer, but ALIGN() itself overflows: an ErrorDataLength near UINT32_MAX (e.g. 0xFFFFFFF9) makes ALIGN(x,8) wrap to 0, so the bounds check passes and the loop advances by zero bytes, spinning indefinitely while 'p' points into stale data. A malicious or MITM SMB server can trigger this against a connecting client; no public exploit identified at time of analysis and EPSS is low (0.21%, 11th percentile).
Out-of-bounds read in the Linux kernel's eCryptfs filesystem packet parser allows a local low-privileged attacker to leak kernel memory or trigger a denial of service by opening a crafted encrypted file with a version 1 header. The root cause is that ecryptfs_parse_packet_set() computes remaining buffer space using a fixed PAGE_SIZE - 8 formula instead of the actual packet-set buffer length, overstating available bytes for version 1 headers where the packet set begins later. Patches are available for stable branches 6.12.109, 6.18.50, 7.2.4, and 7.3-rc1; no public exploit or active exploitation has been identified.
Kernel crash exposure in the OMAP SSI/HSI driver stems from a missing DMA mask initialization on the synthetic HSI controller device allocated via hsi_alloc_controller(), which bypasses the normal OF/platform device setup path and leaves the embedded struct device without a valid dma_mask. Following DMA API hardening changes in the kernel, dma_map_sg() and related helpers enforce the presence of a valid mask before proceeding, causing warnings or kernel panics on any OMAP SoC system where the driver attempts DMA mapping. The vulnerability is tightly hardware-gated to Texas Instruments OMAP-based platforms; patches are confirmed available in kernel stable branches 6.12.109, 6.18.50, 7.2.4, and 7.3-rc1, and no public exploit or KEV listing exists.
Hugetlb reservation accounting in the Linux kernel underflows when a huge-page folio shared between a parent and child process (after fork) is unmapped by the parent first. Since commit df7a6d1f6405 introduced reservation adjustment logic into `__unmap_hugepage_range()`, the function decrements the reservation counter without verifying that the folio's mapcount has reached zero, leaving the child's mapping unaccounted and the counter in an arbitrarily large (underflowed) state until the child unmaps. Affected kernels from 6.9 through the fix commits expose systems using hugetlb-backed workloads to integrity corruption (overcommit of huge pages) and availability risk (kernel memory management instability). No public exploit or active exploitation is identified at time of analysis.
Out-of-bounds read in Linux kernel ACPI APEI GHES ARM hardware error handler arises from using pointer size instead of structure size when accounting for remaining CPER buffer after skipping the cper_sec_proc_arm header. Affected are stable branches 6.12.x through 6.12.109, 6.18.x through 6.18.50, 7.x through 7.2.4, and pre-7.3-rc1, on ARM systems with ACPI APEI support. No public exploit identified at time of analysis; EPSS is 0.20% (10th percentile), and there is no CISA KEV listing, consistent with a low automated exploitation probability.
TRIM/discard is completely broken on 4Kn (4096-byte native sector) block devices in the Linux kernel's libata-scsi layer, causing every ATA DSM TRIM command to fail with 'Parameter list length error' and trigger WARN_ON() splats on each attempt. The bug is in ata_scsi_write_same_xlat(): ata_format_dsm_trim_descr() caps its output at 2048 bytes (ATA_SCSI_RBUF_SIZE), but the caller compares this return value against the logical sector size - on 4Kn devices (sector_size=4096), this equality check always fails and every TRIM is rejected. No public exploit exists; this is a deterministic functional regression on specific hardware. Vendor-released patches are available in stable kernels 6.12.109, 6.18.50, 7.2.4, and 7.3-rc1.
Out-of-bounds read in the Linux kernel Bluetooth subsystem's eir_get_service_data() leaks kernel heap memory into userspace for ISO broadcast sink configurations. When an ISO broadcast sink processes periodic advertising (PA) reports from a remote Bluetooth broadcaster, a PA payload packed with multiple mismatching Service Data fields causes eir_len to undercount by 2 bytes per field iteration - the subtraction uses only the data length (dlen), omitting the 2-byte length+type header - so the pointer eventually reads past hcon->le_per_adv_data[] into the surrounding struct hci_conn; a drifted field matching the BAA UUID causes those bytes to be copied into iso_pi(sk)->base, recoverable by a local or remote-assisted userspace process via getsockopt(BT_ISO_BASE). No public exploit or CISA KEV listing exists; EPSS of 0.17% (6th percentile) reflects low opportunistic risk, though the Bluetooth adjacent-network constraint caps mass exploitation potential.
Remote unauthenticated denial-of-service in AirSane scanner server (versions prior to 0.4.12) allows an attacker to crash the daemon by sending an HTTP POST request with an arbitrarily large Content-Length value, triggering memory exhaustion and a std::bad_alloc exception. The root cause is in httpserver.cpp, where the Content-Length header value is passed directly to std::string::resize() with no upper-bound validation. A secondary variant using non-numeric Content-Length values produces undefined behavior via NaN-to-integer conversion. No public exploit or active exploitation has been identified at time of analysis.
Four coupled error-path defects in the Linux kernel's SUNRPC RPCSEC_GSS option-array XDR decoder (`gssx_dec_option_array`) enable kernel NULL-pointer dereference, a `group_info` refcount leak, and a latent use-after-free during GSS security-context acceptance. Systems acting as NFS servers or clients with Kerberos authentication (sec=krb5/krb5i/krb5p) are in the affected code path; triggering the NULL deref via a failed allocation or malformed GSS response causes a kernel crash (denial-of-service). No public exploit code exists and EPSS sits at the 10th percentile, indicating low current exploitation probability.
Use-after-free in the Linux kernel NFS-over-RDMA server (svcrdma) allows an adjacent-network attacker to trigger kernel memory corruption when an RDMA_CM_EVENT_ADDR_CHANGE event coincides with a replacement listener cm_id allocation failure, leaving sc_cm_id pointing to a destroyed object. Subsequent kernel operations that dereference sc_cm_id - most immediately rdma_disconnect() in svc_rdma_detach() - read or write freed kernel memory, with potential impacts on confidentiality, integrity, and availability. No public exploit or CISA KEV listing exists at time of analysis; EPSS is 0.15% (5th percentile), consistent with the niche NFS-over-RDMA server attack surface.
NULL pointer dereference in the Linux kernel svcrdma subsystem (NFS over RDMA) allows unauthenticated remote clients to crash the kernel by sending a crafted RDMA Read list whose cumulative segment lengths overflow the rq_pages page array. The vulnerability bypasses per-segment length validation by exploiting an edge case where a single segment fills the page array exactly, causing the loop guard to exit prematurely; a subsequent segment then dereferences the NULL sentinel slot at rq_pages[rq_maxpages] during DMA mapping, triggering a kernel panic. No public exploit code or active exploitation (CISA KEV) has been identified, and EPSS is low at 0.21%, but the network-exposed crash vector warrants patching on any host running NFS-over-RDMA.
Race condition in Linux kernel's core scheduler pick_next_task() path allows a local low-privileged attacker to exploit a TOCTOU window when pick_task() releases the rq (run queue) lock mid-selection, potentially leaking information across SMT sibling threads, corrupting forceidle accounting, and destabilizing the scheduler. The vulnerability breaks the cookie-based SMT isolation guarantee of Core Scheduling: an uncookied task can be committed as the sibling of a cookied task during the lock-release window, violating the trust boundary the feature is designed to enforce. No public exploit exists (EPSS 0.19%), and fixes are available in Linux 7.2.4, 7.3-rc1, and 6.19.
Kernel panic via NULL pointer dereference in the Linux kernel qede Ethernet driver (used by QLogic/Marvell FastLinQ NICs) can be triggered under memory pressure conditions during TPA (Transparent Packet Aggregation) fragment processing. A regression introduced by commit 8a8633978b84 omitted the assignment of tpa_info->buffer.data in qede_tpa_start(), causing a 'ghost' buffer descriptor with a NULL data pointer to be reinserted into the active Rx ring upon SKB allocation failure. When hardware subsequently uses the poisoned ring slot, qede_fill_frag_skb() dereferences the NULL pointer, panicking the kernel. No public exploit has been identified at time of analysis, and EPSS is 0.21% (11th percentile).
Memory management order mismatch in the Linux kernel ring-buffer tracing subsystem allows a local low-privileged user to corrupt kernel memory or crash the system by triggering a ring buffer resize concurrently with cached reader page discard. The root cause is that cpu_buffer::free_page stores only a page address without preserving the allocation order; after a resize changes cpu_buffer->subbuf_order, free_pages() is invoked with the new order rather than the original, corrupting the buddy allocator's state. Vendor-released patches exist in Linux 6.18.51, 7.2.4, and 7.3-rc1; no public exploit or CISA KEV listing exists, and EPSS stands at 0.20%.
Kernel memory corruption in the Linux NVMe/TCP host driver (nvme-tcp) allows a malicious or compromised NVMe-over-TCP target to trigger a wild-memory-access write on the connecting host. The receive path gated C2HData PDUs on blk_rq_payload_bytes() alone; for a REQ_OP_WRITE_ZEROES command (no physical segments but non-zero byte count) the request iterator is never initialized, so an attacker-supplied C2HData is copied into whatever stale, driver-private iterator the previous command on that tag left behind. Fixed across stable trees; no public exploit identified at time of analysis and EPSS is low (0.20%).
Integer underflow in the Linux kernel's SCTP stream reconfiguration (RECONF) handler allows a remote SCTP peer to crash the kernel by sending duplicate RECONF response chunks. When a cached RECONF chunk contains an ADD_OUT request alongside other parameters, a duplicate response causes the kernel to roll back `stream->outcnt` twice - the second decrement underflows the unsigned counter and can trigger a kernel panic. EPSS exploitation probability is 0.20% (10th percentile), no public exploit exists, and no CISA KEV listing has been issued, indicating no known active exploitation at time of analysis.
Two compounding flaws in the Linux kernel's Intel Speed Select Technology (ISST) ioctl handler allow a local low-privileged user to trigger out-of-bounds memory access or NULL pointer dereference in kernel space. The `isst_if_clos_assoc()` function applies an off-by-one validator (`>` instead of `>=`) against the `sst_inst[]` array boundary, permitting socket_id equal to `topology_max_packages()` to index one entry past the allocated array; a second path omits a NULL check for valid in-range socket IDs whose package has no bound TPMI SST instance, both dereferences reaching `map_partition_power_domain_id()`. Patched kernel versions 6.12.109, 6.18.50, and 7.2.4 are available; no public exploit has been identified at time of analysis and EPSS (0.20%, 10th percentile) reflects very low observed exploitation activity.
Use-after-free in the Linux kernel's interconnect subsystem allows local low-privileged attackers to corrupt kernel heap memory, potentially enabling privilege escalation to root. The flaw resides in `icc_get()` and `of_icc_get_by_index()`: when `kasprintf()` fails to allocate memory for `path->name`, the error handler calls `kfree(path)` directly while the path's request structures are already linked into interconnect-node hash lists via `hlist_add_head()`, leaving dangling pointers that subsequent interconnect operations dereference. Patched kernel versions 6.12.109, 6.18.50, and 7.2.4 are available; no public exploit or CISA KEV listing exists at time of analysis.
Stack memory disclosure and integer underflow in the Linux kernel's IPMI ipmb subsystem allow a local low-privileged attacker with write access to the IPMB device node to read uninitialized kernel stack bytes and trigger a buffer length underflow in `ipmb_write()`. The root cause is that `ipmb_write()` parses message fields prior to validating the length byte, so a zero-length or undersized write crosses into adjacent stack memory. CVSS rates confidentiality impact as High (stack leak) and availability impact as High (SMBus block-write length underflow), while EPSS at 0.17% reflects the niche attack surface. No public exploit or active exploitation has been identified.
Use-after-free in the Linux kernel's Open vSwitch (OVS) flow deletion path allows a local attacker with low privileges to crash the kernel or potentially escalate privileges. The bug exists in ovs_flow_cmd_del(): when a CMD_DEL netlink command removes a flow from the OVS table, flow->mask is scheduled for RCU-deferred free, but ovs_flow_cmd_fill_info() subsequently reads that pointer after the RCU grace period may have already freed it-a narrow but real race window confirmed by KASAN traces. Reported by Trend Micro's Zero Day Initiative as ZDI-CAN-32042; no public exploit or active exploitation is confirmed at time of analysis.
NULL and error pointer dereference in the Linux kernel's ravb (Renesas Ethernet AVB) driver allows a local attacker with low privileges to crash the kernel or potentially escalate privileges by querying PTP hardware clock state during an invalid lifecycle window. The flaw affects the ravb driver from Linux 4.2 onward, where priv->ptp.clock is either NULL before the first interface open or left holding an error pointer after a failed PTP clock registration. Fixes are confirmed across stable branches in kernels 6.12.109, 6.18.50, 7.2.4, and 7.3-rc1; no public exploit or CISA KEV listing exists.
Out-of-bounds read in the Linux kernel dm-pcache subsystem allows a local low-privileged user to read past segment data boundaries into adjacent control areas, exposing sensitive kernel memory and potentially crashing the kernel. Specifically, cache_replay(), the writeback worker, and the GC worker all miscalculate the tail-kset read length using PCACHE_SEG_SIZE minus the segment offset rather than the actual data-region remainder via cache_seg_remain(). This affects Linux 6.18.x (before 6.18.50), 7.2.x (before 7.2.4), and pre-release 7.3-rc1. No public exploit or active exploitation has been identified; EPSS is 0.17% (6th percentile).
Use-after-free in the Renesas I3C driver allows a local attacker with low privileges to crash the kernel or potentially escalate privileges on systems equipped with Renesas I3C hardware. The flaw stems from a race between a timed-out transfer cleanup path, which frees the transfer struct, and a late-arriving interrupt whose handler dereferences that now-freed memory. No public exploit has been identified, and exploitation requires specific hardware, but the UAF primitive theoretically supports memory corruption beyond a simple crash.
Denial of service in Apache OpenNLP 2.0.0 through 2.5.11 and 3.0.0-M1 through 3.0.0-M5 lets a remote unauthenticated attacker pin CPU or crash a worker thread by feeding crafted text to either of the two built-in RegexNameFinder patterns (EMAIL and URL) obtained via RegexNameFinderFactory.getDefaultRegexNameFinders(). The EMAIL pattern's ambiguous nested quantifiers cause quadratic catastrophic backtracking (~32 KB of local-part characters burns seconds of CPU per find() call), while the URL pattern's nested capturing repetition recurses once per query token, exhausting the thread stack and throwing StackOverflowError (~4 KB, or ~1 KB on -Xss512k worker threads). No public exploit was identified at time of analysis; this is availability-only with no code execution or data exposure.
SymSpellModelSerializer in Apache OpenNLP 3.0.0-M4 and 3.0.0-M5 crashes the host JVM with an OutOfMemoryError when deserializing a crafted SymSpell binary model file whose unigramCount or bigramCount field is set to Integer.MAX_VALUE, forcing a 4-8 GB heap allocation attempt. The root cause (CWE-789: Uncontrolled Memory Allocation) is that these two 32-bit count fields are validated only for non-negativity - no upper bound prevents attacker-controlled pre-sizing of a LinkedHashMap before any dictionary entries are consumed. The attack payload is described by the advisory as under 100 bytes, requires no prior authentication if the target application accepts externally supplied model bytes, and is effective against every deserialization code path in the opennlp-spellcheck extension. No public exploit has been identified at time of analysis, and exposure is substantially limited to the small population running pre-release milestone builds.
Process termination via unrecovered panic in AWS SDK for Go v2's EventStream header decoder allows a remote actor to crash any Go application consuming AWS streaming service responses. Sending a crafted EventStream response frame containing a header value type byte outside the valid range triggers an unhandled Go panic that propagates to terminate the host process. The affected surface includes Bedrock inference and agent runtime packages, making AI/ML workloads on AWS a particularly relevant target; no public exploit or active exploitation has been confirmed.
Use-after-free in Microsoft Edge (Chromium-based) enables an authenticated network attacker to achieve remote code execution with scope change across all supported platforms including Windows, Linux, macOS, Android, and iOS. The vulnerability carries a CVSS 8.5 with high complexity and scope change (S:C), indicating that successful exploitation escapes the browser's sandboxing boundary and affects the underlying host. No public exploit or CISA KEV listing is present at time of analysis, though patch-available status from Microsoft confirms vendor acknowledgment.
Uncontrolled resource consumption in jackson-databind's CoreXMLDeserializers allows unauthenticated remote attackers to trigger multi-minute CPU saturation via a single oversized HTTP request. When jackson-databind deserializes a JSON string into a javax.xml.datatype.Duration or XMLGregorianCalendar field, it passes the raw string directly to JDK's DatatypeFactory, which internally calls BigInteger(String) and BigDecimal(String) constructors that exhibit O(n²) time complexity relative to digit count. Because the malicious digits are embedded in a JSON string token rather than a JSON number token, jackson-core's existing StreamReadConstraints.maxNumberLength guard is bypassed entirely. A few megabytes of input containing millions of digits can consume tens of seconds to minutes of CPU, and a handful of concurrent such requests can exhaust a server's worker thread pool. No public exploit or CISA KEV listing has been identified at time of analysis.
Memory exhaustion DoS in the expressjs/compression Node.js middleware (versions before 1.8.2) allows remote unauthenticated attackers to crash any Express server using the package. When a client aborts an HTTP connection while a compressed response is in flight, the zlib transform stream allocated for that response is never destroyed, permanently leaking its native heap memory. An attacker can automate rapid request-then-abort cycles to exhaust all available server memory, causing process termination. The issue is fixed in 1.8.2 per the OpenJS Foundation advisory; no public exploit code or KEV listing has been identified at time of analysis.
Unauthenticated remote denial of service in Net-SNMP through 5.9.5.2 allows any network-reachable attacker to permanently suspend all SNMP processing by connecting to the SMUX listener and withholding data. The root cause is smux_accept() issuing a blocking recv() call with no timeout on every newly accepted SMUX connection, stalling snmpd's single-threaded main loop indefinitely. Because the block is permanent and snmpd performs all SNMP agent work in that same thread, the entire agent - monitoring, traps, polling - ceases until the process is manually restarted. No public exploit identified at time of analysis, though a researcher gist documenting the behavior is publicly linked in the advisory.
Unauthenticated remote denial-of-service in HPE IceWall Federation Agent and Proxy allows a network attacker to crash or render unavailable the identity federation and authentication proxy service. The root cause is improper handling of incorrect data types (CWE-241), meaning the software fails to gracefully reject or process malformed input, triggering an availability failure. No active exploitation or public exploit code has been identified at time of analysis.
Memory exhaustion in multiparty Node.js library (versions 2.1.0 through <4.3.1) allows any unauthenticated network attacker to crash the host process with a single crafted HTTP request. The parser accumulates multipart part headers into an unbounded in-memory buffer with no size cap, consuming available process memory until the Node.js server is killed. The vulnerability carries no confidentiality or integrity impact - this is a pure availability issue addressed in multiparty 4.3.1, which enforces a maximum on the bytes buffered per part header block.
Denial of service in the RACER swarm-exploration framework (Robotics-STAR-Lab / SYSU STAR Group) lets an attacker who can reach the exploration manager's ROS communication graph drive FastExplorationFSM::optTimerCallback() into an uncontrolled resource-consumption state, stalling or crashing the node and halting multi-robot exploration. The flaw is classified CWE-400 with a CVSS 3.1 base of 7.5 (full availability impact, no confidentiality or integrity impact). No public exploit code or CISA KEV entry was identified in the supplied intelligence, and EPSS is only 0.15% (5th percentile), consistent with the niche academic-robotics deployment of this codebase.
Heap-based buffer overflow in the Sofia IPC daemon's WS-Addressing Action transformation function allows remote unauthenticated attackers to crash Xiongmai XM530 IP cameras or potentially achieve arbitrary code execution by sending a crafted SOAP request containing a wsa5:Action string exceeding 128 bytes. Affected firmware is HMT.CM2005-v220608.1837 and all earlier versions. A public vulnerability research repository exists for this CVE, suggesting exploit code or detailed PoC is available. No public exploit identified as confirmed actively exploited at time of analysis.
A denial-of-service condition in ZJU-FAST-Lab EGO-Planner-v2 (all versions up to commit 5c99a95880401e2599638d567abc0e240396cb42) is triggered by the interaction between the traj_server and poscmd_2_odom nodes and the EGOReplanFSM emergency recovery logic, causing the planner to become unresponsive. The affected software is an open-source trajectory planner for autonomous quadrotors, so a successful denial of service can translate from a crashed planning process directly into loss of autonomous flight control. There is no confirmed evidence of active exploitation (not in CISA KEV) and no public exploit code identified at time of analysis; EPSS is 0.15% (5th percentile), indicating low predicted exploitation activity.
Heap exhaustion in the http4s Ember backend (0.23.x before 0.23.35 and 1.0.0 milestones before 1.0.0-M47) allows a single unauthenticated HTTP/2 connection to open an unbounded number of peer-created streams because SETTINGS_MAX_CONCURRENT_STREAMS is never enforced, and each stream retains per-stream state until the JVM heap is exhausted. The same unchecked allocation is reachable on the client side through server-initiated PUSH_PROMISE frames, since enablePush is likewise not honored. Rated CVSS 7.5 (AV:N/AC:L/PR:N/UI:N, availability-only impact) and fixed in 0.23.35 and 1.0.0-M47 via a Semaphore-gated stream handler; no active exploitation is confirmed (not in CISA KEV) and no public exploit code was identified at time of analysis.
A crafted WebSocket frame can pin a worker thread and exhaust memory in the http4s Scala HTTP library, causing denial of service on servers built with the Ember backend. The flaw lies in WebSocket FrameTranscoder.bodyLength, which rejects extended payload lengths above Integer.MAX_VALUE but accepts negative 64-bit lengths; a remote client that completes a WebSocket handshake can send such a frame, and the decoder then returns an empty frame without advancing its input, so the decode loop spins forever while an ArrayBuffer grows without bound. Versions prior to 0.23.35 and 1.0.0-M47 are affected, and both fix releases are available; no public exploit tooling was identified at time of analysis, though the trigger frame is easily reconstructed from the public regression test added in the fix commit.
Heap memory exhaustion in the Ember HTTP/2 backend of http4s allows an unauthenticated remote peer to crash a server built with withHttp2 by sending PING, SETTINGS, or DATA frames faster than the write loop can drain the outbound frame queue. Because outbound frames are serialized through a single unbounded queue (cats.effect.std.Queue.unbounded) prior to versions 0.23.35 and 1.0.0-M47, acknowledgements and WINDOW_UPDATE responses accumulate without limit when the peer simply stops reading, ultimately triggering an OutOfMemoryError. The same unbounded-queue behavior lets an ember-client be attacked by a hostile HTTP/2 server; CVSS rates it 7.5 with availability-only impact, and no public exploit code or active exploitation is identified in the supplied data.
Unbounded WebSocket message buffering in http4s lets any remote client that completes a WebSocket handshake against an http4s-blaze-server or http4s-ember-server endpoint exhaust server memory, causing a denial of service. The shared decoder accepts declared frame lengths up to Int.MaxValue and defragments continuation frames without any size limit, so oversized frames or a long stream of fragments are accumulated in memory until the JVM heap is exhausted. The flaw is addressed in http4s 0.23.35 and 1.0.0-M47, which enforce a configurable 64 MiB default cap on both individual frames and defragmented messages; no public exploit code or CISA KEV entry is present in the supplied intelligence, and no EPSS score was provided.
Heap-exhaustion denial of service in the http4s DigestAuth server middleware: an inverted stale-nonce comparison in NonceKeeper causes fresh nonces to be discarded while eviction halts at the first stale entry, so the nonce map grows without bound. Any application that protects at least one route with DigestAuth and runs a version prior to 0.23.35 or 1.0.0-M47 is affected, and an unauthenticated remote attacker can drive repeated 401 authentication challenges until the JVM throws OutOfMemoryError and the service collapses. CVSS 7.5 (network, no privileges, no user interaction, availability-only) makes this a straightforward resource-exhaustion attack; no public exploit code or CISA KEV entry was identified at time of analysis, but exploitation requires nothing more than a loop of unauthenticated HTTP requests.
{ "product_name":
Unbounded memory consumption in the web_url_read tool of mcp-searxng before 1.7.1 lets a remote unauthenticated client exhaust the MCP server process by pointing it at an HTTP endpoint that omits (or understates) the Content-Length header. The 5 MiB URL_READ_MAX_CONTENT_LENGTH_BYTES guard is implemented as a HEAD preflight only, so when the header is absent the check short-circuits to false and the full response body is loaded via response.text() and then converted to Markdown, driving both memory and CPU usage toward exhaustion. Exploit code and a Docker-based proof of concept are published in the vendor's GHSA advisory, and the flaw is fixed in 1.7.1; CISA KEV listing and EPSS scoring are not present in the supplied data.
A heap buffer overflow in the IBM MQ client (CWE-787) can be triggered when the client processes crafted messages delivered by a malicious or compromised queue manager, or by an attacker positioned to man-in-the-middle the client-to-manager channel. Successful exploitation can crash the client for a denial-of-service condition or, more seriously, allow arbitrary code execution in the context of the client process. IBM has released a vendor fix via support bulletin 7284943; there is no public exploit code or CISA KEV entry identified at time of analysis, and the elevated complexity (the attacker must control the server side of the connection) keeps real-world risk below its 8.1 CVSS base score for most deployments.
IBM Verify Identity Access (formerly Security Verify Access) can be driven into a denial-of-service state by a remote, unauthenticated attacker who sends crafted requests that the product fails to validate for resource consumption. Only availability is impacted (CVSS 7.5, C:N/I:N/A:H); confidentiality and integrity are untouched. No public exploit code and no CISA KEV entry were identified in the provided intelligence, and the vendor has published a fix; the practical risk is therefore tied to exposure of the affected endpoints rather than to confirmed active exploitation.
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.
Unauthenticated remote denial of service in IBM Verify Identity Access (the current name of IBM Security Verify Access) and its container editions, where crafted incoming requests reach a handler that fails to validate resource consumption, leading to uncontrolled recursion (CWE-674) and service exhaustion. CVSS 7.5 (AV:N/AC:L/PR:N/UI:N, availability-only impact) means any network-reachable instance can be knocked offline without credentials, but confidentiality and integrity are unaffected. IBM has published a vendor fix; no public exploit code or CISA KEV entry was present in the supplied intelligence, and no EPSS score was provided.
IBM MQ message queueing middleware fails to properly restrict XML external entity resolution while processing MQRFH2 message headers, letting an authenticated queue user read server-side files or resources and drive the broker and its Java clients into a denial-of-service state. Affected releases span the 9.1.x, 9.2.x, 9.3.x and 9.4.x LTS/CD streams plus 10.0.0.0 Classes for Java, so any environment running the JMS/XMS or Java messaging client against a current broker is in scope. IBM has published a vendor fix and there is no public exploit code or CISA KEV entry identified at time of analysis, but the confidentiality and availability impact are both rated High (CVSS 8.1, AV:N/AC:L/PR:L/UI:N).
We need to produce a JSON object synthesizing data about CVE-2026-12667. Let me parse the data: - CVE ID: CVE-2026-12667 - Description: IBM MQ 9.1.0.0 through 9.1.0.37 LTS, 9.2.0.0 through 9.2.0.43 LTS, 9.3.0.0 through 9.3.0.41 LTS, 9.3.0.0 through 9.3.5.1 CD, 9.4.0.0 through 9.4.0.25 LTS, 9.4.0.0 through 9.4.5.1 CD, and 10.0.0.0 could allow an authenticated attacker to read files from a vulnerable .NET client or cause limited denial of service due to improper handling of XML external entities in RFH2 folder parsing. - CVSS Score: 7.1 - CVSS Vector: CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:L - CWE: CWE-611 (Improper Restriction of XML External Entity Reference) - Patch: Available from vendor - Reported by: ibm - CPE: cpe:2.3:a:ibm:mq:*:*:*:*:*:*:*:* - Tags: Denial Of Service, IBM, XXE, Mq - References: [cveorg] (patch) https://www.ibm.com/support/pages/node/7284936 No EPSS, no KEV, no POC. So exploitation status: no public exploit identified at time of analysis. Product name: IBM MQ (1-3 words). "IBM MQ". Summary: Must not copy/paraphrase description, original synthesis. Start with impact verb and product. Something like: "Authenticated XML external entity (XXE) injection in IBM MQ's .NET client RFH2 folder parsing allows a remote authenticated attacker to read arbitrary files from the client host or trigger a limited denial of service; CVSS 7.1 with high confidentiality impact and low availability impact. IBM has released patches for all affected LTS and CD release streams (9.1 through 10.0), and no public exploit code or active exploitation has been observed." Wait: "read files from a vulnerable .NET client" — so the impact is on the .NET client that parses RFH2 folders. The attacker sends a crafted message with an RFH2 folder containing XML with external entities; when the .NET client parses it, the XXE is triggered. So the attack vector: AV:N, PR:L (authenticated attacker — presumably requires the ability to inject messages into a queue). Impact: file re
Authenticated denial of service in Atlassian Confluence Data Center allows a low-privileged logged-in user to make the application unavailable to its intended users, with no confidentiality or integrity impact. The flaw was introduced across a broad swath of releases (8.9.0 through 10.2.0) and Atlassian rates it High with a CVSS 4.0 score of 7.1. Fixed builds are 9.2.24 and later on the 9.2 branch and 10.2.17 and later on the 10.2 branch; no public exploit code or confirmed active exploitation was identified at time of analysis.
Memory amplification denial of service in the Tornado Python web framework before 6.5.8 lets unauthenticated remote attackers exhaust server memory with a single crafted multipart/form-data request containing tens of thousands of parts — roughly 600KB of body is enough to build a 100,000-element transient list. Tornado's parse_multipart_form_data splits the body on the boundary before it enforces the max_parts limit, so the safety check fires only after the expensive allocation has already happened. Publicly available exploit code exists (a gist with poc.py), patch status is 'available from vendor' with the fix shipped in 6.5.8, and no CISA KEV listing or EPSS score accompanied 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.
Authenticated users of self-hosted Vikunja instances before 2.6.0 can crash the server by uploading a highly compressed archive through the data-import feature, which is expanded without any size or expansion-ratio limit and consumes tens of gigabytes of memory and disk. The issue is a classic decompression-bomb denial of service (CWE-400) rated 7.1 under CVSS 4.0 (AV:N/AC:L/PR:L/UI:N, availability-only impact); any account able to reach the import endpoint is sufficient, no administrative rights are needed. A vendor fix is available in 2.6.0, and no public exploit code or CISA KEV listing was identified in the supplied intelligence.
Vikunja versions before 2.6.0 omit pixel-count decode limits on avatar and project-background image upload endpoints, so an authenticated user can upload tiny files whose decoded dimensions are enormous and drive unbounded CPU and memory consumption during image processing. Repeated or concurrent uploads can exhaust server resources and deny service to all users of the instance. The flaw is rated CVSS 7.1 with high availability impact only, no public exploit code has been identified at time of analysis, and a vendor fix is available in 2.6.0.
An authenticated user of Vikunja before 2.6.0 can exhaust the memory of migration worker processes by launching a Planka import job that points at a server they control, which advertises a large number of individually size-compliant attachments. Because the migrator enforces per-file size limits but no aggregate memory budget, the worker is consumed and the instance becomes unavailable to every user on it. The flaw is rated 7.1 (CVSS 4.0, availability-only impact), requires low privileges but no user interaction, and no public exploit code or CISA KEV entry is associated with it at time of analysis; the vendor has released a fix in 2.6.0.
Vikunja releases before 2.6.0 can be knocked offline by any authenticated user who uploads a deliberately oversized multipart CSV file to the POST /api/v2/migration/csv/migrate endpoint, because the importer parses rows without bounding cardinality. Each tiny record consumes heap, so a file containing millions of minimal rows drives the process to out-of-memory conditions and terminates the API service, causing a full denial of service for all tenants on that instance. The flaw is rated 7.1 (CVSS 4.0) with high availability impact and low privileges required; exploitability is high and automatable, though no public exploit code and no CISA KEV listing were provided at time of analysis. Notes in the source metadata separately cite a vendor-available patch.
Memory exhaustion in the Vikunja task-filter endpoint lets any authenticated user crash the API service by submitting a filter query containing thousands of nested parentheses, which the recursive expression parser consumes without a depth limit. All deployments before version 2.6.0 are affected (EUVD lists 2.5.0 through 2.5.x), and a single low-privileged account is enough to repeatedly take the backend offline. A vendor advisory and a VulnCheck advisory are published, a patched release exists in 2.6.0, and no public exploit code or CISA KEV entry was identified at time of analysis, though the attack is trivial to reproduce by hand with plain HTTP requests.
FreeRDP clients before 3.31.0 terminate unexpectedly when connected to a malicious RDP server, because the URBDRC (USB redirection) control-transfer request path forwards an attacker-supplied OutputBufferSize of 65536 to the libusb backend without validating it, hitting a reachable assertion. The attacker must control the RDP server and convince a user to connect, so exploitation is network-facing but user-initiated. Impact is limited to availability (client process crash) rather than code execution or data exposure; no public exploit code or CISA KEV listing was identified at time of analysis.
Denial of service in FreeRDP clients before 3.31.0: a rogue or compromised RD Gateway can send a WebSocket Ping frame with a crafted 64-bit extended payload length that wraps an integer inside WinPR's Stream_EnsureRemainingCapacity, producing a double free that crashes the client while the connection is being established. Exploitation requires no credentials on the attacker's side and only a victim client that routes its session through that gateway over the WebSocket transport; the primary impact is loss of availability of the client session. Vendor-released patches are available in FreeRDP 3.31.0, and there is no public exploit code or active exploitation reported in the supplied intelligence.
Remote, unauthenticated attackers can crash any FreeRDP-based RDP server running a version prior to 3.31.0 by sending crafted RDP connection packets that supply zero or oversized DesktopWidth and DesktopHeight values during GCC capability negotiation. The malformed dimensions are used unchecked in multifragment update capability calculations, producing a division-by-zero (CWE-369) or assertion failure that terminates the server process. This is a pure availability impact (CVSS 4.0 base 8.2, VC:N/VI:N/VA:H); no publicly available exploit code has been identified and the flaw is not listed in CISA KEV at time of analysis.
FreeRDP clients before 3.31.0 crash when a malicious RDP server sends a crafted Surface Bits command advertising the NSCodec codec ID — the GDI surface-bits handler dereferences a NULL decoder pointer instead of rejecting the unsupported codec. The victim only has to initiate an RDP session to an attacker-controlled or compromised server; the outcome is a client-side denial of service (application crash), with no memory corruption, code execution, or data exposure reported. No public exploit code and no CISA KEV listing were identified at time of analysis.
FreeRDP versions before 3.31.0 contain a heap buffer overflow in the client-side negotiation routine nego_send_negotiation_request(), which writes the server-supplied LB_LOAD_BALANCE_INFO field into a fixed 512-byte buffer without checking its length. A malicious RDP server — or an attacker positioned as a man-in-the-middle — can trigger the overflow by replying to a client's connection attempt with a Server Redirection PDU carrying an oversized LB_LOAD_BALANCE_INFO value, corrupting heap memory before authentication even completes. The confirmed impact is denial of service or heap corruption (CVSS 7.1, availability-only); no public exploit code and no CISA KEV listing were identified at time of analysis.
We need answer with JSON only. Need include "kielce" in risk_assessment only. Need produce fields. Let's parse data. CVE-2026-91952 FreeRDP before 3.31.0 infinite loop DoS in pool_decode_rect when decoding AVC444 metablocks with more region rectangles than preallocated worker array size. Malicious RDP server sends crafted AVC444 graphics updates causing threaded decode path loop indefinitely, CPU, prevents client operation. CVSS 7.1 vector CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:P/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N. CWE-835 Loop with Unreachable Exit Condition ('Infinite Loop'). CPE freerdp:freerdp all versions? * means all? Affected versions before 3.31.0. Tags Denial Of Service, Freerdp. References advisory GHSA and VulnCheck. Need JSON fields as specified. product_name: "FreeRDP" (1 word maybe). summary original synthesis, not start with "A vulnerability" etc. Mention affected versions before 3.31.0, malicious RDP server, AVC444, infinite loop, CPU DoS, no KEV/POC? Input doesn't say POC or KEV. So "no public exploit identified at time of analysis" maybe in summary? Mention EPSS absent? Not provided. KEV not provided. So no active exploitation confirmed. CVSS 7.1. Do not say zero-day. Start with specific impact verb and product: "An infinite-loop denial of service in FreeRDP before 3.31.0 allows a malicious RDP server..." That starts with "An infinite-loop denial of service" — okay not forbidden? It says Do NOT start with 'A vulnerability', 'A problem', 'A flaw', or 'An issue'. "An infinite-loop..." okay. But maybe better start "Malicious RDP servers can freeze FreeRDP clients..." Let's craft. technical_context: explain RDP, FreeRDP, AVC444, H.264/AVC 4:4:4? AVC444 is a remote desktop graphics codec extension, metablocks, region rectangles, worker array preallocation. CWE-835 infinite loop due to no exit when rectangles exceed worker array size. Threaded decode path. Use CPE. risk_assessment: Must use "kielce" once lowercase. Assess CVSS 4.0 vector AV:N AC:L AT:N PR:N UI:
An out-of-bounds write in FreeRDP's urbdrc (USB redirection) client channel allows a malicious or compromised RDP server to crash the connecting client. When the client loads the urbdrc channel, a crafted 28-byte USB redirection message drives urb_send_current_frame_number_result() to write 4 bytes past a 16-byte heap buffer, which trips an assertion (CWE-617) and terminates the process on builds with verbose asserts enabled. The flaw affects FreeRDP 3.14.0 through 3.30.0 and is fixed in 3.31.0; impact is availability only, and no public exploit code or CISA KEV listing has been identified at time of analysis.
Denial of service in Crawl4AI before 0.9.3 lets untrusted clients exhaust disk, CPU and bandwidth on shared worker processes by choosing the PDF scraping strategy in POST crawl requests and pointing it at large remote PDFs that the server downloads without any size or page limits. The flaw is unauthenticated and remotely triggerable (CVSS 4.0 8.7, AV:N/AC:L/PR:N/UI:N) wherever the Crawl4AI HTTP/server mode is reachable by untrusted clients, while deployments that use Crawl4AI purely as an in-process library with operator-controlled URLs are not exposed. No CISA KEV listing, EPSS score or public exploit code was provided in the available intelligence, so active exploitation and exploit maturity are unconfirmed at time of analysis.
Agent-coderag versions up to and including 1.3.0 execute a repository-supplied Gradle wrapper during the default `agent-coderag sync` dependency-discovery flow, letting an attacker who plants a malicious `gradlew` in a Gradle project run arbitrary code with the victim's operating-system privileges. The flaw is a command-injection design error (CWE-78) in `code_rag/core/manager.py`, where a repository-local `gradlew` is unconditionally preferred over the system `gradle` binary and passed to `asyncio.create_subprocess_exec`; CVSS 3.1 base 8.6 with user interaction required (UI:R) because the victim must index the hostile repository. Upstream fixed the issue in 1.3.1 by using only the PATH-resolved system `gradle`; no public exploit code has been identified and the CVE is not in CISA KEV at time of analysis.
Resource exhaustion in Apple macOS allows a remote, unauthenticated attacker to deny service to legitimate users by feeding input that the operating system fails to bound correctly. Apple fixed the flaw with improved input validation in macOS Tahoe 26.7, macOS Sequoia 15.8, and macOS Golden Gate 27. It is rated 7.5 with an availability-only impact (CWE-400), and while no public exploit code or observed exploitation has been identified, the attack is network-reachable and fully automatable per CISA's SSVC assessment.
A use-after-free in Apple macOS kernel memory management allows a locally installed and executed app to escalate to kernel privileges and run arbitrary code with full system control. Affected releases are macOS before Sequoia 15.8, before Tahoe 26.7, and before Golden Gate 27; Apple shipped fixes in all three branches. The flaw is a local privilege escalation (CVSS 7.8, AV:L/PR:L) rather than a remote code execution bug, and there is no public exploit code or CISA KEV listing identified at time of analysis, so real-world exploitation currently depends on an attacker first getting a malicious app onto the target Mac.
Memory exhaustion in adnanh/webhook through 2.8.3 allows unauthenticated remote attackers to crash the service by streaming multi-gigabyte HTTP request bodies at any configured hook endpoint. The daemon reads the entire request body into memory before it evaluates trigger rules or validates the payload signature, so invalid or unsigned requests are just as effective as valid ones at forcing an out-of-memory kill. SSVC rates the issue as automatable and already exploited by proof-of-concept code, though there is no CISA KEV listing and no vendor-issued fix version identified in the available references.
A regular-expression denial-of-service (ReDoS) flaw in DotVVM's routing engine (DotvvmRoute.IsMatch / DotvvmRouteParser.RouteRegex) lets a remote requester send a long, near-matching URL path and force excessive regex backtracking, tying up a worker thread for roughly one second per request. Only applications that declare routes with multiple unconstrained parameters inside a single path segment (e.g. "edit/{a}-{b}-{c}/done") are exposed; fully constrained or slash-separated routes are unaffected. The issue is fixed in DotVVM 4.2.11, 4.3.15 and 5.0.0-preview09-final, which apply a one-second regex timeout and then permanently switch the route to the .NET non-backtracking engine (or return HTTP 503 on runtimes older than .NET 8). There is no CISA KEV entry and no public exploit code identified at time of analysis, and CISA's SSVC assessment rates Exploitation as none and Automatable as no.
We need to produce a JSON object with the specified fields. Let me synthesize. CVE-2026-25687: race condition in ZPA tunnel handler of Zscaler Client Connector (ZCC), heap corruption → DoS (client crash) and potentially RCE in ZCC process context. CVSS 8.1, vector AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H. CWE-366 (Race Condition within a Thread). Patch available from vendor. Reported by Zscaler. CPE: zscaler client_connector. Tags: RCE, Denial Of Service, Client Connector, Zscaler. SSVC: Exploitation none, Automatable no, Technical Impact total. Affected versions: 4.7 <4.7.0.350, 4.6 <4.6.0.486, 4.8 <4.8.0.267, 4.9 <4.9.0.412. Product name: "Zscaler Client Connector" (1-3 words? "Zscaler Client Connector" is 3 words). Good. Summary: Must not start with "A vulnerability" etc. Start with impact verb and product. Something like: "Heap corruption from a race condition in the Zscaler Client Connector (ZCC) ZPA tunnel handler allows remote unauthenticated attackers to crash the client and potentially execute arbitrary code within the ZCC process, affecting versions 4.6 through 4.9 prior to fixed builds. A vendor patch is available; no public exploit or active exploitation has been confirmed (SSVC exploitation: none)." Actually CVSS vector has PR:N, so unauthenticated remote. But note AV:N with AC:H. We can say remote unauthenticated. But the race condition is in the ZPA tunnel handler — likely reachable over network (ZPA tunnel traffic). So remote attackers. Technical context: CWE-366 race condition within a thread — concurrent access to shared state in ZPA tunnel handler. Zscaler Client Connector is endpoint agent that establishes tunnels to Zscaler cloud (ZPA = Zscaler Private Access). The ZPA tunnel handler processes tunnel traffic; race condition leads to heap corruption. CPE indicates all versions of zscaler:client_connector; EUVD gives specific version ranges. Risk assessment: CVSS 8.1 with AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H. High impact but high complexity (race
LangBot's debug WebSocket server binds to all interfaces on port 5401 and ships with an empty authentication key by default - and the official Docker image together with its docker-compose configuration publish that port to the host, placing every default deployment within reach of any network attacker. A remote unauthenticated attacker who reaches the /plugin/ws endpoint can register an arbitrary debug plugin, after which LangBot broadcasts the full plaintext context of every inbound IM message event to that plugin, enabling mass interception of private chats, forged bot replies, LLM pipeline poisoning via malicious tool registration, knowledge-base enumeration, and persistent denial of service by blocking legitimate plugin installations with the prod_mode flag. No patched version was available at time of publication; SSVC classifies exploitation as proof-of-concept-available and fully automatable.
Configuration injection in the Froxlor hosting control panel (all versions before 2.2.5) lets an authenticated customer embed literal newline characters in a subdomain redirect URL, which are written verbatim into nginx or Apache vhost configuration files when the privileged cron job rebuilds them. Because the injected lines become active web server directives affecting other hosted domains on the same server, a low-privileged tenant can corrupt configs, cause denial of service, or hijack HTTP responses across co-hosted sites. This is a CWE-93 CRLF/newline-injection flaw; there is no public exploit identified at time of analysis and it is not listed in CISA KEV.
Memory exhaustion in File Browser through 2.63.23 allows authenticated users with download permission to trigger unbounded server memory consumption via the subtitle conversion endpoint. The endpoint loads entire .srt, .ass, or .ssa files into memory without size constraints, and concurrent requests amplify impact to a full denial of service. A proof-of-concept exists per SSVC data, though no active exploitation has been confirmed and the vulnerability is not listed in the CISA KEV catalog.
Heap memory exhaustion in filebrowser through 2.63.23 allows any authenticated user to crash the server by sending arbitrarily large WebSocket messages to the /api/command endpoint. The server buffers incoming WebSocket frames before performing permission checks, meaning the DoS path is reachable regardless of whether the EnableExec setting is active or the user holds Execute permission - conditions administrators typically rely on to restrict command access. No public exploit code has been identified at time of analysis, and the vulnerability is not listed in CISA KEV.
Stack-based buffer overflow in the camera driver of Samsung Exynos mobile processors (1330, 1380, 1480, 1580, 1680, 2400, 2500, 2600) allows a local low-privileged attacker to crash the camera subsystem or corrupt adjacent kernel memory by sending a malformed message to the driver. The vendor's CVSS vector (S:C, I:H, A:H) indicates that impact extends beyond a simple denial of service to potential kernel-level integrity corruption, though the advisory text describes only DoS - this discrepancy suggests the stack overflow may be exploitable beyond a crash. No public exploit code or active exploitation (KEV) has been identified at time of analysis.
Null pointer dereference in Zephyr RTOS's MQTT-SN client subsystem allows a malicious or on-path network actor to crash any Zephyr device running an active MQTT-SN session by suppressing gateway keepalive responses. The defect is a missing assignment from the SYS_SLIST_PEEK_HEAD_CONTAINER macro in process_ping(): gw is never populated, remains NULL, and is subsequently dereferenced and passed to the memory slab allocator. The outcome is a kernel panic (with CONFIG_MEM_SLAB_POINTER_VALIDATE) or silent slab free-list corruption on targets where address 0 is writable. No public exploit has been identified at time of analysis, and no KEV listing exists; however, the trivial trigger condition and absence of MQTT-SN authentication make this straightforward for any party controlling or impersonating the client's gateway.
Unauthenticated remote attackers can exhaust the heap of the Apache Storm DRPC server by sending requests with arbitrarily many distinct function names, causing a permanent, unbounded memory leak that accumulates until the process is killed by OOM. Affected is org.apache.storm:storm-server versions 3.0.0 through 3.0.x; version 3.1.0 is the fix release. Because drpc.authorizer is unset in the default configuration, no credentials are required to trigger the condition, making this a zero-configuration DoS against any network-reachable DRPC endpoint. No public exploit code has been identified at time of analysis.
Local privilege escalation to root in Apache Storm's setuid-root `worker-launcher` binary (storm-core 3.0.0) allows a tenant user with code execution on a supervisor node to redirect root-owned `lchown` and `chmod` operations to arbitrary host files via a symlink race. The binary uses FTS to walk worker directories and then re-resolves pathnames at syscall time, creating a TOCTOU window exploitable by any topology user who can write to an intermediate directory component. Because worker relaunches and blob updates repeatedly re-trigger the walk, exploitation is low-cost and repeatable; this defeats the isolation boundary that `supervisor.run.worker.as.user` is designed to enforce. No public exploit is identified at time of analysis.
Stack buffer overflow in SIPp through 3.7.7 allows unauthenticated remote attackers to crash the process by sending a crafted SIP message whose To header tag parameter is 2049 bytes or longer. The get_peer_tag() function in sip_parser.cpp copies the tag value into a fixed-size static buffer without a bounds check, enabling the overflow and resulting in a process crash. No public exploit has been identified at time of analysis, but the trigger condition is trivial to reproduce given the precise byte threshold documented by VulnCheck, and an upstream fix has been committed.
Quadratic time complexity in Nodemailer's addressparser component, versions 9.1.0 through 10.0.4, allows remote unauthenticated attackers to block the Node.js event loop for multiple seconds by supplying crafted email addresses containing RFC 5322 comment-separated atoms. The root cause is a `slice(-1)` call on a growing string accumulator in `_handleAddress` that forces JavaScript engine string-flattening on every token, producing O(n²) parse time - empirical test data in the fix commit shows 1.5 MB of crafted input caused approximately 10 seconds of event-loop stall before the patch. No public exploit code has been identified at time of analysis, and the vulnerability is not listed in CISA KEV; however, the attack payload is trivially constructable by simple string repetition and requires no authentication.
Denial of service in UnrealIRCd's embedded webserver (versions 6.0.5-6.2.6) allows unauthenticated remote attackers to exhaust server memory by sending an HTTP request containing an unbounded number of headers, rendering the IRC daemon unresponsive. The attack surface exists only when a WebSocket or JSON-RPC listener is explicitly enabled, both of which are disabled in the default configuration. The issue is resolved in version 6.2.7 per the vendor advisory, and no public exploit code has been identified at time of analysis.
Uncontrolled memory allocation in vLLM's transcription endpoint before 0.28.0 allows authenticated API clients to crash the entire server process by submitting FLAC audio files with forged, inflated sample rate headers. Because vLLM commonly operates as a shared API server across multiple tenants, a single authenticated attacker can deny service to all concurrent users. VulnCheck reported this with a CVSS 4.0 score of 7.1; no public exploit code or CISA KEV listing has been identified at time of analysis.
Memory corruption in ArduinoBLE's ATT layer write request handler allows an unauthenticated BLE client within Bluetooth range to overflow the ATTClass global object's write buffer by sending a crafted ATT Write Request. Versions prior to 2.0.2 are affected exclusively when at least one BLE characteristic is configured with the BLEEncryption property. No public exploit has been identified at time of analysis, but vendor-released patch version 2.0.2 fully resolves the flaw.
Overlayfs in the Linux kernel exposes an unprivileged local DoS - and a theoretical privilege-escalation path - through a double `end_creating()` call in `ovl_create_real()` when a casefold consistency mismatch is detected during directory creation. The double call unlocks the parent directory's `i_rwsem` twice and calls `dput()` twice on the same dentry: the spurious unlock wedges every subsequent creation under that parent indefinitely, and the spurious `dput()` creates a reference-count imbalance that could underpin a use-after-free chain. The flaw is reachable without root in a user namespace on most modern Linux distributions. No public exploit has been identified at time of analysis and EPSS is 0.18%.
An infinite loop in the Linux kernel's device property subsystem causes system hangs when iterating firmware nodes (fwnodes) with a secondary fwnode containing more than one child. Affected are Linux kernel versions across multiple stable branches (5.10.x through 6.19.x pre-patch). An attacker or malformed firmware blob that produces such a node topology can cause a kernel thread to spin indefinitely, consuming a CPU core and ultimately triggering a watchdog reset or complete system hang. No active exploitation has been confirmed, and EPSS stands at 0.17%.
NFSD's seq-lock-like rq_status_counter protocol in nfsd_dispatch() is broken on three non-success exit paths - cache-hit (RC_REPLY), drop (RC_DROPIT/RQ_DROPME), and encode-error - leaving the counter permanently odd and exposing actively-mutating kernel fields to the lockless reader in nfsd_nl_rpc_status_get_dumpit(). A local low-privileged user on an NFS server who can poll NFSD RPC status via the netlink interface while requests traverse these paths can trigger out-of-bounds reads past the 8-element inline ops array embedded in the rqstp structure, disclosing kernel memory or causing a kernel panic. No public exploit exists (EPSS 0.19%), but the 6.12 LTS series appears unpatched, with fixes confined to 7.2.4 and 7.3-rc1.
NULL pointer dereference in Linux kernel NFSD's NFSv4.2 compound dispatch loop allows remote clients to crash NFS servers built with CONFIG_NFSD_V4_2_INTER_SSC. By inserting an operation such as SETATTR between a foreign PUTFH and SAVEFH within a crafted COMPOUND request, an attacker causes fh_want_write() to dereference a NULL fh_export pointer - killing the nfsd kthread and denying NFS service to all clients. No public exploit has been identified and EPSS stands at 0.22%, reflecting low current exploitation interest.
Persistent denial-of-service in the Linux kernel's NFSv4 server grace-period logic allows any reachable NFSv4 client to hold the server in grace indefinitely by sending CLAIM_PREVIOUS OPEN requests. The root cause is a clock-domain mismatch in clients_still_reclaiming(): the deadline is derived from nn->boot_time (CLOCK_REALTIME, ~1.7 billion seconds since epoch) but compared against ktime_get_boottime_seconds() (CLOCK_BOOTTIME, seconds since last boot - orders of magnitude smaller), making the comparison permanently false. As long as any client continues sending CLAIM_PREVIOUS, the grace period never ends, blocking all non-reclaim NFSv4 operations for every other connected client. No public exploit or active exploitation (CISA KEV) has been identified; EPSS is 0.17%.
POSIX ACL reference-count leaks in Linux kernel NFSD allow a remote attacker to exhaust kernel memory via malformed NFSv4 OPEN compound requests. When the NFS server decodes an OPEN compound carrying valid POSIX ACL createhow4 attributes but encounters a non-seqid-mutating error from claim4 decoding, the compound dispatcher short-circuits before nfsd4_open() executes, permanently leaking both op_pacl and op_dpacl references because OP_OPEN lacked an op_release handler; a secondary leak exists on the NFSv4.0 replay path where op_release was never called from nfsd4_encode_replay(). No public exploit code has been identified and EPSS is 0.20%, but repeated malformed requests against a network-reachable NFSv4 server could drive the kernel toward OOM-induced service failure.
Kernel crash in the Linux libceph client allows a malicious or compromised authenticated Ceph OSD to trigger a BUG_ON assertion in ceph_msg_data_next() by sending a crafted sparse-read reply. The flaw in net/ceph/osd_client.c:osd_sparse_read() validates total data length against summed extents but omits per-extent monotonicity, overlap, and bounds checks, enabling a far-forward extent offset to advance the message-data cursor past the request buffer. Impact is limited to availability - a targeted kernel panic on any Linux client mounting a RADOS-backed filesystem or block device (RBD/CephFS). No public exploit code has been identified at time of analysis, and EPSS is low at 0.20%.
Denial of service and stale-memory disclosure in the Linux kernel SMB/CIFS client (cifs.ko) arise from an integer-overflow in symlink_data()'s error-context parsing loop. A previous fix (commit 7d9a7f1f96cd) checked the post-ALIGN() length against the remaining buffer, but ALIGN() itself overflows: an ErrorDataLength near UINT32_MAX (e.g. 0xFFFFFFF9) makes ALIGN(x,8) wrap to 0, so the bounds check passes and the loop advances by zero bytes, spinning indefinitely while 'p' points into stale data. A malicious or MITM SMB server can trigger this against a connecting client; no public exploit identified at time of analysis and EPSS is low (0.21%, 11th percentile).
Out-of-bounds read in the Linux kernel's eCryptfs filesystem packet parser allows a local low-privileged attacker to leak kernel memory or trigger a denial of service by opening a crafted encrypted file with a version 1 header. The root cause is that ecryptfs_parse_packet_set() computes remaining buffer space using a fixed PAGE_SIZE - 8 formula instead of the actual packet-set buffer length, overstating available bytes for version 1 headers where the packet set begins later. Patches are available for stable branches 6.12.109, 6.18.50, 7.2.4, and 7.3-rc1; no public exploit or active exploitation has been identified.
Kernel crash exposure in the OMAP SSI/HSI driver stems from a missing DMA mask initialization on the synthetic HSI controller device allocated via hsi_alloc_controller(), which bypasses the normal OF/platform device setup path and leaves the embedded struct device without a valid dma_mask. Following DMA API hardening changes in the kernel, dma_map_sg() and related helpers enforce the presence of a valid mask before proceeding, causing warnings or kernel panics on any OMAP SoC system where the driver attempts DMA mapping. The vulnerability is tightly hardware-gated to Texas Instruments OMAP-based platforms; patches are confirmed available in kernel stable branches 6.12.109, 6.18.50, 7.2.4, and 7.3-rc1, and no public exploit or KEV listing exists.
Hugetlb reservation accounting in the Linux kernel underflows when a huge-page folio shared between a parent and child process (after fork) is unmapped by the parent first. Since commit df7a6d1f6405 introduced reservation adjustment logic into `__unmap_hugepage_range()`, the function decrements the reservation counter without verifying that the folio's mapcount has reached zero, leaving the child's mapping unaccounted and the counter in an arbitrarily large (underflowed) state until the child unmaps. Affected kernels from 6.9 through the fix commits expose systems using hugetlb-backed workloads to integrity corruption (overcommit of huge pages) and availability risk (kernel memory management instability). No public exploit or active exploitation is identified at time of analysis.
Out-of-bounds read in Linux kernel ACPI APEI GHES ARM hardware error handler arises from using pointer size instead of structure size when accounting for remaining CPER buffer after skipping the cper_sec_proc_arm header. Affected are stable branches 6.12.x through 6.12.109, 6.18.x through 6.18.50, 7.x through 7.2.4, and pre-7.3-rc1, on ARM systems with ACPI APEI support. No public exploit identified at time of analysis; EPSS is 0.20% (10th percentile), and there is no CISA KEV listing, consistent with a low automated exploitation probability.
TRIM/discard is completely broken on 4Kn (4096-byte native sector) block devices in the Linux kernel's libata-scsi layer, causing every ATA DSM TRIM command to fail with 'Parameter list length error' and trigger WARN_ON() splats on each attempt. The bug is in ata_scsi_write_same_xlat(): ata_format_dsm_trim_descr() caps its output at 2048 bytes (ATA_SCSI_RBUF_SIZE), but the caller compares this return value against the logical sector size - on 4Kn devices (sector_size=4096), this equality check always fails and every TRIM is rejected. No public exploit exists; this is a deterministic functional regression on specific hardware. Vendor-released patches are available in stable kernels 6.12.109, 6.18.50, 7.2.4, and 7.3-rc1.
Out-of-bounds read in the Linux kernel Bluetooth subsystem's eir_get_service_data() leaks kernel heap memory into userspace for ISO broadcast sink configurations. When an ISO broadcast sink processes periodic advertising (PA) reports from a remote Bluetooth broadcaster, a PA payload packed with multiple mismatching Service Data fields causes eir_len to undercount by 2 bytes per field iteration - the subtraction uses only the data length (dlen), omitting the 2-byte length+type header - so the pointer eventually reads past hcon->le_per_adv_data[] into the surrounding struct hci_conn; a drifted field matching the BAA UUID causes those bytes to be copied into iso_pi(sk)->base, recoverable by a local or remote-assisted userspace process via getsockopt(BT_ISO_BASE). No public exploit or CISA KEV listing exists; EPSS of 0.17% (6th percentile) reflects low opportunistic risk, though the Bluetooth adjacent-network constraint caps mass exploitation potential.
Remote unauthenticated denial-of-service in AirSane scanner server (versions prior to 0.4.12) allows an attacker to crash the daemon by sending an HTTP POST request with an arbitrarily large Content-Length value, triggering memory exhaustion and a std::bad_alloc exception. The root cause is in httpserver.cpp, where the Content-Length header value is passed directly to std::string::resize() with no upper-bound validation. A secondary variant using non-numeric Content-Length values produces undefined behavior via NaN-to-integer conversion. No public exploit or active exploitation has been identified at time of analysis.
Four coupled error-path defects in the Linux kernel's SUNRPC RPCSEC_GSS option-array XDR decoder (`gssx_dec_option_array`) enable kernel NULL-pointer dereference, a `group_info` refcount leak, and a latent use-after-free during GSS security-context acceptance. Systems acting as NFS servers or clients with Kerberos authentication (sec=krb5/krb5i/krb5p) are in the affected code path; triggering the NULL deref via a failed allocation or malformed GSS response causes a kernel crash (denial-of-service). No public exploit code exists and EPSS sits at the 10th percentile, indicating low current exploitation probability.
Use-after-free in the Linux kernel NFS-over-RDMA server (svcrdma) allows an adjacent-network attacker to trigger kernel memory corruption when an RDMA_CM_EVENT_ADDR_CHANGE event coincides with a replacement listener cm_id allocation failure, leaving sc_cm_id pointing to a destroyed object. Subsequent kernel operations that dereference sc_cm_id - most immediately rdma_disconnect() in svc_rdma_detach() - read or write freed kernel memory, with potential impacts on confidentiality, integrity, and availability. No public exploit or CISA KEV listing exists at time of analysis; EPSS is 0.15% (5th percentile), consistent with the niche NFS-over-RDMA server attack surface.
NULL pointer dereference in the Linux kernel svcrdma subsystem (NFS over RDMA) allows unauthenticated remote clients to crash the kernel by sending a crafted RDMA Read list whose cumulative segment lengths overflow the rq_pages page array. The vulnerability bypasses per-segment length validation by exploiting an edge case where a single segment fills the page array exactly, causing the loop guard to exit prematurely; a subsequent segment then dereferences the NULL sentinel slot at rq_pages[rq_maxpages] during DMA mapping, triggering a kernel panic. No public exploit code or active exploitation (CISA KEV) has been identified, and EPSS is low at 0.21%, but the network-exposed crash vector warrants patching on any host running NFS-over-RDMA.
Race condition in Linux kernel's core scheduler pick_next_task() path allows a local low-privileged attacker to exploit a TOCTOU window when pick_task() releases the rq (run queue) lock mid-selection, potentially leaking information across SMT sibling threads, corrupting forceidle accounting, and destabilizing the scheduler. The vulnerability breaks the cookie-based SMT isolation guarantee of Core Scheduling: an uncookied task can be committed as the sibling of a cookied task during the lock-release window, violating the trust boundary the feature is designed to enforce. No public exploit exists (EPSS 0.19%), and fixes are available in Linux 7.2.4, 7.3-rc1, and 6.19.
Kernel panic via NULL pointer dereference in the Linux kernel qede Ethernet driver (used by QLogic/Marvell FastLinQ NICs) can be triggered under memory pressure conditions during TPA (Transparent Packet Aggregation) fragment processing. A regression introduced by commit 8a8633978b84 omitted the assignment of tpa_info->buffer.data in qede_tpa_start(), causing a 'ghost' buffer descriptor with a NULL data pointer to be reinserted into the active Rx ring upon SKB allocation failure. When hardware subsequently uses the poisoned ring slot, qede_fill_frag_skb() dereferences the NULL pointer, panicking the kernel. No public exploit has been identified at time of analysis, and EPSS is 0.21% (11th percentile).
Memory management order mismatch in the Linux kernel ring-buffer tracing subsystem allows a local low-privileged user to corrupt kernel memory or crash the system by triggering a ring buffer resize concurrently with cached reader page discard. The root cause is that cpu_buffer::free_page stores only a page address without preserving the allocation order; after a resize changes cpu_buffer->subbuf_order, free_pages() is invoked with the new order rather than the original, corrupting the buddy allocator's state. Vendor-released patches exist in Linux 6.18.51, 7.2.4, and 7.3-rc1; no public exploit or CISA KEV listing exists, and EPSS stands at 0.20%.
Kernel memory corruption in the Linux NVMe/TCP host driver (nvme-tcp) allows a malicious or compromised NVMe-over-TCP target to trigger a wild-memory-access write on the connecting host. The receive path gated C2HData PDUs on blk_rq_payload_bytes() alone; for a REQ_OP_WRITE_ZEROES command (no physical segments but non-zero byte count) the request iterator is never initialized, so an attacker-supplied C2HData is copied into whatever stale, driver-private iterator the previous command on that tag left behind. Fixed across stable trees; no public exploit identified at time of analysis and EPSS is low (0.20%).
Integer underflow in the Linux kernel's SCTP stream reconfiguration (RECONF) handler allows a remote SCTP peer to crash the kernel by sending duplicate RECONF response chunks. When a cached RECONF chunk contains an ADD_OUT request alongside other parameters, a duplicate response causes the kernel to roll back `stream->outcnt` twice - the second decrement underflows the unsigned counter and can trigger a kernel panic. EPSS exploitation probability is 0.20% (10th percentile), no public exploit exists, and no CISA KEV listing has been issued, indicating no known active exploitation at time of analysis.
Two compounding flaws in the Linux kernel's Intel Speed Select Technology (ISST) ioctl handler allow a local low-privileged user to trigger out-of-bounds memory access or NULL pointer dereference in kernel space. The `isst_if_clos_assoc()` function applies an off-by-one validator (`>` instead of `>=`) against the `sst_inst[]` array boundary, permitting socket_id equal to `topology_max_packages()` to index one entry past the allocated array; a second path omits a NULL check for valid in-range socket IDs whose package has no bound TPMI SST instance, both dereferences reaching `map_partition_power_domain_id()`. Patched kernel versions 6.12.109, 6.18.50, and 7.2.4 are available; no public exploit has been identified at time of analysis and EPSS (0.20%, 10th percentile) reflects very low observed exploitation activity.
Use-after-free in the Linux kernel's interconnect subsystem allows local low-privileged attackers to corrupt kernel heap memory, potentially enabling privilege escalation to root. The flaw resides in `icc_get()` and `of_icc_get_by_index()`: when `kasprintf()` fails to allocate memory for `path->name`, the error handler calls `kfree(path)` directly while the path's request structures are already linked into interconnect-node hash lists via `hlist_add_head()`, leaving dangling pointers that subsequent interconnect operations dereference. Patched kernel versions 6.12.109, 6.18.50, and 7.2.4 are available; no public exploit or CISA KEV listing exists at time of analysis.
Stack memory disclosure and integer underflow in the Linux kernel's IPMI ipmb subsystem allow a local low-privileged attacker with write access to the IPMB device node to read uninitialized kernel stack bytes and trigger a buffer length underflow in `ipmb_write()`. The root cause is that `ipmb_write()` parses message fields prior to validating the length byte, so a zero-length or undersized write crosses into adjacent stack memory. CVSS rates confidentiality impact as High (stack leak) and availability impact as High (SMBus block-write length underflow), while EPSS at 0.17% reflects the niche attack surface. No public exploit or active exploitation has been identified.
Use-after-free in the Linux kernel's Open vSwitch (OVS) flow deletion path allows a local attacker with low privileges to crash the kernel or potentially escalate privileges. The bug exists in ovs_flow_cmd_del(): when a CMD_DEL netlink command removes a flow from the OVS table, flow->mask is scheduled for RCU-deferred free, but ovs_flow_cmd_fill_info() subsequently reads that pointer after the RCU grace period may have already freed it-a narrow but real race window confirmed by KASAN traces. Reported by Trend Micro's Zero Day Initiative as ZDI-CAN-32042; no public exploit or active exploitation is confirmed at time of analysis.
NULL and error pointer dereference in the Linux kernel's ravb (Renesas Ethernet AVB) driver allows a local attacker with low privileges to crash the kernel or potentially escalate privileges by querying PTP hardware clock state during an invalid lifecycle window. The flaw affects the ravb driver from Linux 4.2 onward, where priv->ptp.clock is either NULL before the first interface open or left holding an error pointer after a failed PTP clock registration. Fixes are confirmed across stable branches in kernels 6.12.109, 6.18.50, 7.2.4, and 7.3-rc1; no public exploit or CISA KEV listing exists.
Out-of-bounds read in the Linux kernel dm-pcache subsystem allows a local low-privileged user to read past segment data boundaries into adjacent control areas, exposing sensitive kernel memory and potentially crashing the kernel. Specifically, cache_replay(), the writeback worker, and the GC worker all miscalculate the tail-kset read length using PCACHE_SEG_SIZE minus the segment offset rather than the actual data-region remainder via cache_seg_remain(). This affects Linux 6.18.x (before 6.18.50), 7.2.x (before 7.2.4), and pre-release 7.3-rc1. No public exploit or active exploitation has been identified; EPSS is 0.17% (6th percentile).
Use-after-free in the Renesas I3C driver allows a local attacker with low privileges to crash the kernel or potentially escalate privileges on systems equipped with Renesas I3C hardware. The flaw stems from a race between a timed-out transfer cleanup path, which frees the transfer struct, and a late-arriving interrupt whose handler dereferences that now-freed memory. No public exploit has been identified, and exploitation requires specific hardware, but the UAF primitive theoretically supports memory corruption beyond a simple crash.
Denial of service in Apache OpenNLP 2.0.0 through 2.5.11 and 3.0.0-M1 through 3.0.0-M5 lets a remote unauthenticated attacker pin CPU or crash a worker thread by feeding crafted text to either of the two built-in RegexNameFinder patterns (EMAIL and URL) obtained via RegexNameFinderFactory.getDefaultRegexNameFinders(). The EMAIL pattern's ambiguous nested quantifiers cause quadratic catastrophic backtracking (~32 KB of local-part characters burns seconds of CPU per find() call), while the URL pattern's nested capturing repetition recurses once per query token, exhausting the thread stack and throwing StackOverflowError (~4 KB, or ~1 KB on -Xss512k worker threads). No public exploit was identified at time of analysis; this is availability-only with no code execution or data exposure.
SymSpellModelSerializer in Apache OpenNLP 3.0.0-M4 and 3.0.0-M5 crashes the host JVM with an OutOfMemoryError when deserializing a crafted SymSpell binary model file whose unigramCount or bigramCount field is set to Integer.MAX_VALUE, forcing a 4-8 GB heap allocation attempt. The root cause (CWE-789: Uncontrolled Memory Allocation) is that these two 32-bit count fields are validated only for non-negativity - no upper bound prevents attacker-controlled pre-sizing of a LinkedHashMap before any dictionary entries are consumed. The attack payload is described by the advisory as under 100 bytes, requires no prior authentication if the target application accepts externally supplied model bytes, and is effective against every deserialization code path in the opennlp-spellcheck extension. No public exploit has been identified at time of analysis, and exposure is substantially limited to the small population running pre-release milestone builds.
Process termination via unrecovered panic in AWS SDK for Go v2's EventStream header decoder allows a remote actor to crash any Go application consuming AWS streaming service responses. Sending a crafted EventStream response frame containing a header value type byte outside the valid range triggers an unhandled Go panic that propagates to terminate the host process. The affected surface includes Bedrock inference and agent runtime packages, making AI/ML workloads on AWS a particularly relevant target; no public exploit or active exploitation has been confirmed.
Use-after-free in Microsoft Edge (Chromium-based) enables an authenticated network attacker to achieve remote code execution with scope change across all supported platforms including Windows, Linux, macOS, Android, and iOS. The vulnerability carries a CVSS 8.5 with high complexity and scope change (S:C), indicating that successful exploitation escapes the browser's sandboxing boundary and affects the underlying host. No public exploit or CISA KEV listing is present at time of analysis, though patch-available status from Microsoft confirms vendor acknowledgment.
Uncontrolled resource consumption in jackson-databind's CoreXMLDeserializers allows unauthenticated remote attackers to trigger multi-minute CPU saturation via a single oversized HTTP request. When jackson-databind deserializes a JSON string into a javax.xml.datatype.Duration or XMLGregorianCalendar field, it passes the raw string directly to JDK's DatatypeFactory, which internally calls BigInteger(String) and BigDecimal(String) constructors that exhibit O(n²) time complexity relative to digit count. Because the malicious digits are embedded in a JSON string token rather than a JSON number token, jackson-core's existing StreamReadConstraints.maxNumberLength guard is bypassed entirely. A few megabytes of input containing millions of digits can consume tens of seconds to minutes of CPU, and a handful of concurrent such requests can exhaust a server's worker thread pool. No public exploit or CISA KEV listing has been identified at time of analysis.
Memory exhaustion DoS in the expressjs/compression Node.js middleware (versions before 1.8.2) allows remote unauthenticated attackers to crash any Express server using the package. When a client aborts an HTTP connection while a compressed response is in flight, the zlib transform stream allocated for that response is never destroyed, permanently leaking its native heap memory. An attacker can automate rapid request-then-abort cycles to exhaust all available server memory, causing process termination. The issue is fixed in 1.8.2 per the OpenJS Foundation advisory; no public exploit code or KEV listing has been identified at time of analysis.
Unauthenticated remote denial of service in Net-SNMP through 5.9.5.2 allows any network-reachable attacker to permanently suspend all SNMP processing by connecting to the SMUX listener and withholding data. The root cause is smux_accept() issuing a blocking recv() call with no timeout on every newly accepted SMUX connection, stalling snmpd's single-threaded main loop indefinitely. Because the block is permanent and snmpd performs all SNMP agent work in that same thread, the entire agent - monitoring, traps, polling - ceases until the process is manually restarted. No public exploit identified at time of analysis, though a researcher gist documenting the behavior is publicly linked in the advisory.
Unauthenticated remote denial-of-service in HPE IceWall Federation Agent and Proxy allows a network attacker to crash or render unavailable the identity federation and authentication proxy service. The root cause is improper handling of incorrect data types (CWE-241), meaning the software fails to gracefully reject or process malformed input, triggering an availability failure. No active exploitation or public exploit code has been identified at time of analysis.
Memory exhaustion in multiparty Node.js library (versions 2.1.0 through <4.3.1) allows any unauthenticated network attacker to crash the host process with a single crafted HTTP request. The parser accumulates multipart part headers into an unbounded in-memory buffer with no size cap, consuming available process memory until the Node.js server is killed. The vulnerability carries no confidentiality or integrity impact - this is a pure availability issue addressed in multiparty 4.3.1, which enforces a maximum on the bytes buffered per part header block.
Denial of service in the RACER swarm-exploration framework (Robotics-STAR-Lab / SYSU STAR Group) lets an attacker who can reach the exploration manager's ROS communication graph drive FastExplorationFSM::optTimerCallback() into an uncontrolled resource-consumption state, stalling or crashing the node and halting multi-robot exploration. The flaw is classified CWE-400 with a CVSS 3.1 base of 7.5 (full availability impact, no confidentiality or integrity impact). No public exploit code or CISA KEV entry was identified in the supplied intelligence, and EPSS is only 0.15% (5th percentile), consistent with the niche academic-robotics deployment of this codebase.
Heap-based buffer overflow in the Sofia IPC daemon's WS-Addressing Action transformation function allows remote unauthenticated attackers to crash Xiongmai XM530 IP cameras or potentially achieve arbitrary code execution by sending a crafted SOAP request containing a wsa5:Action string exceeding 128 bytes. Affected firmware is HMT.CM2005-v220608.1837 and all earlier versions. A public vulnerability research repository exists for this CVE, suggesting exploit code or detailed PoC is available. No public exploit identified as confirmed actively exploited at time of analysis.
A denial-of-service condition in ZJU-FAST-Lab EGO-Planner-v2 (all versions up to commit 5c99a95880401e2599638d567abc0e240396cb42) is triggered by the interaction between the traj_server and poscmd_2_odom nodes and the EGOReplanFSM emergency recovery logic, causing the planner to become unresponsive. The affected software is an open-source trajectory planner for autonomous quadrotors, so a successful denial of service can translate from a crashed planning process directly into loss of autonomous flight control. There is no confirmed evidence of active exploitation (not in CISA KEV) and no public exploit code identified at time of analysis; EPSS is 0.15% (5th percentile), indicating low predicted exploitation activity.