Denial Of Service
Monthly
Permanent denial of service in Motor.NET before 0.19.0: any party able to publish to a queue or topic that a Motor.NET service consumes can crash that service and keep it down. The service deserializes incoming payloads with JsonNetDeserializer.Deserialize, and structurally valid JSON that does not match the handler's expected type raises a JsonSerializationException that the code does not catch, so TypedConsumerService.SingleMessageConsumeAsync classifies the message as a CriticalFailure and calls StopApplication. Because the offending message is never acknowledged, it is redelivered on restart and re-triggers the shutdown, producing a crash-loop that persists until the message is removed or the library is upgraded; the CVSS 4.0 score is 8.7 (AV:N/AC:L/PR:N/UI:N/VA:H) with no confidentiality or integrity impact, and no public exploit code was identified at time of analysis.
A single crafted Kafka record can permanently crash-loop a Motor.NET service running any version before 0.19.0 that consumes a topic with CloudEvents message handling. The record declares content-type application/cloudevents+json but omits the mandatory CloudEvent attributes id, source and type, causing an ArgumentException inside KafkaMessageConsumer.SingleMessageHandlingAsync that the framework treats as fatal; the host stops, and because the same poison record is re-consumed on every restart, the service never recovers until the message is removed or the software is upgraded. The authoritative CVSS 4.0 vector (8.7, AV:N/AC:L/AT:N/PR:N/UI:N/VA:H) rates this as a high-availability-impact, unauthenticated denial of service with no confidentiality or integrity exposure; the key environmental precondition is that the attacker can produce records to the consumed topic (Kafka-level produce authorization). No public exploit code or confirmed active exploitation was identified at time of analysis, and the vendor has released 0.19.0 as the complete fix.
Denial of service in Motor.NET before 0.19.0 lets anyone able to publish to a queue the service consumes crash it permanently by sending a message with crafted CloudEvents headers - an invalid cloudEvents:motorversion value, a header whose value is a single double-quote, or an unparsable typed attribute such as cloudEvents:time. The malformed header raises an unhandled exception in ExtractCloudEvent that the host treats as fatal; because the message is never acknowledged, the broker redelivers it after restart and the crash repeats as a permanent crash-loop. The impact is availability-only (CVSS 8.7, VA:H) and requires broker-side publish access rather than authentication to Motor.NET itself; no public exploit was identified at time of analysis.
Quadratic-time parsing in Mistune's inline link parser (all versions through 3.2.1) lets remote attackers stall server-side markdown rendering by submitting a small payload made of long runs of unmatched '[' characters. Each '[' triggers a probe of parse_link_text/parse_link_label, so work grows O(N^2): roughly 30+ seconds of CPU for a 16 KB input and proportionally worse as input grows, with the nested '[[..]]' shape costing a further ~4x. The vendor advisory reports a reproducible benchmark proof-of-concept against mistune 3.2.1 in default configuration (no plugins or options required); there is no CISA KEV listing and no confirmed in-the-wild exploitation, and the impact is CPU time only - no memory growth, no confidentiality or integrity effect - so sustained denial requires repeated requests to exhaust a worker or thread pool, most acutely on single-threaded WSGI deployments.
Remote unauthenticated attackers can exhaust server CPU by submitting markdown with deeply nested square brackets to Mistune 3.0.0 through 3.2.1, where the parse_link_text helper degrades to quadratic time and a roughly 16KB payload can occupy a worker for over a minute. Impact is availability-only with no confidentiality or integrity effect, and exposure is bounded by whether an endpoint actually passes untrusted markdown to mistune. A public proof-of-concept reference exists in the upstream advisory; it is not verified as working exploit code, there is no CISA KEV entry, and the this_week decision label reflects a credible but non-emergency priority.
Quadratic-time CPU exhaustion in Mistune 3.0.0 through 3.2.1 allows unauthenticated remote attackers to tie up a rendering worker for minutes using only a few tens of kilobytes of Markdown built from unbalanced '[' or '
Signals point to a real but bounded availability risk rather than a top-tier emergency: the MongoDB C# Driver's BSON library interns every BsonSymbol produced during deserialization and retains those symbols for the lifetime of the process with no cap, so an unauthenticated remote party able to get attacker-influenced BSON or JSON deserialized can drive unbounded memory growth and deny service to the host application (availability-only; assessed CVSS 3.1 AV:N/AC:H/PR:N/UI:N/C:N/I:N/A:H, CNA CVSS 4.0 score 8.2). Exposure is conditional - the application must actually deserialize untrusted content through the driver and that content must carry BSON symbol values; merely having the driver installed is not a trigger, and no MongoDB server component is affected. No public exploit code or confirmed active exploitation has been identified at time of analysis; version 3.13.0 removes symbol interning as the fix.
Authenticated clients of the Crawl4AI Docker API server (0.9.0 through 0.9.4) can freeze the entire service, including the /health endpoint, by submitting a catastrophic-backtracking regular expression in a /crawl request body through RegexExtractionStrategy.custom, RegexChunking.patterns, or a schema field of type 'regex' in the JSON extraction strategies. Python's re engine does not release the GIL while matching, so a single crafted pattern combined with an attacker-influenced page blocks the event loop and the crawl deadline cannot interrupt it. Exploitation requires a valid API bearer token and a crawl target whose text triggers the backtracking; a public proof-of-concept request is published in the vendor advisory, no active exploitation has been reported, and the vendor fixed the issue in 0.9.5. The library imported directly is not exposed unless the host application itself feeds untrusted regexes into these strategies.
CPU exhaustion in libexpat through 2.9.0 lets remote attackers stall XML parsing workloads by supplying a moderately sized document with a large DTD and nested entity definitions, because the billion-laughs amplification check in accountingDiffTolerated() aborts too late and each parse can burn tens of seconds of CPU. Unauthenticated remote attackers (CVSS:4.0 AV:N/AC:L/AT:N/PR:N/UI:N) can trigger it wherever an application passes untrusted XML to libexpat with DTD processing and internal general/parameter entity expansion enabled; impact is limited to availability, with no disclosure or modification. No confirmed active exploitation is known, though a public PoC reference exists; it is not verified as working exploit code, and vendor patch material is available upstream (PR #1433).
Unbounded memory leak in the msgpack-python C extension (msgpack._cmsgpack) allows a remote attacker to exhaust process memory by repeatedly sending malformed MessagePack maps, with approximately 4 MiB retained per failed parse and roughly 100 MiB after 25 failed parses as measured by the reporter. The flaw is triggered when strict_map_key=True (the default) rejects a map key that is neither str nor bytes: the decoded value has already been constructed and is never released, affecting both unpackb() and the streaming Unpacker API in releases up to 1.2.2. No public exploit code or active exploitation is identified in the available data, but the CVSS of 7.5 is availability-only (C:N/I:N/A:H), so impact is limited to denial of service rather than data disclosure or code execution; a public proof-of-concept reference exists in the advisory, though it is not verified as working exploit code.
SmileCDR HAPI FHIR can be driven into heap exhaustion and service crashes by unauthenticated attackers who submit highly compressed payloads that are expanded during server-side decompression. The CVSS:4.0 score of 8.7 (assessed as CVSS:3.1 AV:N/AC:L/PR:N/UI:N/C:N/I:N/A:H) reflects an availability-only impact: no confidentiality or integrity loss, no privileges or user interaction required, but a high availability hit against network-reachable FHIR endpoints that decompress untrusted input. No public exploit code has been identified at the time of analysis and the CVE is not listed in CISA KEV; the vuln.today decision label is this_week, so treatment should be scheduled in the current patch cycle rather than treated as an emergency, and deployed instances that accept compressed request bodies should be prioritized.
PhpSpreadsheet's SpreadsheetML reader (Reader\Xml) can be induced to pass a DOCTYPE/ENTITY block into SimpleXML despite its entity-blocking scanner, because the scanner's html_entity_decode callback runs after the pattern check and the content is scanned twice. Files containing a double-encoded less-than character (&#60; or &#x3C;) smuggle a literal '<!DOCTYPE' past both passes, so processing attacker-controlled XML with the Xml reader yields entity-expansion denial of service on all PHP versions, plus local file read or SSRF where external entities are still resolved (PHP before 8.0, libxml2 before 2.9, or any PHP build with LIBXML_NOENT enabled). Affected releases are 3.3.0 through 3.10.7 and 4.0.0 through 5.9.0; vendor fixes are released as 3.10.8 and 5.10.0. No CISA KEV listing or confirmed active exploitation is reported, but the vendor advisory itself publishes a proof-of-concept that was reportedly exercised against PhpSpreadsheet 5.9.0, and the priority signal places remediation in the this-week band rather than emergency.
Denial of service in the PhpSpreadsheet ODS reader allows unauthenticated remote attackers to exhaust a PHP worker's CPU and memory by loading a small crafted .ods file, terminating the process. Every maintained branch is affected - 1.x up to and including 1.30.6, 2.0-2.1 up to 2.1.18, 2.2-2.4 up to 2.4.7, 3.x up to 3.10.7 and 4.x-5.x up to 5.9.0 - whenever an application parses an attacker-supplied OpenDocument Spreadsheet using the default read filter, since no authentication or special configuration is required. A public proof-of-concept payload is documented in the vendor advisory; it is not verified as working exploit code, and no confirmed active exploitation (CISA KEV) is recorded at time of analysis.
PhpSpreadsheet's XLSX reader can be tricked into parsing attacker-supplied DOCTYPE and entity declarations, because the XML security scanner in XmlScanner::scan() only anticipates NUL-byte interleaving and is defeated by ISO-2022 escape sequences inserted into the declaration. Any application that accepts user-uploaded spreadsheets and processes them with the Xlsx reader is affected, and the prerequisite is only that the file be opened; the CVSS vector shows PR:N/UI:N, so no authentication or interaction beyond the file upload is needed. The reliably exploitable outcome is entity-expansion denial of service on all supported PHP versions (a ~2 KB XLSX produced ~45 MB peak memory in the advisory's testing, roughly a 22,000:1 amplification); local file read and SSRF through external SYSTEM entities are conditional on PHP < 8.0 / libxml2 < 2.9 or on the application explicitly passing LIBXML_NOENT. A public proof-of-concept is published in the GitHub security advisory at https://github.com/PHPOffice/PhpSpreadsheet/security/advisories/GHSA-3x94-m842-937c; there is no CISA KEV confirmation of active exploitation, and the priority decision is to patch this week.
Excessive CPU consumption during DNS name decompression in dnspython before 2.9.0 lets remote attackers degrade or deny service to applications that decode untrusted DNS wire-format messages. The parser function dns.name.from_wire_parser follows attacker-controlled compression pointer chains without a hop limit, so a single crafted message can force long pointer traversal; no authentication, user interaction, or non-default configuration is required once untrusted DNS data reaches the parser (CVSS 3.1 AV:N/AC:L/PR:N/UI:N, A:H). The exposure is bounded by architecture rather than by exploit complexity: resolvers, DNS clients and any service that decodes DNS queries or responses from untrusted peers are affected, while systems that only parse trusted internal DNS data are not. No public exploit was identified at time of analysis and the issue is not in CISA KEV; the vendor-released patch is version 2.9.0, and the vendor states there are no known workarounds.
Denial of service in the Python GraphQL-core library (3.0.0 through 3.2.11, plus 3.3.0 pre-releases before 3.3.0rc1) lets an unauthenticated remote attacker pin a worker process for minutes with a single shallow, highly compressible query. The OverlappingFieldsCanBeMerged rule, which ships in specified_rules and is therefore enabled by default on every GraphQL-core server, re-compares flattened inline fragments at every nesting level and can be driven into quadratic - worst case O(n²·m²) - CPU work during validation, before any resolver, depth limiter or complexity analyzer runs; the vendor advisory documents a 364 KB query consuming one core for over 300 seconds on 3.2.8. Severity is availability-only (no confidentiality or integrity impact) and the CVSS 3.1 vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H scores 7.5; the advisory includes a proof-of-concept query generator with measured timings, which is a public PoC reference rather than independently verified working exploit code, and there is no confirmed active exploitation (no CISA KEV listing) at time of analysis. Fixed versions 3.2.12 and 3.3.0rc1 enforce a 250,000-comparison budget.
Memory exhaustion in GraphQL-core 3.2.3 through 3.2.11 and 3.3.0a3 through 3.3.0rc0 lets a remote unauthenticated attacker OOM-kill a GraphQL worker by sending a highly compressible comment-padded document that sails past the parser's max_tokens guard. The lexer still allocates and chains a Token object for every comment it skips, but those tokens were invisible to the counter, so a 38 MB payload parsed under max_tokens=5 produced roughly +92 MB of resident memory per the vendor's measurements. The vendor advisory ships a proof-of-concept that demonstrates the bypass; the flaw is not listed in CISA KEV and no in-the-wild exploitation is reported, and the vendor rates the impact as availability-only (CVSS 7.5, A:H, no confidentiality or integrity effect).
Crafted HTTP requests sent to Next.js App Router applications that define at least one Server Action can drive excessive CPU consumption and block processing of further requests in the same Node.js process, producing a denial of service rated 8.2 (CWE-770, uncontrolled resource allocation). The affected ranges are 13.0.0 through 15.5.27 and 16.0.0 through 16.4.0; applications on the Pages Router, and App Router applications with no Server Actions defined, are explicitly not vulnerable, which substantially narrows the exposed population. Exploitation requires no authentication and no user interaction, and no public exploit code has been identified at time of analysis, with no CISA KEV entry and no EPSS signal supplied; the vendor states that upgrading to 15.5.28 or 16.4.1 is the only complete fix, with no workaround available.
Unauthenticated memory exhaustion in LiteSpeed's lsquic HTTP/3 stack (all versions before 4.9.0) allows a remote attacker to drive a server or client into high memory use and ultimately denial of service by sending header blocks that pile up completed header sets faster than the application claims them, a technique the reporters call a QPACK amplification attack. Any lsquic-based endpoint that exposes HTTP/3 over UDP/QUIC is affected with default settings, since the library applied no cap on buffered header sets; no credentials or user interaction are needed and the vector scores 7.5 (AV:N/AC:L/PR:N/UI:N, availability-only). There is confirmed vendor activity in the form of upstream pull request 666 and the 4.9.0 release notes, and a public research write-up from FoxIO describes the attack class, but no public exploit code was identified at time of analysis and the issue is not in CISA KEV.
Denial of service in body-parser before 1.20.9 and in 2.0.0 through 2.3.x allows any remote unauthenticated client to terminate a Node.js or Express process that uses default body parsing. A request with a Content-Encoding such as gzip and trailing bytes after a valid compressed frame causes the finished decompression stream to emit an error with no error listener, resulting in an uncaught exception. The vendor has released fixes in 2.4.0 and 1.20.9; no public exploit code was identified in the supplied intelligence at time of analysis.
Applications that configure the multer upload middleware (npm package, versions 2.0.0 through 2.4.x) with an asynchronous fileFilter can be crashed by a single aborted multipart request, taking down the entire Node.js process. When a request is aborted while one file's async fileFilter callback is still pending, that callback later resumes a parsing stream multer has already torn down, and the underlying busboy parser emits an error on a stream with no remaining error listener, which Node raises as an unhandled 'error' event and exits. The bug is unauthenticated and requires no user interaction (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H, 7.5), but it is conditional: uploads using a synchronous fileFilter, and the storage engine choice, do not matter. No public exploit code has been identified at time of analysis, and the issue is fixed in multer 2.5.0.
Unauthenticated remote attackers can crash an Erlang node running the Dropbox esaml SAML library (versions 0.1.0 and later) by sending a crafted SAMLResponse to the SAML assertion-consumer (ACS) or single-logout (SLO) endpoint. Because esaml_binding:decode_response/2 hands the decoded XML to xmerl_scan:string/2 before any signature check, every distinct element name, attribute name, and namespace prefix in the document is interned as a permanent Erlang atom; a few dozen requests carrying documents with thousands of unique names exhaust the roughly one-million-entry atom table and abort the VM. The CNA rates this CVSS 4.0 at 8.7, while independent assessment aligns with a CVSS:3.1 vector of AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H - pure availability impact with no confidentiality or integrity loss; no public exploit code was identified at time of analysis and there is no CISA KEV listing, though the flaw is trivially reachable by anyone who can POST to a SAML endpoint.
Unauthenticated memory exhaustion in the Erlang esaml SAML library (version 0.1.0 and later) lets a remote attacker crash the hosting Erlang VM by posting a small, highly compressed SAMLResponse to a service provider's assertion-consumer (ACS) or single-logout endpoint. Because esaml_binding:decode_response/2 base64-decodes and inflates the attacker-supplied payload with zlib:unzip/1 before any signature check, a deflate stream with roughly 1000:1 amplification expands tens of kilobytes into hundreds of megabytes of binary and then gigabytes of charlist heap, so only a handful of requests are needed to take the node down. No public exploit code or confirmed active exploitation was identified at time of analysis, but the issue is trivially reachable on any internet-facing esaml service provider and upstream fix commits are already published.
Unauthenticated remote denial of service in the Erlang SAML library esaml (0.1.0 and later) allows an attacker to abort an Erlang node by exhausting its atom table through unsigned LogoutResponse messages posted to a Service Provider's single-logout endpoint. Because esaml_sp:validate_logout_response/2 only verifies a signature when one is present, an unsigned message reaches esaml:decode_logout_response/1, where status_code_map/1 converts each unrecognized StatusCode into a permanent atom with list_to_atom/1 - roughly one million distinct status values (the VM's default atom-table ceiling) are enough to crash the node. The CNA scored this CVSS 4.0 8.2 with the AT:P attack-requirement metric; the practical barrier is volume and time rather than skill, and no public exploit code or confirmed active exploitation was identified at time of analysis.
A truncated HTTP/2 PING frame sent to an HAProxy frontend that accepts HTTP/2 causes the H2 multiplexer to spin in a busy loop, driving worker-thread CPU to saturation and degrading or denying service to legitimate traffic (CVSS 7.5, availability-only). Only deployments with an HTTP/2-enabled listener (h2 in the bind line / ALPN negotiation) are reachable; HTTP/1.x-only setups are unaffected, and no authentication or user interaction is required against an H2 endpoint. Sustained impact depends on the attacker holding many concurrent connections open, since each connection is eventually cleared by 'timeout http-request'; no public exploit code was identified at the time of analysis and there is no KEV confirmation of active exploitation.
HAProxy 3.x releases before 3.4.6 can be driven into persistent denial of service by remote unauthenticated attackers sending a single crafted HTTP/1.1 chunked request: the H1 multiplexer's chunk parser re-processes the same bytes without advancing, pinning one worker thread at 100% CPU even after timeouts fire and the client disconnects. Repeating the request exhausts the worker pool until the process is manually restarted, so confidentiality and integrity are unaffected but availability is fully lost. Attack surface is standard HTTP/1.1 frontend traffic (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H, CWE-835); no public exploit code or CISA KEV entry is associated with this CVE at time of analysis, and the upstream fix commit is published in the HAProxy git repository.
Unauthenticated resource exhaustion in the Drupal Webform contrib module (all versions from 0.0.0 through 6.2.12, plus 6.3.0 through 6.3.1) lets remote attackers drive excessive resource allocation and degrade or deny availability of a Drupal site; confidentiality and integrity are unaffected (CVSS 3.1 7.5, AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H). The flaw is CWE-770 (allocation without limits or throttling), and the vendor's stated remediation is to update the module per Drupal security advisory sa-contrib-2026-170. Risk signals are moderate rather than urgent: EPSS is 0.19% (8th percentile), there is no CISA KEV listing, and no public exploit has been identified at time of analysis, though the priority label places remediation in the current week.
Unauthenticated denial of service in Splunk Enterprise deployments below 10.4.4, 10.2.8, and 10.0.11 is reachable through the Edge Processor package download endpoint in the Splunk Pipeline Builders app. Repeated unauthenticated requests force the Splunk daemon to read and base64-encode the complete package in memory for each request, exhausting system resources and potentially making the service stop responding (CVSS 7.5, CWE-400). Exposure is limited to instances running Edge Processor/Pipeline Builders, and no public exploit identified at time of analysis.
Splunk Enterprise deployments running Splunk Web in versions below 10.4.4, 10.2.8, 10.0.11, and 9.4.16 can be taken offline by a single unauthenticated request to the Splunk Web JavaScript logger endpoint, which parses request headers without any limit and can be driven into an unresponsive state. Recovery requires a full restart of Splunk Enterprise, making this an easy-to-repeat availability attack against any instance whose Splunk Web port is reachable by the attacker. The assessed vector (AV:N/AC:L/PR:N/UI:N, availability-only) is consistent with a low-complexity denial of service; no public exploit code or confirmed active exploitation was identified at time of analysis, and the risk is bounded to service availability with no implied data loss or compromise.
Denial of service in toml-node versions prior to 5.0.1 allows remote attackers to synchronously block the Node.js event loop by supplying a TOML document containing an extremely long hexadecimal, octal, binary, or decimal integer literal. Because toml.parse() calls BigInt() on the digit sequence before the 64-bit range check runs in reduceInteger(), conversion cost grows super-linearly with digit count, so a single crafted document can stall all concurrent work in the process (CVSS 3.1: 7.5, AV:N/AC:L/PR:N/UI:N, availability-only). The flaw is exploitable only where an application parses attacker-influenced TOML; there is no confidentiality or integrity impact, no public exploit code identified at time of analysis, and the issue is fixed in 5.0.1.
Denial of service in Django REST Framework prior to 3.18.2 allows a remote unauthenticated client to pin synchronous workers for prolonged periods by sending Accept headers packed with a large number of media-type tokens carrying oversized quoted parameter values and many semicolons; the Content-Type header is a secondary, now length-capped, vector. Impact is availability-only (CVSS 3.1 AV:N/AC:L/PR:N/UI:N/A:H, 7.5) and self-healing once requests stop, but on WSGI/sync worker deployments (e.g. gunicorn sync workers) concurrent requests can exhaust the worker pool and take the service offline. No public exploit code and no active exploitation are indicated in the supplied intelligence, and no EPSS score was provided, so this reads as a genuine but bounded availability risk rather than an emergency.
Apache Traffic Server 9.2.15, 10.1.4, and 10.2.0 can be crashed by a remote, unauthenticated attacker through a null pointer dereference (CWE-476) while proxying chunked origin responses. The race requires a client POST request to be torn down via TCP half-close at the moment ATS reparses a chunked origin response from a reentrant write-ready callback, so it is timing-dependent and typically needs repeated attempts; impact is limited to denial of service (crash), with no code execution, data disclosure, or integrity loss. No public exploit was identified at time of analysis, and vendor-released fixed versions are 9.2.16, 10.1.5, and 10.2.1.
Apache Traffic Server 9.2.15, 10.1.4, and 10.2.0 can be driven into a denial-of-service condition through log amplification: each internal '@' header stripped from a request causes an unthrottled diagnostic log line, so requests carrying many such headers generate disproportionate log writes. The issue is network-reachable and unauthenticated, with high availability impact when log storage, rotation, or retention is constrained; no public exploit identified at time of analysis.
Memory exhaustion in Pillow's SPIDER image plugin lets an attacker-supplied multi-frame .spider stack file bypass the library's decompression-bomb guard and drive allocations of several gigabytes. The flaw lives in SpiderImageFile.seek(), which reparses a later frame and overwrites self._size without re-running Image._decompression_bomb_check(), so a tiny file with a benign first frame passes Image.open() and then declares enormous dimensions in a later mode-F frame; loading that frame consumes roughly four bytes per attacker-declared pixel. All Pillow releases prior to 13.0.0 that parse SPIDER stacks are affected, and exploitation requires the application to open an untrusted multi-frame SPIDER file and then seek past frame 0 - a niche, passive-interaction path consistent with the assessed CVSS 3.1 vector AV:N/AC:L/PR:N/UI:R/S:U/C:N/I:N/A:H. No public exploit code or confirmed active exploitation was identified at time of analysis, and the vendor patch is 13.0.0.
Out-of-bounds read in Pillow's TIFF encoder before 13.0.0 can disclose process memory or crash the application when TIFF tag values are inconsistent with the actual image data. The issue is not triggered by simply opening a malicious image; exploitation requires an application to save a TIFF using attacker-influenced tag metadata, and the assessed vector (AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:L) reflects that high precondition. The Pillow 13.0.0 release notes rank it High severity, but no CISA KEV listing, EPSS score, or public exploit code was provided, so no public exploit identified at time of analysis.
Denial of service in Pillow's legacy TarIO parser (all versions before 13.0.0) allows remote unauthenticated attackers to supply a crafted TAR archive that traps the parser in an uncontrolled loop, consuming a full CPU core and never returning. The exposure is sharply limited because only applications that directly call PIL.TarIO.TarIO on attacker-controlled archives are affected; the default Image.open() path does not reach this parser. No public exploit code was identified at time of analysis, and the raw 7.1 CVSS 4.0 score should be weighed against the non-default API prerequisite and required user or service interaction to supply the malicious archive.
Denial of service in Pillow's PDF append workflow (versions 5.1.0 through 12.x) lets an unauthenticated remote attacker exhaust CPU by supplying a compact malformed PDF whose /Pages /Kids arrays repeatedly reference the same child /Pages node. When an application calls Image.save(..., format="PDF", append=True) on that attacker-controlled file, PdfParser.linearize_page_tree() expands the shared subtree as if it were distinct pages, producing exponentially many page references before the append completes. Impact is availability-only (C:N/I:N/A:H) and reachability is narrow - ordinary image decoding and standalone PDF export never hit this code path - so no public exploit code has been identified at time of analysis and the issue is fixed in Pillow 13.0.0.
Denial of service in Apache Traffic Server 8.0.0 through 8.1.11, 9.0.0 through 9.2.15, 10.0.0 through 10.1.4, and 10.2.0 allows unauthenticated remote attackers to exhaust a downstream HTTP/1.1 origin's worker pool by sending a truncated HTTP/2 Content-Length value that ATS mishandles when relaying the request. The issue is limited to proxy deployments where ATS accepts HTTP/2 client connections and forwards traffic to an HTTP/1.1 origin; no public exploit code or confirmed active exploitation was identified at time of analysis.
A stale-state flaw in Zephyr's experimental MCTP-over-I2C+GPIO target binding lets any device that can act as an I2C controller on the shared management bus crash the target with two ordinary transactions, because the selected pseudo-register and data-phase flag are never reset at STOP. The resulting NULL pointer dereference occurs inside the I2C target interrupt handler, so on parts that trap address-0 accesses it produces a fatal exception and device reset or halt; on other parts it degrades to a malformed packet buffer reaching the MCTP stack and silently dropped transmit packets. Only builds with CONFIG_MCTP and CONFIG_MCTP_I2C_GPIO_TARGET=y are exposed, exploitation requires a pre-existing foothold on the same physical bus (compromised peer endpoint or BMC), and no public exploit code was identified at time of analysis; the authoritative assessment scores this as availability-only impact (CVSS:3.1/AV:A/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H) rather than the C:L/I:N variant in the published vector.
Crafted IPv4 packets addressed to the router itself can drive the line card network processor on Cisco IOS XR into repeated resets, causing traffic loss and - under sustained attack - a persistent shutdown state that requires manual line card reload to recover. The flaw affects carrier-grade hardware specifically: Cisco ASR 9000 and 9900 Series Aggregation Services Routers plus the ASR 9902 and 9903 Compact High-Performance Routers. No authentication or user interaction is needed (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H, 7.5), and no public exploit code or confirmed active exploitation was identified at time of analysis; exposure is bounded by the fact that transit IPv4 traffic and all IPv6 traffic cannot trigger the fault.
Denial of service in Cisco Expressway Series allows unauthenticated remote attackers to crash the service by triggering a null pointer dereference (CWE-476). The issue is network-reachable, low complexity, and requires no user interaction, with impact limited to availability; exact affected and fixed releases must be confirmed in Cisco's advisory. No public exploit code was identified at time of analysis, and exposure is greatest for internet-facing or externally reachable Expressway deployments.
Network-reachable assertion failures in Cisco Expressway Series (TelePresence VCS Expressway) can terminate video and collaboration signaling services, allowing remote attackers to cause a denial-of-service condition without authentication or user interaction (CVSS 7.5). The vendor advisory describes multiple internally discovered CWE-617 reachable-assertion issues fixed in hardening releases; the exact affected and fixed versions are not specified in the provided data. No public exploit was identified at time of analysis, and there is no CISA KEV confirmation of active exploitation.
Denial of service in the Cisco Snort 3 detection engine allows remote, unauthenticated attackers to exhaust device resources and disrupt inspection services on any deployment actively processing attacker-influenced traffic - including Cisco Secure Firewall Threat Defense (FTD), the Cisco UTD Snort IPS engine, Cisco Cyber Vision, and upstream Open Source Snort 3. The flaw is rooted in uncontrolled resource consumption (CWE-400) and carries a CVSS 3.1 base score of 7.5 (AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H), meaning only availability is impacted - there is no confidentiality or integrity loss and no code execution. No public exploit code or in-the-wild activity has been identified at time of analysis, and no CISA KEV listing exists; exploitation is gated by the requirement that the target actually sits inline as an IDS/IPS or firewall in a traffic-inspection path.
Unbounded resource retention in PowerDNS Recursor lets remote clients drive memory exhaustion by repeatedly opening and closing DNS-over-TCP connections. Recursor 5.3.x before 5.3.12 and 5.4.x before 5.4.8 retain entries for closed TCP client addresses until process exit, so the effect is cumulative over long uptimes and can degrade or deny DNS service; the independent assessment is availability-only (CVSS 3.1: 5.3, AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L). No public exploit code or CISA KEV entry was identified at time of analysis.
Denial of service in PowerDNS Recursor 5.3.x before 5.3.12 and 5.4.x before 5.4.8, where a crafted DNS packet containing malicious name-compression pointers can drive runaway decompression of DNS names and exhaust the recursor's CPU, degrading or halting resolution service (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H, 7.5 High). Unauthenticated remote attackers can trigger it either by sending the packet straight to a reachable recursor port or by luring the recursor into resolving a name hosted on an attacker-controlled authoritative server that returns the malformed response - so open or internet-reachable resolvers carry the most risk, while recursors restricted to trusted internal clients are far less exposed. The flaw ships as part of PowerDNS Security Advisory 2026-12, a seven-CVE bundle whose most serious member (CVE-2026-85555) is a DNSSEC validation bypass and cache poisoning; no CVSS vector, EPSS score, KEV listing or public exploit code was present in the source data, so no quantitative exploitation signal is available and the assessment rests on the vendor advisory text alone (no public exploit identified at time of analysis).
TCP connection exhaustion in PowerDNS Recursor lets a remote client drive the resolver to hold excessive numbers of sockets in TIME_WAIT state, degrading or blocking further DNS-over-TCP service. The flaw affects Recursor release trains 5.3.x before 5.3.12 and 5.4.x before 5.4.8 and is one of seven CVEs bundled in PowerDNS Security Advisory 2026-12, where it is rated Medium/Low rather than High. No authentication is needed (assessed vector AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L), impact is limited to availability with no effect on data integrity or confidentiality, and no public exploit code or active exploitation was identified at time of analysis; no EPSS or KEV data was supplied.
Out-of-bounds read in PowerDNS Recursor's findScopeZero routine affects releases prior to 5.3.12 and 5.4.8 and is reachable by unauthenticated remote clients only when the non-default use-incoming-edns-subnet option is enabled and a scope-zero answer is produced for a query. Triggering the flaw yields limited information exposure and a plausible denial of service via process crash, reflected in the assessed CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:L vector, where the high attack complexity stems from the narrow configuration and answer-shape prerequisites rather than attacker-side effort. No public exploit code, EPSS score, or CISA KEV listing was identified at time of analysis, so exploitation has not been observed and quantitative risk signals are absent; this CVE is one of six issues bundled in PowerDNS Security Advisory 2026-12 and is among the lower-severity members of that set.
PowerDNS Recursor releases before 5.3.12 and 5.4.8 can be tied up by a client that opens a connection to the built-in web server/REST API and never completes the TLS handshake, leaving the single web handler occupied and denying web/API access to legitimate users. The flaw is one of seven issues fixed in PowerDNS Security Advisory 2026-12 (tracked as CVE-2026-85557) and only matters where the web server is enabled and reachable, since it is disabled or bound to loopback by default; DNS resolution, cache integrity and DNSSEC validation are unaffected. No public exploit code, EPSS score or CISA KEV listing was provided, and no vendor CVSS vector accompanies this specific CVE, so the recorded impact is limited to availability of the web interface (assessed CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L).
DNSSEC validation bypass and cache poisoning in PowerDNS Recursor lets remote, unauthenticated attackers inject forged DNS records into a validating resolver's cache by abusing the out-of-band NS resolution path, where delegation and NS data obtained during recursion is accepted without adequate DNSSEC validation (assessed CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:H/A:N - integrity-only impact, high attack complexity). Affected are the 5.3.x train prior to 5.3.12 and the 5.4.x train prior to 5.4.8; the flaw is only practically reachable when an attacker can induce the recursor to resolve a name in a zone whose authoritative responses they control or spoof, for example by luring a client query for an attacker-chosen domain. No public exploit code or confirmed active exploitation was identified at time of analysis, and the EPSS score was not available in the supplied data, but the integrity impact on DNSSEC-validating deployments makes this the highest-priority item in advisory SA-2026-12.
Remote unauthenticated denial of service in BIND 9 recursive resolvers can be triggered by a crafted query sequence when at least two RPZ zones are configured with conflicting recursive-only settings. The issue affects BIND 9.16.0 through 9.18.50, 9.20.0 through 9.20.29, 9.21.0 through 9.21.26, and corresponding -S1 subscription releases, causing an assertion failure and server exit when an RD=1 query is followed by an RD=0 query. ISC advisories reference fixed releases 9.20.30 and 9.21.27; no public exploit code or CISA KEV entry was identified at time of analysis, and exposure is limited to recursive servers with the uncommon mixed-RPZ configuration.
Assertion failure and null-pointer abort in ISC BIND 9 recursive resolvers terminate the named process, denying DNS resolution to every downstream client until the service is restarted. Only resolvers configured with BOTH a 'redirect' zone AND 'qname-minimization strict' are affected - a non-default, operator-chosen combination that sharply narrows the realistically exposed population, so despite the 7.5 CVSS base score all qualitative signals point to a low real-world priority. The trigger is unauthenticated and remote (PR:N/AV:N, no user interaction), but no public exploit identified at time of analysis and no active exploitation has been reported; vendor fixes are published as 9.20.30 and 9.21.27.
PHPOffice PHPPresentation through version 1.2.0 will pass attacker-controlled serialized data to unserialize() when IOFactory::load() is pointed at a crafted presentation file, letting remote unauthenticated attackers instantiate arbitrary PHP objects inside the hosting application. Where a usable gadget chain exists in the application or its dependencies, that object instantiation escalates to code execution or arbitrary file writes; without gadgets the realistic outcome is object injection or denial of service, which is why the assessed vector marks a present attack requirement (AT:P) rather than an unconditional network exploit. No confirmed active exploitation (CISA KEV) and no public exploit identified at time of analysis, but the library is widely embedded in document-conversion and upload pipelines, so exposure depends entirely on whether the vulnerable Serialized reader path is reachable from untrusted input.
Authenticated NoSQL injection in Trudesk through 1.2.11 lets any low-privilege account manipulate the MongoDB $where clause in GET /api/v1/tickets/search, causing high-impact denial of service by looping mongod and exposing full ticket documents including internal notes. The flaw requires a valid account, so exploitation is not possible pre-auth, and no public exploit was identified at time of analysis. Risk is moderate and centered on malicious or compromised low-privilege users rather than anonymous internet attackers.
Remote unauthenticated denial of service affects Apache bRPC versions before 1.18.0 when the optional RTMP service is enabled, where a single crafted RTMP Abort control message that names the chunk stream currently being parsed triggers a heap use-after-free and crashes the server. Exploitation requires the non-default ServerOptions.rtmp_service configuration and network reachability to the RTMP listener; with that precondition, no credentials, user interaction, or application-level request are needed. The assessed impact is availability-only (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H), and no public exploit code, EPSS score, or CISA KEV entry was identified at time of analysis.
Memory exhaustion in Apache bRPC versions prior to 1.18.0 lets remote attackers force excessive preallocation on servers that terminate Baidu ubrpc traffic through a NsheadService wired to UbrpcMcpack2Adaptor and parse it with code generated by the mcpack2pb protoc plugin. Because the mcpack request body is parsed before authentication (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H), a single unauthenticated request declaring an array item count such as 0x7fffffff can drive roughly 16GB of virtual memory reservation and may abort the process (SIGABRT) on memory-constrained hosts, making this an availability-only denial of service (CWE-770). There is no CVSS score or EPSS probability in the source data, the CVE is not in CISA KEV, and no public exploit code is referenced, so exploitation status is no public exploit identified at time of analysis; the vulnerable adaptor wiring is non-default, so stock bRPC servers that do not use it are unaffected.
Memory and CPU exhaustion in Apache bRPC's Redis-protocol handler lets remote unauthenticated clients inflate server resource usage by sending long inline commands that never terminate with CRLF. The parser sizes its allocation from the current connection buffer and copies the whole buffer before it has located the inline command boundary, because the redis_max_allocation_size guard was enforced only for RESP bulk strings and arrays, not for the inline path. Every bRPC release before 1.18.0 is affected when the optional Redis protocol is enabled and reachable by untrusted clients; impact is limited to denial of service (availability only), with no code execution or data disclosure reported. No public exploit code was identified at time of analysis and there is no confirmed active exploitation (CISA KEV), though the assessed vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L reflects a low-complexity, unauthenticated network trigger.
Memory exhaustion in Apache bRPC's RTMP AMF parser allows unauthenticated remote attackers to consume excessive resources on servers that opt into the RTMP component. Versions before 1.18.0 accept attacker-declared long-string lengths and array element counts without first confirming the payload contains enough bytes, so a small AMF message can force oversized allocations before parsing fails. RTMP is not enabled by default, and no public exploit has been identified at time of analysis.
Denial of service in Apache bRPC servers that accept HTTP/2 traffic is caused by a malformed DATA frame with the PADDED flag set and a zero-length payload. An unauthenticated remote attacker with network reachability to an h2c- or TLS-ALPN-exposed HTTP/2 listener can trigger an integer underflow and out-of-bounds read, leading to high availability impact; versions before 1.18.0 are affected. No public exploit was identified at time of analysis, and the independent assessment rates the issue as network-reachable, low complexity, and requiring no privileges or user interaction.
Denial of service in Apache bRPC before 1.18.0 allows an unauthenticated remote RTMP peer to crash the bRPC process by sending deeply nested AMF objects or arrays, exhausting the native stack. Exploitation is limited to deployments that use bRPC's RTMP support and parse AMF data from untrusted peers, such as RTMP servers handling client command messages or RTMP clients handling server responses, metadata, or cue points. No CISA KEV confirmation or public exploit code was identified at time of analysis.
Remote denial of service in Apache bRPC (versions before 1.18.0) allows a malicious, compromised, or faulty Redis server peer to crash a bRPC Redis client process by returning a deeply nested RESP array reply, driving the reply parser into unbounded recursion until the stack is exhausted. Applications that use bRPC as a Redis client with ChannelOptions.protocol="redis" and that accept replies from an untrusted or tampered-with Redis peer are affected, while deployments that only ever connect to trusted, operator-controlled Redis instances are effectively not exposed. No authentication or user interaction on the client is required to trigger the crash once a crafted reply is delivered; no public exploit code, EPSS score, or CISA KEV listing was identified at time of analysis.
Remote unauthenticated denial of service in Apache bRPC before 1.18.0 allows attackers to exhaust the server thread stack by sending a crafted HTTP/2 header block containing many consecutive HPACK Dynamic Table Size Update entries, potentially crashing or hanging the process. The root cause is uncontrolled recursion (CWE-674) in bRPC's built-in HPACK decoder, reachable when the built-in HTTP/2 server accepts traffic from untrusted networks. No public exploit code was identified at time of analysis, and the assessed CVSS vector is AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H, so impact is limited to availability.
Denial of service in NLnet Labs Unbound versions 1.22.0 through 1.26.1, where a remote DNS-over-QUIC client can crash the resolver by triggering a QUIC path migration at the moment a mesh reply still holds a pointer to the associated DoQ stream. The resulting use-after-free dereferences the freed stream when the pending reply completes, terminating the process and interrupting DNS resolution for all clients of that instance. Exploitation is race-dependent and only possible when DoQ is enabled and Unbound runs multi-threaded; no public exploit code was identified and the issue is not confirmed actively exploited (CISA KEV), so this is a genuine but bounded availability risk rather than a wormable or data-exposing flaw.
Denial of service in Eclipse OpenJ9 through 0.63.0 affects deployments that enable the optional JITServer mode, where unauthenticated remote attackers or a malicious JITServer peer can crash the daemon or a client JVM with malformed protocol messages. The vendor CVSS 4.0 score is 8.7 (High), and the assessed CVSS 3.1 vector AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H confirms remote, low-complexity, availability-only impact with no confidentiality or integrity loss. JITServer is not enabled in default JVM operation, so standard OpenJ9 installations are not exposed; no public exploit or CISA KEV entry was identified at time of analysis.
Denial of service in Eclipse OpenJ9's optional JITServer component (versions prior to 0.63.0) allows an unauthenticated remote attacker to crash the server process by sending a crafted DataDescriptor message. The flaw is an unsigned integer underflow caused by missing cross-field validation of _size, _paddingSize, and _dataOffset, leading to an out-of-bounds heap read and SIGSEGV. Exploitation requires JITServer to be enabled and network-reachable; no public exploit code or CISA KEV entry was identified at time of analysis, and the assessed impact is availability-only (CVSS 3.1 base 7.5), narrower than the headline CVSS 4.0 score of 8.7 suggests.
Remote denial of service in Eclipse OpenJ9 JITServer before 0.63.0 lets unauthenticated network attackers exhaust compilation worker threads by sending crafted TCP connections with a malicious 4-byte length prefix. The integer overflow in MessageBuffer::computeRequiredCapacity() drives a worker thread into an infinite loop at roughly 100% CPU, so a small number of connections can silently stop compilation services for all connected client JVMs. The vendor base score of 8.7 reflects availability-only impact; no public exploit was identified at time of analysis.
Unauthenticated remote attackers can exhaust the daily rate-limit allowance for payment gateway verification in the Formidable Forms WordPress plugin on sites running versions 6.5 through 6.x with a payment gateway configured, causing legitimate verification attempts to be refused for up to 24 hours. The plugin records the rate-limit timestamp without verifying the caller's identity, so the issue is a bounded availability impact against the verification function rather than a full site outage or system compromise. Publicly available exploit code exists, a vendor patch is available in version 7.0, and the independently assessed impact is availability-only.
Remote denial of service in Apache bRPC through 1.17.0: a peer speaking HTTP/2 or gRPC over the h2 protocol can abort the server process by filling the HPACK dynamic table's fixed-size ring buffer. Because the table is sized for a 34-byte entry while the true minimum entry is 33 bytes, 121 minimal header entries (121 x 33 = 3993 bytes) fit within the default 4096-byte table without triggering byte-usage-based eviction, and the 121st insertion trips CHECK(!_header_queue.full()) and kills the process. The h2 parser is enabled by default on every bRPC port and HPACK decoding happens before any Authenticator check, so exposure to an untrusted network is the only precondition; there is no data exposure or code execution, only availability loss (assessed CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H). No CISA KEV entry, no public exploit code, and no EPSS score were provided with this record, so active exploitation status is unconfirmed and no public exploit was identified at time of analysis.
Remote denial of service in Apache bRPC through 1.17.0 allows an unauthenticated HTTP/2 or gRPC peer to abort the server process by sending a header block that compresses into 121 minimal 33-byte HPACK entries. Because the default 4096-byte dynamic table evicts based on byte usage (121 x 33 = 3993 bytes, still under the cap) while its ring buffer is sized for only 120 slots, the 121st AddHeader trips CHECK(!_header_queue.full()) and terminates the process. No public exploit code or CISA KEV listing was identified at time of analysis, but the h2 HPACK decoder runs on the shared server port before any Authenticator check, so every reachable h2/gRPC listener on an affected version is exposed without credentials or user interaction.
Remote denial of service in Apache bRPC through 1.17.0 allows attackers to crash servers or exhaust memory by sending crafted HTTP/2 (h2) frames to the default-enabled h2 parser, which processes peer-controlled stream and header metadata before authentication. The vendor advisory describes remote unauthenticated exploitation; the assessed impact is availability-only (CVSS 3.1 AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H). No public exploit code or CISA KEV entry was identified at time of analysis.
Apache bRPC through 1.17.0 can be remotely aborted by an unauthenticated attacker who sends a crafted HTTP/2 HEADERS block that exhausts the HPACK dynamic-table ring buffer while staying under the byte-based eviction threshold. The h2 parser is enabled by default on bRPC server ports and processes headers before Authenticator checks, so any exposed h2/gRPC listener is sufficient; the assessed impact is availability-only (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H). No public exploit identified at time of analysis, and no KEV listing or EPSS score was present in the supplied data.
Remote unauthenticated denial of service in Apache bRPC's HPACK decoder can abort the server process when a peer sends a crafted HTTP/2 header block. bRPC through 1.17.0 is affected because the default 4096-byte HPACK dynamic table allocates about 120 ring-buffer slots while eviction is tracked by bytes, allowing 121 minimal 33-byte entries to fill the queue and trip a CHECK assertion on the next AddHeader. The h2 parser is enabled by default and runs before authentication, so any exposed HTTP/2 or gRPC listener is at risk; no public exploit code was identified at time of analysis, and 1.18.0 fixes the issue.
Denial of service in Apache bRPC before 1.18.0 allows remote attackers to crash the server by sending deeply nested mcpack data to endpoints serving nshead_mcpack, ubrpc, or public_pbrpc requests. The issue is reachable only when the service uses a recursive/self-referential protobuf schema and the endpoint is exposed to untrusted networks; the assessed vector is AV:N/AC:L/PR:N/UI:N with availability-only impact. No public exploit identified at time of analysis.
Remote unauthenticated denial of service in Apache bRPC through 1.17.0 allows any peer that can reach an HTTP/2 or gRPC (h2) listener to abort the server process with a single crafted HEADERS frame. The HPACK dynamic table's ring buffer is sized by byte capacity rather than slot count, so a peer can insert 121 minimal 33-byte entries (3,993 bytes, still under the default 4,096-byte budget) without triggering eviction and then hit a CHECK(!_header_queue.full()) assertion that terminates the process. The h2 parser is enabled by default on every bRPC server port and decodes headers before any Authenticator check, so no credentials, user interaction, or non-default configuration are required; no public exploit code was identified at time of analysis, but a vendor fix exists and the flaw was addressed together with a large batch of other HTTP/2 parsing issues in bRPC 1.18.0.
Remote unauthenticated attackers can crash Apache bRPC 1.17.0 and earlier servers that expose HTTP/2, h2, or gRPC endpoints by sending a crafted HPACK header block that exhausts the dynamic-table ring buffer and triggers an assertion abort. The issue is reachable during HPACK decoding before any Authenticator check, and the assessed CVSS 3.1 score is 7.5 High (AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H). No public exploit was identified at time of analysis; the vendor released bRPC 1.18.0 and an upstream patch in PR #3462.
Denial of service in Apache bRPC through 1.17.0 lets a remote attacker abort the server process by sending a single crafted HTTP/2 header block that overruns the HPACK decoder's dynamic-table slot accounting. Because the h2 parser runs on the shared listening port before any Authenticator check, no credentials are needed whenever an h2/gRPC-enabled port is reachable from an untrusted network; 121 minimal 33-byte entries fit inside the 4096-byte byte budget but exceed the 120 available ring-buffer slots, tripping CHECK(!_header_queue.full()) and terminating the process. Impact is availability-only (no disclosure or modification), the condition is removed in bRPC 1.18.0, and no public exploit identified at time of analysis.
Remote pre-authentication denial of service in Apache bRPC through 1.17.0 lets any network peer that can reach an HTTP/2 or gRPC (h2) listener abort the server process. The HPACK decoder sizes its default 4096-byte dynamic table for 33-byte minimum entries (roughly 120 slots) but evicts only on byte usage, so 121 minimal 33-byte insertions (3993 bytes) fit under the byte cap while exhausting the fixed slot ring buffer; the 121st AddHeader trips CHECK(!_header_queue.full()) and terminates the process. No public exploit identified at time of analysis, and because the h2 parser decodes headers before any Authenticator check, no credentials or user interaction are required.
A crafted HPACK header block can abort Apache bRPC 1.17.0 and earlier that serve HTTP/2 (h2) or gRPC on a network-reachable port. The default 4096-byte HPACK dynamic table is sized for 34-byte entries, so only about 120 ring-buffer slots exist while eviction is driven by byte usage; 121 minimal 33-byte headers (3,993 bytes total) are therefore accepted without evicting anything, and the 121st insertion trips CHECK(!_header_queue.full()) and terminates the process. The flaw is availability-only and unauthenticated per the assessed vector (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H) because decoding happens before any Authenticator runs; it is fixed in bRPC 1.18.0, and no CISA KEV listing, EPSS score, or public exploit code was supplied in the input data, so no public exploit has been identified at time of analysis.
Remote denial of service in Apache bRPC 1.17.0 and earlier allows unauthenticated network peers to drive disproportionate memory consumption or abort the process by sending input whose declared or decompressed size is never validated. Five parser paths are affected: gzip/zlib/snappy decompression (only the compressed body is counted against -max_body_size, so a 64 MiB cap-compliant body can expand to tens of GiB), AMF string readers that allocate the wire-declared length before the bytes arrive, RTMP chunk headers that re-declare a message length mid-stream into an ever-growing reassembly buffer, the unbounded meta_size field of the sofa-pbrpc parser, and mcpack2pb, which reserves on wire-declared item counts and underflows its buffer size when a string value_size is 0, triggering an abort. The flaw is fixed in 1.18.0 and no public exploit code was identified at time of analysis; reachability is limited to deployments that actually accept compressed bodies or rtmp/sofa-pbrpc/mcpack2pb traffic from untrusted clients, since these parsers run before any Authenticator check.
Unauthenticated remote attackers can crash an Apache bRPC server running version 1.17.0 or earlier by sending a crafted HPACK header block that fills the decoder's dynamic-table ring buffer past its capacity, tripping an internal CHECK() assertion and aborting the process. The flaw affects any bRPC deployment that exposes an HTTP/2 or gRPC (h2) listener to an untrusted network, since HPACK decoding happens before any Authenticator check and the h2 parser is enabled by default on bRPC ports. Impact is limited to availability (process termination); no code execution or data exposure is described, and no public exploit code was identified at time of analysis. This is one of 35 fixes bundled into the bRPC 1.18.0 security release, which also addresses related h2 resource-consumption and decompression issues.
Denial of service in adm-zip 0.6.0 and earlier lets remote attackers terminate the entire Node.js host process by submitting a single malformed ZIP file to any of the library's asynchronous decompression APIs. The flaw sits in inflateAsync (methods/inflater.js), which pipes attacker-controlled DEFLATE bytes into a zlib.createInflateRaw stream without registering an 'error' listener, so a corrupted stream or an over-large inflation surfaces as an uncatchable uncaught exception that kills the process and every in-flight request with it. Only applications that call readFileAsync, readAsTextAsync, extractAllToAsync, or ZipEntry.getDataAsync on untrusted input on versions <= 0.6.0 are exposed; the synchronous APIs are unaffected, and there is no data disclosure or code execution - impact is availability only. No public exploit code was identified at the time of analysis, and the issue is not listed in CISA KEV.
Dulwich before 1.2.9 can be driven into an infinite recursion loop by a crafted packfile containing an OFS_DELTA object whose delta_offset is 0, causing CPU exhaustion during clone, fetch, cat-file, or internal object lookups. The flaw is availability-only and requires the victim or service to process an attacker-supplied packfile; publicly available exploit code exists.
Remote unauthenticated memory exhaustion in vLLM 0.7.3 through versions before 0.31.0 allows clients to terminate shared multimodal inference servers by supplying oversized per-request mm_processor_kwargs, such as large image sizes, patch counts, crop counts, or frame counts, to processors including Isaac, InternVL, and H2OVL. Deployments are exposed when they serve a multimodal model and accept client-controlled mm_processor_kwargs without validation; text-only models and hardened configurations that disable or clamp these parameters are not affected. No public exploit was identified at time of analysis, and the impact is availability-only, with the highest practical risk for multi-tenant or internet-exposed inference APIs.
Unauthenticated memory exhaustion in vLLM versions after 0.24.0 and before 0.31.0 can crash the API process when the MOSS-Audio model is served, because request-controlled mm_processor_kwargs values are cached in an unbounded dictionary. Attackers can send distinct or oversized mel_config values to the /tokenize endpoint, which bypasses --api-key authentication, steadily consuming host memory until the service dies. No public exploit identified at time of analysis; the vendor-released fix is vLLM 0.31.0, and only deployments actively serving MOSS-Audio with a reachable /tokenize endpoint are exposed.
Denial of service in vLLM before 0.31.0 allows remote attackers to crash the entire serving instance by forging or reusing a client-supplied multimodal UUID so the P0/P1 processor caches disagree on token placeholders and hit a fatal mismatch. The flaw is reachable only when the deployment serves a multimodal model and exposes the multimodal inference API; text-only vLLM deployments are not affected. No public exploit was identified at time of analysis, and authentication requirements are conflicting in the supplied data: the CVSS 4.0 vector is PR:L while the independent assessment uses PR:N and the description says unauthenticated, so exposure depends on whether the endpoint is protected (vLLM's default server has no authentication per the assessment).
Client-side denial of service in FreeRDP 3.0.0 through 3.31.x occurs when the smartcard emulation backend processes a server-supplied SCardGetStatusChange reader-state entry with a null szReader name, causing a null-pointer dereference in strcmp/_wcscmp that crashes the RDP client process. Exploitation is server-to-client: an attacker-controlled or compromised RDP server can trigger the crash against a victim who connects using a FreeRDP client with the smartcard channel/emulation enabled, requiring the user to initiate or accept the session. Impact is limited to availability (client crash) with no code execution or data disclosure, and no public exploit has been identified at time of analysis.
Denial of service in FreeRDP 3.32.0: a use-after-free in the drdynvc dynamic virtual channel allows a malicious RDP server to crash the client during teardown by sending one crafted Virtual Channel PDU with a totalLength smaller than the chunk size. The victim must initiate an RDP session to the attacker-controlled or compromised server (UI:P), so passive clients are not exposed; the assessed impact is availability-only with no confidentiality or integrity loss. No public exploit code was identified at time of analysis, and this CVE is not in CISA KEV.
IBM QRadar SIEM 7.6.0.0 through 7.6.0.4 and 7.5.0 through 7.5.0 Update Package 16 Interim Fix 001 contains an out-of-bounds read that a network-reachable attacker can trigger without authentication or user interaction. Successful exploitation can crash or degrade the service for a high availability impact and may expose limited adjacent memory, yielding low confidentiality impact; no public exploit has been identified at time of analysis and no KEV status is provided.
Unauthenticated remote resource exhaustion in IBM QRadar SIEM allows a network-reachable attacker to degrade or take down the security monitoring platform without credentials or user interaction, per the CVSS 3.1 vector (AV:N/AC:L/PR:N/UI:N, availability-only impact). The vendor scored the issue 8.6 with Scope:Changed, while our independent assessment treats scope as unchanged (7.5-equivalent), since no downstream component beyond the QRadar service itself is demonstrated; either way the practical effect is loss of SIEM visibility while the attack runs. No public exploit code or confirmed active exploitation was identified at time of analysis, and the affected version range is not disclosed in the supplied data - it must be confirmed against IBM advisory node 7292244 before exposure decisions are made.
Oracle Essbase 21.8.3.0.0 can be hung or repeatedly crashed by unauthenticated remote attackers who can reach the Security component over TCP, producing a complete denial of service with no confidentiality or integrity impact. The CVSS 3.1 vector is AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H (7.5), so no credentials or user interaction are required according to the supplied assessment. No public exploit identified at time of analysis, and no CISA KEV entry is present; exposure is highest for internet-reachable or broadly accessible Essbase deployments.
Permanent denial of service in Motor.NET before 0.19.0: any party able to publish to a queue or topic that a Motor.NET service consumes can crash that service and keep it down. The service deserializes incoming payloads with JsonNetDeserializer.Deserialize, and structurally valid JSON that does not match the handler's expected type raises a JsonSerializationException that the code does not catch, so TypedConsumerService.SingleMessageConsumeAsync classifies the message as a CriticalFailure and calls StopApplication. Because the offending message is never acknowledged, it is redelivered on restart and re-triggers the shutdown, producing a crash-loop that persists until the message is removed or the library is upgraded; the CVSS 4.0 score is 8.7 (AV:N/AC:L/PR:N/UI:N/VA:H) with no confidentiality or integrity impact, and no public exploit code was identified at time of analysis.
A single crafted Kafka record can permanently crash-loop a Motor.NET service running any version before 0.19.0 that consumes a topic with CloudEvents message handling. The record declares content-type application/cloudevents+json but omits the mandatory CloudEvent attributes id, source and type, causing an ArgumentException inside KafkaMessageConsumer.SingleMessageHandlingAsync that the framework treats as fatal; the host stops, and because the same poison record is re-consumed on every restart, the service never recovers until the message is removed or the software is upgraded. The authoritative CVSS 4.0 vector (8.7, AV:N/AC:L/AT:N/PR:N/UI:N/VA:H) rates this as a high-availability-impact, unauthenticated denial of service with no confidentiality or integrity exposure; the key environmental precondition is that the attacker can produce records to the consumed topic (Kafka-level produce authorization). No public exploit code or confirmed active exploitation was identified at time of analysis, and the vendor has released 0.19.0 as the complete fix.
Denial of service in Motor.NET before 0.19.0 lets anyone able to publish to a queue the service consumes crash it permanently by sending a message with crafted CloudEvents headers - an invalid cloudEvents:motorversion value, a header whose value is a single double-quote, or an unparsable typed attribute such as cloudEvents:time. The malformed header raises an unhandled exception in ExtractCloudEvent that the host treats as fatal; because the message is never acknowledged, the broker redelivers it after restart and the crash repeats as a permanent crash-loop. The impact is availability-only (CVSS 8.7, VA:H) and requires broker-side publish access rather than authentication to Motor.NET itself; no public exploit was identified at time of analysis.
Quadratic-time parsing in Mistune's inline link parser (all versions through 3.2.1) lets remote attackers stall server-side markdown rendering by submitting a small payload made of long runs of unmatched '[' characters. Each '[' triggers a probe of parse_link_text/parse_link_label, so work grows O(N^2): roughly 30+ seconds of CPU for a 16 KB input and proportionally worse as input grows, with the nested '[[..]]' shape costing a further ~4x. The vendor advisory reports a reproducible benchmark proof-of-concept against mistune 3.2.1 in default configuration (no plugins or options required); there is no CISA KEV listing and no confirmed in-the-wild exploitation, and the impact is CPU time only - no memory growth, no confidentiality or integrity effect - so sustained denial requires repeated requests to exhaust a worker or thread pool, most acutely on single-threaded WSGI deployments.
Remote unauthenticated attackers can exhaust server CPU by submitting markdown with deeply nested square brackets to Mistune 3.0.0 through 3.2.1, where the parse_link_text helper degrades to quadratic time and a roughly 16KB payload can occupy a worker for over a minute. Impact is availability-only with no confidentiality or integrity effect, and exposure is bounded by whether an endpoint actually passes untrusted markdown to mistune. A public proof-of-concept reference exists in the upstream advisory; it is not verified as working exploit code, there is no CISA KEV entry, and the this_week decision label reflects a credible but non-emergency priority.
Quadratic-time CPU exhaustion in Mistune 3.0.0 through 3.2.1 allows unauthenticated remote attackers to tie up a rendering worker for minutes using only a few tens of kilobytes of Markdown built from unbalanced '[' or '![' characters. The flaw is reachable from every documented default entry point (mistune.html, mistune.create_markdown(...), Markdown.parse) in both the default escape=False renderer and the escape=True safe mode, with no plugin or non-default option required, so any application that renders attacker-supplied Markdown is exposed. Impact is limited to availability - no confidentiality, integrity, or cross-process effect (scope Unchanged) - and a public proof-of-concept with reproducible timing measurements exists; deployments that require authentication before accepting Markdown effectively raise the bar to PR:L (CVSS 3.1 ~6.5 per the advisory's conservative variant). No CISA KEV listing or confirmed active exploitation is reported at time of analysis; upgrading to 3.3.0 is the fix.
Quadratic denial of service in Mistune's Markdown parser (parse_link_text) affects versions 3.2.0, 3.2.1 and main prior to the 3.3.0 fix: a string consisting of many unmatched opening square brackets followed by a character and many closing brackets causes repeated full forward scans, so cost grows with the square of input length. Per the Python package advisory, roughly 25 KB of brackets stalls a worker past 15 seconds on an Apple M2 and about 50 KB stretches past two minutes, letting remote attackers wedge any server-side caller of mistune.html() that renders attacker-influenced markdown, such as comment fields, web forms or federated content feeds. The CVSS 4.0 base score is 8.7 (AV:N/AC:L/AT:N/PR:N/UI:N, high availability only), exploit code in the form of a timing benchmark is published in the advisory, and no CISA KEV entry or other active-exploitation signal is present in the supplied data.
Signals point to a real but bounded availability risk rather than a top-tier emergency: the MongoDB C# Driver's BSON library interns every BsonSymbol produced during deserialization and retains those symbols for the lifetime of the process with no cap, so an unauthenticated remote party able to get attacker-influenced BSON or JSON deserialized can drive unbounded memory growth and deny service to the host application (availability-only; assessed CVSS 3.1 AV:N/AC:H/PR:N/UI:N/C:N/I:N/A:H, CNA CVSS 4.0 score 8.2). Exposure is conditional - the application must actually deserialize untrusted content through the driver and that content must carry BSON symbol values; merely having the driver installed is not a trigger, and no MongoDB server component is affected. No public exploit code or confirmed active exploitation has been identified at time of analysis; version 3.13.0 removes symbol interning as the fix.
Authenticated clients of the Crawl4AI Docker API server (0.9.0 through 0.9.4) can freeze the entire service, including the /health endpoint, by submitting a catastrophic-backtracking regular expression in a /crawl request body through RegexExtractionStrategy.custom, RegexChunking.patterns, or a schema field of type 'regex' in the JSON extraction strategies. Python's re engine does not release the GIL while matching, so a single crafted pattern combined with an attacker-influenced page blocks the event loop and the crawl deadline cannot interrupt it. Exploitation requires a valid API bearer token and a crawl target whose text triggers the backtracking; a public proof-of-concept request is published in the vendor advisory, no active exploitation has been reported, and the vendor fixed the issue in 0.9.5. The library imported directly is not exposed unless the host application itself feeds untrusted regexes into these strategies.
CPU exhaustion in libexpat through 2.9.0 lets remote attackers stall XML parsing workloads by supplying a moderately sized document with a large DTD and nested entity definitions, because the billion-laughs amplification check in accountingDiffTolerated() aborts too late and each parse can burn tens of seconds of CPU. Unauthenticated remote attackers (CVSS:4.0 AV:N/AC:L/AT:N/PR:N/UI:N) can trigger it wherever an application passes untrusted XML to libexpat with DTD processing and internal general/parameter entity expansion enabled; impact is limited to availability, with no disclosure or modification. No confirmed active exploitation is known, though a public PoC reference exists; it is not verified as working exploit code, and vendor patch material is available upstream (PR #1433).
Unbounded memory leak in the msgpack-python C extension (msgpack._cmsgpack) allows a remote attacker to exhaust process memory by repeatedly sending malformed MessagePack maps, with approximately 4 MiB retained per failed parse and roughly 100 MiB after 25 failed parses as measured by the reporter. The flaw is triggered when strict_map_key=True (the default) rejects a map key that is neither str nor bytes: the decoded value has already been constructed and is never released, affecting both unpackb() and the streaming Unpacker API in releases up to 1.2.2. No public exploit code or active exploitation is identified in the available data, but the CVSS of 7.5 is availability-only (C:N/I:N/A:H), so impact is limited to denial of service rather than data disclosure or code execution; a public proof-of-concept reference exists in the advisory, though it is not verified as working exploit code.
SmileCDR HAPI FHIR can be driven into heap exhaustion and service crashes by unauthenticated attackers who submit highly compressed payloads that are expanded during server-side decompression. The CVSS:4.0 score of 8.7 (assessed as CVSS:3.1 AV:N/AC:L/PR:N/UI:N/C:N/I:N/A:H) reflects an availability-only impact: no confidentiality or integrity loss, no privileges or user interaction required, but a high availability hit against network-reachable FHIR endpoints that decompress untrusted input. No public exploit code has been identified at the time of analysis and the CVE is not listed in CISA KEV; the vuln.today decision label is this_week, so treatment should be scheduled in the current patch cycle rather than treated as an emergency, and deployed instances that accept compressed request bodies should be prioritized.
PhpSpreadsheet's SpreadsheetML reader (Reader\Xml) can be induced to pass a DOCTYPE/ENTITY block into SimpleXML despite its entity-blocking scanner, because the scanner's html_entity_decode callback runs after the pattern check and the content is scanned twice. Files containing a double-encoded less-than character (&#60; or &#x3C;) smuggle a literal '<!DOCTYPE' past both passes, so processing attacker-controlled XML with the Xml reader yields entity-expansion denial of service on all PHP versions, plus local file read or SSRF where external entities are still resolved (PHP before 8.0, libxml2 before 2.9, or any PHP build with LIBXML_NOENT enabled). Affected releases are 3.3.0 through 3.10.7 and 4.0.0 through 5.9.0; vendor fixes are released as 3.10.8 and 5.10.0. No CISA KEV listing or confirmed active exploitation is reported, but the vendor advisory itself publishes a proof-of-concept that was reportedly exercised against PhpSpreadsheet 5.9.0, and the priority signal places remediation in the this-week band rather than emergency.
Denial of service in the PhpSpreadsheet ODS reader allows unauthenticated remote attackers to exhaust a PHP worker's CPU and memory by loading a small crafted .ods file, terminating the process. Every maintained branch is affected - 1.x up to and including 1.30.6, 2.0-2.1 up to 2.1.18, 2.2-2.4 up to 2.4.7, 3.x up to 3.10.7 and 4.x-5.x up to 5.9.0 - whenever an application parses an attacker-supplied OpenDocument Spreadsheet using the default read filter, since no authentication or special configuration is required. A public proof-of-concept payload is documented in the vendor advisory; it is not verified as working exploit code, and no confirmed active exploitation (CISA KEV) is recorded at time of analysis.
PhpSpreadsheet's XLSX reader can be tricked into parsing attacker-supplied DOCTYPE and entity declarations, because the XML security scanner in XmlScanner::scan() only anticipates NUL-byte interleaving and is defeated by ISO-2022 escape sequences inserted into the declaration. Any application that accepts user-uploaded spreadsheets and processes them with the Xlsx reader is affected, and the prerequisite is only that the file be opened; the CVSS vector shows PR:N/UI:N, so no authentication or interaction beyond the file upload is needed. The reliably exploitable outcome is entity-expansion denial of service on all supported PHP versions (a ~2 KB XLSX produced ~45 MB peak memory in the advisory's testing, roughly a 22,000:1 amplification); local file read and SSRF through external SYSTEM entities are conditional on PHP < 8.0 / libxml2 < 2.9 or on the application explicitly passing LIBXML_NOENT. A public proof-of-concept is published in the GitHub security advisory at https://github.com/PHPOffice/PhpSpreadsheet/security/advisories/GHSA-3x94-m842-937c; there is no CISA KEV confirmation of active exploitation, and the priority decision is to patch this week.
Excessive CPU consumption during DNS name decompression in dnspython before 2.9.0 lets remote attackers degrade or deny service to applications that decode untrusted DNS wire-format messages. The parser function dns.name.from_wire_parser follows attacker-controlled compression pointer chains without a hop limit, so a single crafted message can force long pointer traversal; no authentication, user interaction, or non-default configuration is required once untrusted DNS data reaches the parser (CVSS 3.1 AV:N/AC:L/PR:N/UI:N, A:H). The exposure is bounded by architecture rather than by exploit complexity: resolvers, DNS clients and any service that decodes DNS queries or responses from untrusted peers are affected, while systems that only parse trusted internal DNS data are not. No public exploit was identified at time of analysis and the issue is not in CISA KEV; the vendor-released patch is version 2.9.0, and the vendor states there are no known workarounds.
Denial of service in the Python GraphQL-core library (3.0.0 through 3.2.11, plus 3.3.0 pre-releases before 3.3.0rc1) lets an unauthenticated remote attacker pin a worker process for minutes with a single shallow, highly compressible query. The OverlappingFieldsCanBeMerged rule, which ships in specified_rules and is therefore enabled by default on every GraphQL-core server, re-compares flattened inline fragments at every nesting level and can be driven into quadratic - worst case O(n²·m²) - CPU work during validation, before any resolver, depth limiter or complexity analyzer runs; the vendor advisory documents a 364 KB query consuming one core for over 300 seconds on 3.2.8. Severity is availability-only (no confidentiality or integrity impact) and the CVSS 3.1 vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H scores 7.5; the advisory includes a proof-of-concept query generator with measured timings, which is a public PoC reference rather than independently verified working exploit code, and there is no confirmed active exploitation (no CISA KEV listing) at time of analysis. Fixed versions 3.2.12 and 3.3.0rc1 enforce a 250,000-comparison budget.
Memory exhaustion in GraphQL-core 3.2.3 through 3.2.11 and 3.3.0a3 through 3.3.0rc0 lets a remote unauthenticated attacker OOM-kill a GraphQL worker by sending a highly compressible comment-padded document that sails past the parser's max_tokens guard. The lexer still allocates and chains a Token object for every comment it skips, but those tokens were invisible to the counter, so a 38 MB payload parsed under max_tokens=5 produced roughly +92 MB of resident memory per the vendor's measurements. The vendor advisory ships a proof-of-concept that demonstrates the bypass; the flaw is not listed in CISA KEV and no in-the-wild exploitation is reported, and the vendor rates the impact as availability-only (CVSS 7.5, A:H, no confidentiality or integrity effect).
Crafted HTTP requests sent to Next.js App Router applications that define at least one Server Action can drive excessive CPU consumption and block processing of further requests in the same Node.js process, producing a denial of service rated 8.2 (CWE-770, uncontrolled resource allocation). The affected ranges are 13.0.0 through 15.5.27 and 16.0.0 through 16.4.0; applications on the Pages Router, and App Router applications with no Server Actions defined, are explicitly not vulnerable, which substantially narrows the exposed population. Exploitation requires no authentication and no user interaction, and no public exploit code has been identified at time of analysis, with no CISA KEV entry and no EPSS signal supplied; the vendor states that upgrading to 15.5.28 or 16.4.1 is the only complete fix, with no workaround available.
Unauthenticated memory exhaustion in LiteSpeed's lsquic HTTP/3 stack (all versions before 4.9.0) allows a remote attacker to drive a server or client into high memory use and ultimately denial of service by sending header blocks that pile up completed header sets faster than the application claims them, a technique the reporters call a QPACK amplification attack. Any lsquic-based endpoint that exposes HTTP/3 over UDP/QUIC is affected with default settings, since the library applied no cap on buffered header sets; no credentials or user interaction are needed and the vector scores 7.5 (AV:N/AC:L/PR:N/UI:N, availability-only). There is confirmed vendor activity in the form of upstream pull request 666 and the 4.9.0 release notes, and a public research write-up from FoxIO describes the attack class, but no public exploit code was identified at time of analysis and the issue is not in CISA KEV.
Denial of service in body-parser before 1.20.9 and in 2.0.0 through 2.3.x allows any remote unauthenticated client to terminate a Node.js or Express process that uses default body parsing. A request with a Content-Encoding such as gzip and trailing bytes after a valid compressed frame causes the finished decompression stream to emit an error with no error listener, resulting in an uncaught exception. The vendor has released fixes in 2.4.0 and 1.20.9; no public exploit code was identified in the supplied intelligence at time of analysis.
Applications that configure the multer upload middleware (npm package, versions 2.0.0 through 2.4.x) with an asynchronous fileFilter can be crashed by a single aborted multipart request, taking down the entire Node.js process. When a request is aborted while one file's async fileFilter callback is still pending, that callback later resumes a parsing stream multer has already torn down, and the underlying busboy parser emits an error on a stream with no remaining error listener, which Node raises as an unhandled 'error' event and exits. The bug is unauthenticated and requires no user interaction (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H, 7.5), but it is conditional: uploads using a synchronous fileFilter, and the storage engine choice, do not matter. No public exploit code has been identified at time of analysis, and the issue is fixed in multer 2.5.0.
Unauthenticated remote attackers can crash an Erlang node running the Dropbox esaml SAML library (versions 0.1.0 and later) by sending a crafted SAMLResponse to the SAML assertion-consumer (ACS) or single-logout (SLO) endpoint. Because esaml_binding:decode_response/2 hands the decoded XML to xmerl_scan:string/2 before any signature check, every distinct element name, attribute name, and namespace prefix in the document is interned as a permanent Erlang atom; a few dozen requests carrying documents with thousands of unique names exhaust the roughly one-million-entry atom table and abort the VM. The CNA rates this CVSS 4.0 at 8.7, while independent assessment aligns with a CVSS:3.1 vector of AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H - pure availability impact with no confidentiality or integrity loss; no public exploit code was identified at time of analysis and there is no CISA KEV listing, though the flaw is trivially reachable by anyone who can POST to a SAML endpoint.
Unauthenticated memory exhaustion in the Erlang esaml SAML library (version 0.1.0 and later) lets a remote attacker crash the hosting Erlang VM by posting a small, highly compressed SAMLResponse to a service provider's assertion-consumer (ACS) or single-logout endpoint. Because esaml_binding:decode_response/2 base64-decodes and inflates the attacker-supplied payload with zlib:unzip/1 before any signature check, a deflate stream with roughly 1000:1 amplification expands tens of kilobytes into hundreds of megabytes of binary and then gigabytes of charlist heap, so only a handful of requests are needed to take the node down. No public exploit code or confirmed active exploitation was identified at time of analysis, but the issue is trivially reachable on any internet-facing esaml service provider and upstream fix commits are already published.
Unauthenticated remote denial of service in the Erlang SAML library esaml (0.1.0 and later) allows an attacker to abort an Erlang node by exhausting its atom table through unsigned LogoutResponse messages posted to a Service Provider's single-logout endpoint. Because esaml_sp:validate_logout_response/2 only verifies a signature when one is present, an unsigned message reaches esaml:decode_logout_response/1, where status_code_map/1 converts each unrecognized StatusCode into a permanent atom with list_to_atom/1 - roughly one million distinct status values (the VM's default atom-table ceiling) are enough to crash the node. The CNA scored this CVSS 4.0 8.2 with the AT:P attack-requirement metric; the practical barrier is volume and time rather than skill, and no public exploit code or confirmed active exploitation was identified at time of analysis.
A truncated HTTP/2 PING frame sent to an HAProxy frontend that accepts HTTP/2 causes the H2 multiplexer to spin in a busy loop, driving worker-thread CPU to saturation and degrading or denying service to legitimate traffic (CVSS 7.5, availability-only). Only deployments with an HTTP/2-enabled listener (h2 in the bind line / ALPN negotiation) are reachable; HTTP/1.x-only setups are unaffected, and no authentication or user interaction is required against an H2 endpoint. Sustained impact depends on the attacker holding many concurrent connections open, since each connection is eventually cleared by 'timeout http-request'; no public exploit code was identified at the time of analysis and there is no KEV confirmation of active exploitation.
HAProxy 3.x releases before 3.4.6 can be driven into persistent denial of service by remote unauthenticated attackers sending a single crafted HTTP/1.1 chunked request: the H1 multiplexer's chunk parser re-processes the same bytes without advancing, pinning one worker thread at 100% CPU even after timeouts fire and the client disconnects. Repeating the request exhausts the worker pool until the process is manually restarted, so confidentiality and integrity are unaffected but availability is fully lost. Attack surface is standard HTTP/1.1 frontend traffic (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H, CWE-835); no public exploit code or CISA KEV entry is associated with this CVE at time of analysis, and the upstream fix commit is published in the HAProxy git repository.
Unauthenticated resource exhaustion in the Drupal Webform contrib module (all versions from 0.0.0 through 6.2.12, plus 6.3.0 through 6.3.1) lets remote attackers drive excessive resource allocation and degrade or deny availability of a Drupal site; confidentiality and integrity are unaffected (CVSS 3.1 7.5, AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H). The flaw is CWE-770 (allocation without limits or throttling), and the vendor's stated remediation is to update the module per Drupal security advisory sa-contrib-2026-170. Risk signals are moderate rather than urgent: EPSS is 0.19% (8th percentile), there is no CISA KEV listing, and no public exploit has been identified at time of analysis, though the priority label places remediation in the current week.
Unauthenticated denial of service in Splunk Enterprise deployments below 10.4.4, 10.2.8, and 10.0.11 is reachable through the Edge Processor package download endpoint in the Splunk Pipeline Builders app. Repeated unauthenticated requests force the Splunk daemon to read and base64-encode the complete package in memory for each request, exhausting system resources and potentially making the service stop responding (CVSS 7.5, CWE-400). Exposure is limited to instances running Edge Processor/Pipeline Builders, and no public exploit identified at time of analysis.
Splunk Enterprise deployments running Splunk Web in versions below 10.4.4, 10.2.8, 10.0.11, and 9.4.16 can be taken offline by a single unauthenticated request to the Splunk Web JavaScript logger endpoint, which parses request headers without any limit and can be driven into an unresponsive state. Recovery requires a full restart of Splunk Enterprise, making this an easy-to-repeat availability attack against any instance whose Splunk Web port is reachable by the attacker. The assessed vector (AV:N/AC:L/PR:N/UI:N, availability-only) is consistent with a low-complexity denial of service; no public exploit code or confirmed active exploitation was identified at time of analysis, and the risk is bounded to service availability with no implied data loss or compromise.
Denial of service in toml-node versions prior to 5.0.1 allows remote attackers to synchronously block the Node.js event loop by supplying a TOML document containing an extremely long hexadecimal, octal, binary, or decimal integer literal. Because toml.parse() calls BigInt() on the digit sequence before the 64-bit range check runs in reduceInteger(), conversion cost grows super-linearly with digit count, so a single crafted document can stall all concurrent work in the process (CVSS 3.1: 7.5, AV:N/AC:L/PR:N/UI:N, availability-only). The flaw is exploitable only where an application parses attacker-influenced TOML; there is no confidentiality or integrity impact, no public exploit code identified at time of analysis, and the issue is fixed in 5.0.1.
Denial of service in Django REST Framework prior to 3.18.2 allows a remote unauthenticated client to pin synchronous workers for prolonged periods by sending Accept headers packed with a large number of media-type tokens carrying oversized quoted parameter values and many semicolons; the Content-Type header is a secondary, now length-capped, vector. Impact is availability-only (CVSS 3.1 AV:N/AC:L/PR:N/UI:N/A:H, 7.5) and self-healing once requests stop, but on WSGI/sync worker deployments (e.g. gunicorn sync workers) concurrent requests can exhaust the worker pool and take the service offline. No public exploit code and no active exploitation are indicated in the supplied intelligence, and no EPSS score was provided, so this reads as a genuine but bounded availability risk rather than an emergency.
Apache Traffic Server 9.2.15, 10.1.4, and 10.2.0 can be crashed by a remote, unauthenticated attacker through a null pointer dereference (CWE-476) while proxying chunked origin responses. The race requires a client POST request to be torn down via TCP half-close at the moment ATS reparses a chunked origin response from a reentrant write-ready callback, so it is timing-dependent and typically needs repeated attempts; impact is limited to denial of service (crash), with no code execution, data disclosure, or integrity loss. No public exploit was identified at time of analysis, and vendor-released fixed versions are 9.2.16, 10.1.5, and 10.2.1.
Apache Traffic Server 9.2.15, 10.1.4, and 10.2.0 can be driven into a denial-of-service condition through log amplification: each internal '@' header stripped from a request causes an unthrottled diagnostic log line, so requests carrying many such headers generate disproportionate log writes. The issue is network-reachable and unauthenticated, with high availability impact when log storage, rotation, or retention is constrained; no public exploit identified at time of analysis.
Memory exhaustion in Pillow's SPIDER image plugin lets an attacker-supplied multi-frame .spider stack file bypass the library's decompression-bomb guard and drive allocations of several gigabytes. The flaw lives in SpiderImageFile.seek(), which reparses a later frame and overwrites self._size without re-running Image._decompression_bomb_check(), so a tiny file with a benign first frame passes Image.open() and then declares enormous dimensions in a later mode-F frame; loading that frame consumes roughly four bytes per attacker-declared pixel. All Pillow releases prior to 13.0.0 that parse SPIDER stacks are affected, and exploitation requires the application to open an untrusted multi-frame SPIDER file and then seek past frame 0 - a niche, passive-interaction path consistent with the assessed CVSS 3.1 vector AV:N/AC:L/PR:N/UI:R/S:U/C:N/I:N/A:H. No public exploit code or confirmed active exploitation was identified at time of analysis, and the vendor patch is 13.0.0.
Out-of-bounds read in Pillow's TIFF encoder before 13.0.0 can disclose process memory or crash the application when TIFF tag values are inconsistent with the actual image data. The issue is not triggered by simply opening a malicious image; exploitation requires an application to save a TIFF using attacker-influenced tag metadata, and the assessed vector (AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:L) reflects that high precondition. The Pillow 13.0.0 release notes rank it High severity, but no CISA KEV listing, EPSS score, or public exploit code was provided, so no public exploit identified at time of analysis.
Denial of service in Pillow's legacy TarIO parser (all versions before 13.0.0) allows remote unauthenticated attackers to supply a crafted TAR archive that traps the parser in an uncontrolled loop, consuming a full CPU core and never returning. The exposure is sharply limited because only applications that directly call PIL.TarIO.TarIO on attacker-controlled archives are affected; the default Image.open() path does not reach this parser. No public exploit code was identified at time of analysis, and the raw 7.1 CVSS 4.0 score should be weighed against the non-default API prerequisite and required user or service interaction to supply the malicious archive.
Denial of service in Pillow's PDF append workflow (versions 5.1.0 through 12.x) lets an unauthenticated remote attacker exhaust CPU by supplying a compact malformed PDF whose /Pages /Kids arrays repeatedly reference the same child /Pages node. When an application calls Image.save(..., format="PDF", append=True) on that attacker-controlled file, PdfParser.linearize_page_tree() expands the shared subtree as if it were distinct pages, producing exponentially many page references before the append completes. Impact is availability-only (C:N/I:N/A:H) and reachability is narrow - ordinary image decoding and standalone PDF export never hit this code path - so no public exploit code has been identified at time of analysis and the issue is fixed in Pillow 13.0.0.
Denial of service in Apache Traffic Server 8.0.0 through 8.1.11, 9.0.0 through 9.2.15, 10.0.0 through 10.1.4, and 10.2.0 allows unauthenticated remote attackers to exhaust a downstream HTTP/1.1 origin's worker pool by sending a truncated HTTP/2 Content-Length value that ATS mishandles when relaying the request. The issue is limited to proxy deployments where ATS accepts HTTP/2 client connections and forwards traffic to an HTTP/1.1 origin; no public exploit code or confirmed active exploitation was identified at time of analysis.
A stale-state flaw in Zephyr's experimental MCTP-over-I2C+GPIO target binding lets any device that can act as an I2C controller on the shared management bus crash the target with two ordinary transactions, because the selected pseudo-register and data-phase flag are never reset at STOP. The resulting NULL pointer dereference occurs inside the I2C target interrupt handler, so on parts that trap address-0 accesses it produces a fatal exception and device reset or halt; on other parts it degrades to a malformed packet buffer reaching the MCTP stack and silently dropped transmit packets. Only builds with CONFIG_MCTP and CONFIG_MCTP_I2C_GPIO_TARGET=y are exposed, exploitation requires a pre-existing foothold on the same physical bus (compromised peer endpoint or BMC), and no public exploit code was identified at time of analysis; the authoritative assessment scores this as availability-only impact (CVSS:3.1/AV:A/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H) rather than the C:L/I:N variant in the published vector.
Crafted IPv4 packets addressed to the router itself can drive the line card network processor on Cisco IOS XR into repeated resets, causing traffic loss and - under sustained attack - a persistent shutdown state that requires manual line card reload to recover. The flaw affects carrier-grade hardware specifically: Cisco ASR 9000 and 9900 Series Aggregation Services Routers plus the ASR 9902 and 9903 Compact High-Performance Routers. No authentication or user interaction is needed (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H, 7.5), and no public exploit code or confirmed active exploitation was identified at time of analysis; exposure is bounded by the fact that transit IPv4 traffic and all IPv6 traffic cannot trigger the fault.
Denial of service in Cisco Expressway Series allows unauthenticated remote attackers to crash the service by triggering a null pointer dereference (CWE-476). The issue is network-reachable, low complexity, and requires no user interaction, with impact limited to availability; exact affected and fixed releases must be confirmed in Cisco's advisory. No public exploit code was identified at time of analysis, and exposure is greatest for internet-facing or externally reachable Expressway deployments.
Network-reachable assertion failures in Cisco Expressway Series (TelePresence VCS Expressway) can terminate video and collaboration signaling services, allowing remote attackers to cause a denial-of-service condition without authentication or user interaction (CVSS 7.5). The vendor advisory describes multiple internally discovered CWE-617 reachable-assertion issues fixed in hardening releases; the exact affected and fixed versions are not specified in the provided data. No public exploit was identified at time of analysis, and there is no CISA KEV confirmation of active exploitation.
Denial of service in the Cisco Snort 3 detection engine allows remote, unauthenticated attackers to exhaust device resources and disrupt inspection services on any deployment actively processing attacker-influenced traffic - including Cisco Secure Firewall Threat Defense (FTD), the Cisco UTD Snort IPS engine, Cisco Cyber Vision, and upstream Open Source Snort 3. The flaw is rooted in uncontrolled resource consumption (CWE-400) and carries a CVSS 3.1 base score of 7.5 (AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H), meaning only availability is impacted - there is no confidentiality or integrity loss and no code execution. No public exploit code or in-the-wild activity has been identified at time of analysis, and no CISA KEV listing exists; exploitation is gated by the requirement that the target actually sits inline as an IDS/IPS or firewall in a traffic-inspection path.
Unbounded resource retention in PowerDNS Recursor lets remote clients drive memory exhaustion by repeatedly opening and closing DNS-over-TCP connections. Recursor 5.3.x before 5.3.12 and 5.4.x before 5.4.8 retain entries for closed TCP client addresses until process exit, so the effect is cumulative over long uptimes and can degrade or deny DNS service; the independent assessment is availability-only (CVSS 3.1: 5.3, AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L). No public exploit code or CISA KEV entry was identified at time of analysis.
Denial of service in PowerDNS Recursor 5.3.x before 5.3.12 and 5.4.x before 5.4.8, where a crafted DNS packet containing malicious name-compression pointers can drive runaway decompression of DNS names and exhaust the recursor's CPU, degrading or halting resolution service (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H, 7.5 High). Unauthenticated remote attackers can trigger it either by sending the packet straight to a reachable recursor port or by luring the recursor into resolving a name hosted on an attacker-controlled authoritative server that returns the malformed response - so open or internet-reachable resolvers carry the most risk, while recursors restricted to trusted internal clients are far less exposed. The flaw ships as part of PowerDNS Security Advisory 2026-12, a seven-CVE bundle whose most serious member (CVE-2026-85555) is a DNSSEC validation bypass and cache poisoning; no CVSS vector, EPSS score, KEV listing or public exploit code was present in the source data, so no quantitative exploitation signal is available and the assessment rests on the vendor advisory text alone (no public exploit identified at time of analysis).
TCP connection exhaustion in PowerDNS Recursor lets a remote client drive the resolver to hold excessive numbers of sockets in TIME_WAIT state, degrading or blocking further DNS-over-TCP service. The flaw affects Recursor release trains 5.3.x before 5.3.12 and 5.4.x before 5.4.8 and is one of seven CVEs bundled in PowerDNS Security Advisory 2026-12, where it is rated Medium/Low rather than High. No authentication is needed (assessed vector AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L), impact is limited to availability with no effect on data integrity or confidentiality, and no public exploit code or active exploitation was identified at time of analysis; no EPSS or KEV data was supplied.
Out-of-bounds read in PowerDNS Recursor's findScopeZero routine affects releases prior to 5.3.12 and 5.4.8 and is reachable by unauthenticated remote clients only when the non-default use-incoming-edns-subnet option is enabled and a scope-zero answer is produced for a query. Triggering the flaw yields limited information exposure and a plausible denial of service via process crash, reflected in the assessed CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:L/I:N/A:L vector, where the high attack complexity stems from the narrow configuration and answer-shape prerequisites rather than attacker-side effort. No public exploit code, EPSS score, or CISA KEV listing was identified at time of analysis, so exploitation has not been observed and quantitative risk signals are absent; this CVE is one of six issues bundled in PowerDNS Security Advisory 2026-12 and is among the lower-severity members of that set.
PowerDNS Recursor releases before 5.3.12 and 5.4.8 can be tied up by a client that opens a connection to the built-in web server/REST API and never completes the TLS handshake, leaving the single web handler occupied and denying web/API access to legitimate users. The flaw is one of seven issues fixed in PowerDNS Security Advisory 2026-12 (tracked as CVE-2026-85557) and only matters where the web server is enabled and reachable, since it is disabled or bound to loopback by default; DNS resolution, cache integrity and DNSSEC validation are unaffected. No public exploit code, EPSS score or CISA KEV listing was provided, and no vendor CVSS vector accompanies this specific CVE, so the recorded impact is limited to availability of the web interface (assessed CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L).
DNSSEC validation bypass and cache poisoning in PowerDNS Recursor lets remote, unauthenticated attackers inject forged DNS records into a validating resolver's cache by abusing the out-of-band NS resolution path, where delegation and NS data obtained during recursion is accepted without adequate DNSSEC validation (assessed CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:H/A:N - integrity-only impact, high attack complexity). Affected are the 5.3.x train prior to 5.3.12 and the 5.4.x train prior to 5.4.8; the flaw is only practically reachable when an attacker can induce the recursor to resolve a name in a zone whose authoritative responses they control or spoof, for example by luring a client query for an attacker-chosen domain. No public exploit code or confirmed active exploitation was identified at time of analysis, and the EPSS score was not available in the supplied data, but the integrity impact on DNSSEC-validating deployments makes this the highest-priority item in advisory SA-2026-12.
Remote unauthenticated denial of service in BIND 9 recursive resolvers can be triggered by a crafted query sequence when at least two RPZ zones are configured with conflicting recursive-only settings. The issue affects BIND 9.16.0 through 9.18.50, 9.20.0 through 9.20.29, 9.21.0 through 9.21.26, and corresponding -S1 subscription releases, causing an assertion failure and server exit when an RD=1 query is followed by an RD=0 query. ISC advisories reference fixed releases 9.20.30 and 9.21.27; no public exploit code or CISA KEV entry was identified at time of analysis, and exposure is limited to recursive servers with the uncommon mixed-RPZ configuration.
Assertion failure and null-pointer abort in ISC BIND 9 recursive resolvers terminate the named process, denying DNS resolution to every downstream client until the service is restarted. Only resolvers configured with BOTH a 'redirect' zone AND 'qname-minimization strict' are affected - a non-default, operator-chosen combination that sharply narrows the realistically exposed population, so despite the 7.5 CVSS base score all qualitative signals point to a low real-world priority. The trigger is unauthenticated and remote (PR:N/AV:N, no user interaction), but no public exploit identified at time of analysis and no active exploitation has been reported; vendor fixes are published as 9.20.30 and 9.21.27.
PHPOffice PHPPresentation through version 1.2.0 will pass attacker-controlled serialized data to unserialize() when IOFactory::load() is pointed at a crafted presentation file, letting remote unauthenticated attackers instantiate arbitrary PHP objects inside the hosting application. Where a usable gadget chain exists in the application or its dependencies, that object instantiation escalates to code execution or arbitrary file writes; without gadgets the realistic outcome is object injection or denial of service, which is why the assessed vector marks a present attack requirement (AT:P) rather than an unconditional network exploit. No confirmed active exploitation (CISA KEV) and no public exploit identified at time of analysis, but the library is widely embedded in document-conversion and upload pipelines, so exposure depends entirely on whether the vulnerable Serialized reader path is reachable from untrusted input.
Authenticated NoSQL injection in Trudesk through 1.2.11 lets any low-privilege account manipulate the MongoDB $where clause in GET /api/v1/tickets/search, causing high-impact denial of service by looping mongod and exposing full ticket documents including internal notes. The flaw requires a valid account, so exploitation is not possible pre-auth, and no public exploit was identified at time of analysis. Risk is moderate and centered on malicious or compromised low-privilege users rather than anonymous internet attackers.
Remote unauthenticated denial of service affects Apache bRPC versions before 1.18.0 when the optional RTMP service is enabled, where a single crafted RTMP Abort control message that names the chunk stream currently being parsed triggers a heap use-after-free and crashes the server. Exploitation requires the non-default ServerOptions.rtmp_service configuration and network reachability to the RTMP listener; with that precondition, no credentials, user interaction, or application-level request are needed. The assessed impact is availability-only (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H), and no public exploit code, EPSS score, or CISA KEV entry was identified at time of analysis.
Memory exhaustion in Apache bRPC versions prior to 1.18.0 lets remote attackers force excessive preallocation on servers that terminate Baidu ubrpc traffic through a NsheadService wired to UbrpcMcpack2Adaptor and parse it with code generated by the mcpack2pb protoc plugin. Because the mcpack request body is parsed before authentication (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H), a single unauthenticated request declaring an array item count such as 0x7fffffff can drive roughly 16GB of virtual memory reservation and may abort the process (SIGABRT) on memory-constrained hosts, making this an availability-only denial of service (CWE-770). There is no CVSS score or EPSS probability in the source data, the CVE is not in CISA KEV, and no public exploit code is referenced, so exploitation status is no public exploit identified at time of analysis; the vulnerable adaptor wiring is non-default, so stock bRPC servers that do not use it are unaffected.
Memory and CPU exhaustion in Apache bRPC's Redis-protocol handler lets remote unauthenticated clients inflate server resource usage by sending long inline commands that never terminate with CRLF. The parser sizes its allocation from the current connection buffer and copies the whole buffer before it has located the inline command boundary, because the redis_max_allocation_size guard was enforced only for RESP bulk strings and arrays, not for the inline path. Every bRPC release before 1.18.0 is affected when the optional Redis protocol is enabled and reachable by untrusted clients; impact is limited to denial of service (availability only), with no code execution or data disclosure reported. No public exploit code was identified at time of analysis and there is no confirmed active exploitation (CISA KEV), though the assessed vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L reflects a low-complexity, unauthenticated network trigger.
Memory exhaustion in Apache bRPC's RTMP AMF parser allows unauthenticated remote attackers to consume excessive resources on servers that opt into the RTMP component. Versions before 1.18.0 accept attacker-declared long-string lengths and array element counts without first confirming the payload contains enough bytes, so a small AMF message can force oversized allocations before parsing fails. RTMP is not enabled by default, and no public exploit has been identified at time of analysis.
Denial of service in Apache bRPC servers that accept HTTP/2 traffic is caused by a malformed DATA frame with the PADDED flag set and a zero-length payload. An unauthenticated remote attacker with network reachability to an h2c- or TLS-ALPN-exposed HTTP/2 listener can trigger an integer underflow and out-of-bounds read, leading to high availability impact; versions before 1.18.0 are affected. No public exploit was identified at time of analysis, and the independent assessment rates the issue as network-reachable, low complexity, and requiring no privileges or user interaction.
Denial of service in Apache bRPC before 1.18.0 allows an unauthenticated remote RTMP peer to crash the bRPC process by sending deeply nested AMF objects or arrays, exhausting the native stack. Exploitation is limited to deployments that use bRPC's RTMP support and parse AMF data from untrusted peers, such as RTMP servers handling client command messages or RTMP clients handling server responses, metadata, or cue points. No CISA KEV confirmation or public exploit code was identified at time of analysis.
Remote denial of service in Apache bRPC (versions before 1.18.0) allows a malicious, compromised, or faulty Redis server peer to crash a bRPC Redis client process by returning a deeply nested RESP array reply, driving the reply parser into unbounded recursion until the stack is exhausted. Applications that use bRPC as a Redis client with ChannelOptions.protocol="redis" and that accept replies from an untrusted or tampered-with Redis peer are affected, while deployments that only ever connect to trusted, operator-controlled Redis instances are effectively not exposed. No authentication or user interaction on the client is required to trigger the crash once a crafted reply is delivered; no public exploit code, EPSS score, or CISA KEV listing was identified at time of analysis.
Remote unauthenticated denial of service in Apache bRPC before 1.18.0 allows attackers to exhaust the server thread stack by sending a crafted HTTP/2 header block containing many consecutive HPACK Dynamic Table Size Update entries, potentially crashing or hanging the process. The root cause is uncontrolled recursion (CWE-674) in bRPC's built-in HPACK decoder, reachable when the built-in HTTP/2 server accepts traffic from untrusted networks. No public exploit code was identified at time of analysis, and the assessed CVSS vector is AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H, so impact is limited to availability.
Denial of service in NLnet Labs Unbound versions 1.22.0 through 1.26.1, where a remote DNS-over-QUIC client can crash the resolver by triggering a QUIC path migration at the moment a mesh reply still holds a pointer to the associated DoQ stream. The resulting use-after-free dereferences the freed stream when the pending reply completes, terminating the process and interrupting DNS resolution for all clients of that instance. Exploitation is race-dependent and only possible when DoQ is enabled and Unbound runs multi-threaded; no public exploit code was identified and the issue is not confirmed actively exploited (CISA KEV), so this is a genuine but bounded availability risk rather than a wormable or data-exposing flaw.
Denial of service in Eclipse OpenJ9 through 0.63.0 affects deployments that enable the optional JITServer mode, where unauthenticated remote attackers or a malicious JITServer peer can crash the daemon or a client JVM with malformed protocol messages. The vendor CVSS 4.0 score is 8.7 (High), and the assessed CVSS 3.1 vector AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H confirms remote, low-complexity, availability-only impact with no confidentiality or integrity loss. JITServer is not enabled in default JVM operation, so standard OpenJ9 installations are not exposed; no public exploit or CISA KEV entry was identified at time of analysis.
Denial of service in Eclipse OpenJ9's optional JITServer component (versions prior to 0.63.0) allows an unauthenticated remote attacker to crash the server process by sending a crafted DataDescriptor message. The flaw is an unsigned integer underflow caused by missing cross-field validation of _size, _paddingSize, and _dataOffset, leading to an out-of-bounds heap read and SIGSEGV. Exploitation requires JITServer to be enabled and network-reachable; no public exploit code or CISA KEV entry was identified at time of analysis, and the assessed impact is availability-only (CVSS 3.1 base 7.5), narrower than the headline CVSS 4.0 score of 8.7 suggests.
Remote denial of service in Eclipse OpenJ9 JITServer before 0.63.0 lets unauthenticated network attackers exhaust compilation worker threads by sending crafted TCP connections with a malicious 4-byte length prefix. The integer overflow in MessageBuffer::computeRequiredCapacity() drives a worker thread into an infinite loop at roughly 100% CPU, so a small number of connections can silently stop compilation services for all connected client JVMs. The vendor base score of 8.7 reflects availability-only impact; no public exploit was identified at time of analysis.
Unauthenticated remote attackers can exhaust the daily rate-limit allowance for payment gateway verification in the Formidable Forms WordPress plugin on sites running versions 6.5 through 6.x with a payment gateway configured, causing legitimate verification attempts to be refused for up to 24 hours. The plugin records the rate-limit timestamp without verifying the caller's identity, so the issue is a bounded availability impact against the verification function rather than a full site outage or system compromise. Publicly available exploit code exists, a vendor patch is available in version 7.0, and the independently assessed impact is availability-only.
Remote denial of service in Apache bRPC through 1.17.0: a peer speaking HTTP/2 or gRPC over the h2 protocol can abort the server process by filling the HPACK dynamic table's fixed-size ring buffer. Because the table is sized for a 34-byte entry while the true minimum entry is 33 bytes, 121 minimal header entries (121 x 33 = 3993 bytes) fit within the default 4096-byte table without triggering byte-usage-based eviction, and the 121st insertion trips CHECK(!_header_queue.full()) and kills the process. The h2 parser is enabled by default on every bRPC port and HPACK decoding happens before any Authenticator check, so exposure to an untrusted network is the only precondition; there is no data exposure or code execution, only availability loss (assessed CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H). No CISA KEV entry, no public exploit code, and no EPSS score were provided with this record, so active exploitation status is unconfirmed and no public exploit was identified at time of analysis.
Remote denial of service in Apache bRPC through 1.17.0 allows an unauthenticated HTTP/2 or gRPC peer to abort the server process by sending a header block that compresses into 121 minimal 33-byte HPACK entries. Because the default 4096-byte dynamic table evicts based on byte usage (121 x 33 = 3993 bytes, still under the cap) while its ring buffer is sized for only 120 slots, the 121st AddHeader trips CHECK(!_header_queue.full()) and terminates the process. No public exploit code or CISA KEV listing was identified at time of analysis, but the h2 HPACK decoder runs on the shared server port before any Authenticator check, so every reachable h2/gRPC listener on an affected version is exposed without credentials or user interaction.
Remote denial of service in Apache bRPC through 1.17.0 allows attackers to crash servers or exhaust memory by sending crafted HTTP/2 (h2) frames to the default-enabled h2 parser, which processes peer-controlled stream and header metadata before authentication. The vendor advisory describes remote unauthenticated exploitation; the assessed impact is availability-only (CVSS 3.1 AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H). No public exploit code or CISA KEV entry was identified at time of analysis.
Apache bRPC through 1.17.0 can be remotely aborted by an unauthenticated attacker who sends a crafted HTTP/2 HEADERS block that exhausts the HPACK dynamic-table ring buffer while staying under the byte-based eviction threshold. The h2 parser is enabled by default on bRPC server ports and processes headers before Authenticator checks, so any exposed h2/gRPC listener is sufficient; the assessed impact is availability-only (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H). No public exploit identified at time of analysis, and no KEV listing or EPSS score was present in the supplied data.
Remote unauthenticated denial of service in Apache bRPC's HPACK decoder can abort the server process when a peer sends a crafted HTTP/2 header block. bRPC through 1.17.0 is affected because the default 4096-byte HPACK dynamic table allocates about 120 ring-buffer slots while eviction is tracked by bytes, allowing 121 minimal 33-byte entries to fill the queue and trip a CHECK assertion on the next AddHeader. The h2 parser is enabled by default and runs before authentication, so any exposed HTTP/2 or gRPC listener is at risk; no public exploit code was identified at time of analysis, and 1.18.0 fixes the issue.
Denial of service in Apache bRPC before 1.18.0 allows remote attackers to crash the server by sending deeply nested mcpack data to endpoints serving nshead_mcpack, ubrpc, or public_pbrpc requests. The issue is reachable only when the service uses a recursive/self-referential protobuf schema and the endpoint is exposed to untrusted networks; the assessed vector is AV:N/AC:L/PR:N/UI:N with availability-only impact. No public exploit identified at time of analysis.
Remote unauthenticated denial of service in Apache bRPC through 1.17.0 allows any peer that can reach an HTTP/2 or gRPC (h2) listener to abort the server process with a single crafted HEADERS frame. The HPACK dynamic table's ring buffer is sized by byte capacity rather than slot count, so a peer can insert 121 minimal 33-byte entries (3,993 bytes, still under the default 4,096-byte budget) without triggering eviction and then hit a CHECK(!_header_queue.full()) assertion that terminates the process. The h2 parser is enabled by default on every bRPC server port and decodes headers before any Authenticator check, so no credentials, user interaction, or non-default configuration are required; no public exploit code was identified at time of analysis, but a vendor fix exists and the flaw was addressed together with a large batch of other HTTP/2 parsing issues in bRPC 1.18.0.
Remote unauthenticated attackers can crash Apache bRPC 1.17.0 and earlier servers that expose HTTP/2, h2, or gRPC endpoints by sending a crafted HPACK header block that exhausts the dynamic-table ring buffer and triggers an assertion abort. The issue is reachable during HPACK decoding before any Authenticator check, and the assessed CVSS 3.1 score is 7.5 High (AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H). No public exploit was identified at time of analysis; the vendor released bRPC 1.18.0 and an upstream patch in PR #3462.
Denial of service in Apache bRPC through 1.17.0 lets a remote attacker abort the server process by sending a single crafted HTTP/2 header block that overruns the HPACK decoder's dynamic-table slot accounting. Because the h2 parser runs on the shared listening port before any Authenticator check, no credentials are needed whenever an h2/gRPC-enabled port is reachable from an untrusted network; 121 minimal 33-byte entries fit inside the 4096-byte byte budget but exceed the 120 available ring-buffer slots, tripping CHECK(!_header_queue.full()) and terminating the process. Impact is availability-only (no disclosure or modification), the condition is removed in bRPC 1.18.0, and no public exploit identified at time of analysis.
Remote pre-authentication denial of service in Apache bRPC through 1.17.0 lets any network peer that can reach an HTTP/2 or gRPC (h2) listener abort the server process. The HPACK decoder sizes its default 4096-byte dynamic table for 33-byte minimum entries (roughly 120 slots) but evicts only on byte usage, so 121 minimal 33-byte insertions (3993 bytes) fit under the byte cap while exhausting the fixed slot ring buffer; the 121st AddHeader trips CHECK(!_header_queue.full()) and terminates the process. No public exploit identified at time of analysis, and because the h2 parser decodes headers before any Authenticator check, no credentials or user interaction are required.
A crafted HPACK header block can abort Apache bRPC 1.17.0 and earlier that serve HTTP/2 (h2) or gRPC on a network-reachable port. The default 4096-byte HPACK dynamic table is sized for 34-byte entries, so only about 120 ring-buffer slots exist while eviction is driven by byte usage; 121 minimal 33-byte headers (3,993 bytes total) are therefore accepted without evicting anything, and the 121st insertion trips CHECK(!_header_queue.full()) and terminates the process. The flaw is availability-only and unauthenticated per the assessed vector (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H) because decoding happens before any Authenticator runs; it is fixed in bRPC 1.18.0, and no CISA KEV listing, EPSS score, or public exploit code was supplied in the input data, so no public exploit has been identified at time of analysis.
Remote denial of service in Apache bRPC 1.17.0 and earlier allows unauthenticated network peers to drive disproportionate memory consumption or abort the process by sending input whose declared or decompressed size is never validated. Five parser paths are affected: gzip/zlib/snappy decompression (only the compressed body is counted against -max_body_size, so a 64 MiB cap-compliant body can expand to tens of GiB), AMF string readers that allocate the wire-declared length before the bytes arrive, RTMP chunk headers that re-declare a message length mid-stream into an ever-growing reassembly buffer, the unbounded meta_size field of the sofa-pbrpc parser, and mcpack2pb, which reserves on wire-declared item counts and underflows its buffer size when a string value_size is 0, triggering an abort. The flaw is fixed in 1.18.0 and no public exploit code was identified at time of analysis; reachability is limited to deployments that actually accept compressed bodies or rtmp/sofa-pbrpc/mcpack2pb traffic from untrusted clients, since these parsers run before any Authenticator check.
Unauthenticated remote attackers can crash an Apache bRPC server running version 1.17.0 or earlier by sending a crafted HPACK header block that fills the decoder's dynamic-table ring buffer past its capacity, tripping an internal CHECK() assertion and aborting the process. The flaw affects any bRPC deployment that exposes an HTTP/2 or gRPC (h2) listener to an untrusted network, since HPACK decoding happens before any Authenticator check and the h2 parser is enabled by default on bRPC ports. Impact is limited to availability (process termination); no code execution or data exposure is described, and no public exploit code was identified at time of analysis. This is one of 35 fixes bundled into the bRPC 1.18.0 security release, which also addresses related h2 resource-consumption and decompression issues.
Denial of service in adm-zip 0.6.0 and earlier lets remote attackers terminate the entire Node.js host process by submitting a single malformed ZIP file to any of the library's asynchronous decompression APIs. The flaw sits in inflateAsync (methods/inflater.js), which pipes attacker-controlled DEFLATE bytes into a zlib.createInflateRaw stream without registering an 'error' listener, so a corrupted stream or an over-large inflation surfaces as an uncatchable uncaught exception that kills the process and every in-flight request with it. Only applications that call readFileAsync, readAsTextAsync, extractAllToAsync, or ZipEntry.getDataAsync on untrusted input on versions <= 0.6.0 are exposed; the synchronous APIs are unaffected, and there is no data disclosure or code execution - impact is availability only. No public exploit code was identified at the time of analysis, and the issue is not listed in CISA KEV.
Dulwich before 1.2.9 can be driven into an infinite recursion loop by a crafted packfile containing an OFS_DELTA object whose delta_offset is 0, causing CPU exhaustion during clone, fetch, cat-file, or internal object lookups. The flaw is availability-only and requires the victim or service to process an attacker-supplied packfile; publicly available exploit code exists.
Remote unauthenticated memory exhaustion in vLLM 0.7.3 through versions before 0.31.0 allows clients to terminate shared multimodal inference servers by supplying oversized per-request mm_processor_kwargs, such as large image sizes, patch counts, crop counts, or frame counts, to processors including Isaac, InternVL, and H2OVL. Deployments are exposed when they serve a multimodal model and accept client-controlled mm_processor_kwargs without validation; text-only models and hardened configurations that disable or clamp these parameters are not affected. No public exploit was identified at time of analysis, and the impact is availability-only, with the highest practical risk for multi-tenant or internet-exposed inference APIs.
Unauthenticated memory exhaustion in vLLM versions after 0.24.0 and before 0.31.0 can crash the API process when the MOSS-Audio model is served, because request-controlled mm_processor_kwargs values are cached in an unbounded dictionary. Attackers can send distinct or oversized mel_config values to the /tokenize endpoint, which bypasses --api-key authentication, steadily consuming host memory until the service dies. No public exploit identified at time of analysis; the vendor-released fix is vLLM 0.31.0, and only deployments actively serving MOSS-Audio with a reachable /tokenize endpoint are exposed.
Denial of service in vLLM before 0.31.0 allows remote attackers to crash the entire serving instance by forging or reusing a client-supplied multimodal UUID so the P0/P1 processor caches disagree on token placeholders and hit a fatal mismatch. The flaw is reachable only when the deployment serves a multimodal model and exposes the multimodal inference API; text-only vLLM deployments are not affected. No public exploit was identified at time of analysis, and authentication requirements are conflicting in the supplied data: the CVSS 4.0 vector is PR:L while the independent assessment uses PR:N and the description says unauthenticated, so exposure depends on whether the endpoint is protected (vLLM's default server has no authentication per the assessment).
Client-side denial of service in FreeRDP 3.0.0 through 3.31.x occurs when the smartcard emulation backend processes a server-supplied SCardGetStatusChange reader-state entry with a null szReader name, causing a null-pointer dereference in strcmp/_wcscmp that crashes the RDP client process. Exploitation is server-to-client: an attacker-controlled or compromised RDP server can trigger the crash against a victim who connects using a FreeRDP client with the smartcard channel/emulation enabled, requiring the user to initiate or accept the session. Impact is limited to availability (client crash) with no code execution or data disclosure, and no public exploit has been identified at time of analysis.
Denial of service in FreeRDP 3.32.0: a use-after-free in the drdynvc dynamic virtual channel allows a malicious RDP server to crash the client during teardown by sending one crafted Virtual Channel PDU with a totalLength smaller than the chunk size. The victim must initiate an RDP session to the attacker-controlled or compromised server (UI:P), so passive clients are not exposed; the assessed impact is availability-only with no confidentiality or integrity loss. No public exploit code was identified at time of analysis, and this CVE is not in CISA KEV.
IBM QRadar SIEM 7.6.0.0 through 7.6.0.4 and 7.5.0 through 7.5.0 Update Package 16 Interim Fix 001 contains an out-of-bounds read that a network-reachable attacker can trigger without authentication or user interaction. Successful exploitation can crash or degrade the service for a high availability impact and may expose limited adjacent memory, yielding low confidentiality impact; no public exploit has been identified at time of analysis and no KEV status is provided.
Unauthenticated remote resource exhaustion in IBM QRadar SIEM allows a network-reachable attacker to degrade or take down the security monitoring platform without credentials or user interaction, per the CVSS 3.1 vector (AV:N/AC:L/PR:N/UI:N, availability-only impact). The vendor scored the issue 8.6 with Scope:Changed, while our independent assessment treats scope as unchanged (7.5-equivalent), since no downstream component beyond the QRadar service itself is demonstrated; either way the practical effect is loss of SIEM visibility while the attack runs. No public exploit code or confirmed active exploitation was identified at time of analysis, and the affected version range is not disclosed in the supplied data - it must be confirmed against IBM advisory node 7292244 before exposure decisions are made.
Oracle Essbase 21.8.3.0.0 can be hung or repeatedly crashed by unauthenticated remote attackers who can reach the Security component over TCP, producing a complete denial of service with no confidentiality or integrity impact. The CVSS 3.1 vector is AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H (7.5), so no credentials or user interaction are required according to the supplied assessment. No public exploit identified at time of analysis, and no CISA KEV entry is present; exposure is highest for internet-reachable or broadly accessible Essbase deployments.