Command injection in the device-information handler of the TOTOLINK A950RG router (firmware 4.1.2cu.5268_B20221104) allows an attacker who can influence the MAC-address value processed by getDeviceInfo to execute arbitrary shell commands on the device. The CVSS vector requires high privileges (PR:H) to reach the vulnerable code path, and the available sources do not specify how a malicious MAC value reaches /var/lib/misc/udhcpd.leases - whether via unauthenticated LAN DHCP traffic or authenticated configuration - so the precise unauthenticated reachability is not confirmed. A public PoC reference exists but is not verified as working exploit code; there is no confirmed active exploitation (no CISA KEV listing), and EPSS is 2.74% (86th percentile), which places it in the upper band of predicted exploitation likelihood and is consistent with the 'this_week' prioritisation label.
Command injection in the Cloud Upgrade component of TOTOLINK A950RG firmware 4.1.2cu.5268_B20221104 allows a remote attacker with a low-privilege authenticated session to pass a crafted FileName value into an OS command through the setCloudUpgradeFW function. The assessed CVSS 3.1 score is 8.6 (AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H), and EPSS is 1.55% (75th percentile), indicating high technical severity but not top-percentile exploitation likelihood. A public PoC reference exists; it is not verified as working exploit code, and no CISA KEV entry is present.
CRLF injection in the SNMP Handler of Oray Pugongying X5 Pro 7.6.1 lets a low-privileged authenticated user inject newline sequences through the info.community argument handled by snmp_set, potentially causing the root-running snmpd to execute commands as root. The issue is network reachable, requires authentication (CVSS 4.0 score 8.6, PR:L), and affects both the router and subsequent systems; a public PoC reference exists but is not verified as working exploit code, and no CISA KEV active exploitation is reported. The vendor was contacted early about the disclosure and did not respond, and no vendor-released patch is identified at time of analysis.
A CRLF injection (CWE-93) in the L2TP Authentication component of the Oray Pugongying X5 Pro router running firmware 7.6.1 allows an attacker who controls the L2TP Username argument to insert carriage-return/line-feed sequences into authentication handling, which the vendor-tagged data classifies as code injection and the published reference describes as reaching root code execution. The CVSS 4.0 vector scores 8.5 (AV:N/AC:L/AT:N/PR:H/UI:N) with high confidentiality, integrity and availability impact on both the vulnerable device and subsequent systems, but exploitation requires high privileges (PR:H) - authenticated or configuration-level access to the device's L2TP setup - rather than anonymous reach. A public PoC reference exists (a Notion write-up titled for root code execution via L2TP username injection); it is not verified as working exploit code, and there is no CISA KEV listing, though the current priority decision is to remediate this week. The vendor was contacted about the disclosure and did not respond.
Hardcoded static AES key embedded in the TTLock Android app (versions before v9.8.260718) and its associated smart-lock firmware allows an attacker within Bluetooth Low Energy range to capture a pairing exchange and decrypt the per-pairing session key transported during it. With the recovered persistent session key the attacker can forge authenticated BLE commands to unlock the lock, enroll fingerprints or access codes, and carry out administrative actions without authentication or user interaction (CVSS 4.0 8.6, AV:A/AC:L/PR:N/UI:N, VC:H/VI:H). The exposure is bounded by physical proximity to the lock rather than internet reach, and a public PoC reference exists but is not verified as working exploit code; there is no CISA KEV confirmation of active exploitation.
CRLF injection in the ser2net serial-handling component of Oray Pugongying X5 Pro 7.6.1 lets a network-positioned attacker with low-level privileges manipulate the regmsg argument so that an unescaped newline sequence in a serial banner overwrites an existing file on the device. The vendor-supplied CVSS 4.0 score is 8.6 (High); the independently assessed vector is CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:H/A:H, reflecting high integrity and availability impact with lower confidentiality impact, and requiring PR:L - the attacker must already hold a low-privilege foothold/authentication on the device rather than being fully unauthenticated. No CISA KEV entry exists and no EPSS value was supplied; a public PoC reference exists, but it is not verified as working exploit code. The vendor was contacted and did not respond, so no official fix or vendor-stated mitigation is documented.
CRLF injection in the Dnsmasq-related DNS group configuration of Oray Pugongying X5 Pro 7.6.1 can let an attacker with authenticated administrative access inject newline sequences that, per the referenced PoC, are executed as a root script, with high confidentiality, integrity, and availability impact. Exploitation requires administrative access to the DNS group feature plus user interaction, so exposure depends on whether the management interface is reachable and how well admin access is protected; a public PoC reference exists but is not verified as working exploit code, and no CISA KEV confirmation was provided. The vendor did not respond to disclosure and no official fixed version is listed, making this a conditional risk rather than a drop-everything emergency.
CRLF injection in the IPsec Policy Handler of Oray Pugongying X5 Pro 7.6.1 allows an authenticated high-privilege administrator to place unsanitized secproto values into /etc/ipsec.conf and inject additional IPsec configuration directives that execute as root. The CVSS 4.0 score is 8.5 with PR:H and no user interaction, and the independent assessment rates confidentiality, integrity and availability impact as high for both the device and downstream systems. A public PoC reference exists; it is not verified as working exploit code, no CISA KEV entry is provided, the vendor did not respond, and no fixed version is published.
CRLF injection in the Dnsmasq configuration path of Oray Pugongying X5 Pro 7.6.1 lets a user who already holds administrative access to the router's management interface inject newline-delimited Dnsmasq directives through the DHCP lease-time (rent_time) parameter, corrupting the device's DNS/DHCP behaviour and enabling configuration tampering and potential information disclosure. Exploitation is remote over the network with no user interaction, but the CVSS vector carries PR:H - an attacker must first obtain admin credentials, which is the primary limiting factor, so this is not reachable by an unauthenticated outsider. A public PoC reference (a Notion write-up titled 'Dnsmasq Directive Injection Through DHCP Lease Time') exists but is not verified as working exploit code; the vendor was contacted and did not respond, so no fixed version is available, and the disclosure is tracked for remediation this week.
Improper Secure Boot signature verification in Yoosee GM8135S Camera 21.00.01.36 lets a local attacker who already holds high privileges get a manipulated firmware image accepted (CWE-347), undermining firmware trust and enabling tampering or persistence. The CVSS 4.0 base score is 8.4, but the attack path is local and requires existing privileged access, with no remote or unauthenticated vector described; the exact trigger is not detailed in the available sources. A public PoC reference exists; it is not verified as working exploit code, and no CISA KEV entry confirms active exploitation; the disclosure is prioritized this_week.
Unrestricted file upload in the PPPoE WAN Setting Handler of Oray Pugongying X5 Pro 7.6.1 lets an attacker who already holds administrative access to the router management interface manipulate the pppd_options argument so that plugin_init writes attacker-controlled content to /tmp/upload_files, ultimately enabling code execution on the device. Per the vendor CVSS vector the attack is network-reachable and low-complexity but requires high privileges (PR:H), which is the main limiting factor. A public PoC reference exists (a Notion write-up describing root code execution via the PPPoE plugin option) but it is not verified as working exploit code, and the vendor was contacted about the disclosure without responding, so no patch is available. This is a genuine but lower-tier priority rather than a drop-everything emergency, though it should be handled this week.
Command injection in the apmib_set handler of firewall.so on TOTOLINK A950RG routers running firmware 4.1.2cu.5268_B20221104 lets a remote attacker who holds low-privilege credentials execute arbitrary commands on the device, with high impact on confidentiality, integrity and availability. The vulnerable configuration routine is reachable over the network but requires at least low-privilege authentication (PR:L) plus reachability of the management interface, which is typically LAN-side unless remote administration is enabled; no user interaction is involved and attack complexity is low. A public exploit disclosure reference (a GitHub gist) is cited in the advisory and the vendor has not published a fixed firmware build in the available data, while EPSS sits at 1.88% (79th percentile) and priority is rated this_week rather than emergency.
A logic issue was addressed with improved state management. This issue is fixed in macOS Golden Gate 27.2, macOS Sequoia 15.8.2, macOS Tahoe 26.7.2. An unprivileged attacker may be able to execute arbitrary code as root.
An out-of-bounds write issue was addressed with improved bounds checking. This issue is fixed in macOS Golden Gate 27, macOS Sequoia 15.8.2, macOS Tahoe 26.7.2. Processing a maliciously crafted file may lead to arbitrary code execution.
Memory corruption in Apple's iOS, iPadOS and macOS file parsing code can lead to unexpected app termination or arbitrary code execution when a victim opens a maliciously crafted file. Affected are iOS/iPadOS releases before 26.7 and 27.2, macOS Sequoia before 15.8.2, macOS Tahoe before 26.7.2, and macOS Golden Gate before 27.2; the flaw is reachable by a remote unauthenticated attacker (AV:N/PR:N) but requires user interaction (UI:R) to open the file. No public exploit code and no confirmed active exploitation are known at time of analysis, and EPSS is low at 0.64% (50th percentile), though the potential impact of successful exploitation is total per SSVC.
A use-after-free issue was addressed with improved memory management. This issue is fixed in Safari 27.2, iOS 26.7.2 and iPadOS 26.7.2, iOS 27.2 and iPadOS 27.2, macOS Golden Gate 27.2, tvOS 27.2, visionOS 27.2, watchOS 27.2. Processing maliciously crafted web content may lead to arbitrary code execution.
Arbitrary code execution in Apple WebKit can be triggered when a user processes maliciously crafted web content on affected Apple platforms, with successful exploitation leading to memory corruption and potential full compromise of confidentiality, integrity, and availability (CVSS 8.8). The flaw affects Safari before 27.2, iOS/iPadOS before 26.7.2 and before 27.2, macOS Golden Gate before 27.2, and tvOS, visionOS, and watchOS before 27.2; it requires user interaction but no authentication. No confirmed active exploitation or public exploit code was identified at time of analysis, and EPSS is 0.57% (47th percentile).
A type confusion issue was addressed with improved memory handling. This issue is fixed in Safari 27.2, iOS 26.7.2 and iPadOS 26.7.2, iOS 27.2 and iPadOS 27.2, macOS Golden Gate 27.2, visionOS 27.2. Processing maliciously crafted web content may lead to arbitrary code execution.
A use-after-free issue was addressed with improved memory management. This issue is fixed in Safari 27.2, iOS 27.2 and iPadOS 27.2, macOS Golden Gate 27.2, tvOS 27.2, visionOS 27.2, watchOS 27.2. Processing maliciously crafted web content may lead to arbitrary code execution.
Arbitrary code execution in Apple's WebKit rendering engine affects Safari, iOS and iPadOS, macOS Golden Gate, and visionOS users who process maliciously crafted web content. The flaw is a use-after-free (CWE-416) that Apple addressed with improved memory management, and the vendor's only complete fix is updating to the listed releases. No active exploitation is confirmed (CISA SSVC exploitation: none) and no public exploit code is confirmed; EPSS is 0.54% (45th percentile), so this is a serious but not emergency-level patching priority - patch within this week.
A use-after-free issue was addressed with improved memory management. This issue is fixed in Safari 27.2, iOS 26.7.2 and iPadOS 26.7.2, iOS 27.2 and iPadOS 27.2, macOS Golden Gate 27.2, visionOS 27.2. Processing maliciously crafted web content may lead to arbitrary code execution.
Improper access restrictions in Apple macOS before Golden Gate 27.2 allow a maliciously crafted archive to cause unexpected app termination or arbitrary code execution when a user unpacks it. The CVSS vector indicates network reachability, no required privileges (PR:N), and mandatory user interaction (UI:R); no public exploit was identified at time of analysis, and EPSS is 0.51% (44th percentile). Apple has released macOS Golden Gate 27.2 as the vendor fix.
Authenticated inference callers on vLLM scale-out deployments that explicitly enable the shared-memory multimodal processor cache (--mm-processor-cache-type shm) can plant forged address and monotonic_id values inside the public features.kwargs_data envelope, causing a worker to dereference attacker-chosen ring-buffer memory and pass those bytes to pickle.loads(), which yields arbitrary Python execution in the worker process. The chain runs over ordinary public HTTP routes (/v1/chat/completions/render, /v1/chat/completions, /inference/v1/generate) and succeeds even with VLLM_ALLOW_INSECURE_SERIALIZATION=0 because it abuses the built-in SHM object deserializer rather than the optional insecure-serialization flag. Exposure is limited to deployments combining a scale-out topology with the non-default SHM cache mode; the flaw is fixed in vLLM 0.32.0, CISA KEV does not list it, and a proof-of-concept chain is described in the upstream advisory but no public working exploit code was confirmed at time of analysis.
An integer overflow was addressed with improved input validation. This issue is fixed in iOS 26.7.2 and iPadOS 26.7.2, iOS 27 and iPadOS 27, macOS Golden Gate 27, macOS Sequoia 15.8, macOS Tahoe 26.7. Processing a maliciously crafted image may lead to arbitrary code execution.
Unauthenticated ClickHouse SQL injection in OpenPanel before 3.0.0 lets remote attackers read other tenants' analytics data by injecting SQL through the projectId path segment of the /live/visitors/:projectId WebSocket, a handler that skips the session and project-access checks its sibling endpoints enforce. A second, authenticated variant lets any project member inject through the unescaped timeframe start/end values in cohort.preview; in both cases the payload reaches the shared ClickHouse events table that is isolated only by a WHERE project_id predicate, so extraction is cross-tenant. Exploitation requires no credentials for the primary vector and no special configuration, though data recovery is inferential (UNION or boolean/count oracles against a uniq(profile_id) count) rather than a bulk dump; the vendor advisory includes a benign demonstration query, but it is not verified as working exploit code, the vendor has released 3.0.0 as a fix, CISA KEV does not list the issue, SSVC exploitation is 'none', and EPSS is low at 0.43% (37th percentile).
VMime 0.9.2 and earlier contain an integer underflow in uuEncoder::decode that turns a crafted uuencoded MIME part into a stack-based buffer overflow, letting remote attackers crash a mail-processing application or potentially corrupt memory toward code execution. The trigger is a uuencoded line whose length character is 'b' but which is not a valid 'begin' marker: the marker probe consumes more bytes than the short line accounts for, the remaining-length subtraction wraps to a huge value, and the decode loop then writes past its fixed-size stack buffer. Exploitation requires that the target application feeds untrusted uuencoded content to VMime's decoder and that a victim or their mail software actually processes the message (network-reachable, unauthenticated, requires user interaction); no active exploitation is confirmed (SSVC exploitation: none), EPSS is a modest 0.73% (54th percentile), and while an upstream fix exists in PR/commit form, a public technical writeup exists that is not verified as working exploit code.
A permissive CORS misconfiguration in Transmission 4.0.0 through 4.1.2 exposes the RPC anti-CSRF token (X-Transmission-Session-Id) to any web origin, so a malicious page can read the token from the 409 handshake response and replay it to drive the RPC interface as the user. Successful replay allows settings tampering, download-directory changes and torrent or data deletion, and escalates to code execution where the script-torrent-done feature is usable, but only defeats the anti-CSRF check - an RPC password or an enabled IP whitelist blocks the replay. Exploitation requires the RPC web server (default port 9091) to be reachable from the victim's browser and the victim to visit the attacker's page; no KEV entry exists, EPSS is low at 0.53% (44th percentile), and a public researcher write-up with a PoC reference exists but is not verified here as working exploit code.
An authorization bypass in HashiCorp Consul's service mesh proxy registration lets an authenticated user who holds only service:write permission apply Envoy escape-hatch (custom proxy/bootstrap) configuration that is normally reserved for holders of mesh:write, effectively escalating the privileges of an existing mesh participant. Consul Community Edition before 2.1.0 and Consul Enterprise before 2.1.0 and before 1.21.19 are affected; the CVSS 3.1 score is 8.8 (AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H), with the PR:L metric reflecting that a valid ACL token with service:write is required, so this is not anonymous remote compromise. No public exploit code or confirmed active exploitation was identified at time of analysis, and the vendor states the issue is fixed in Consul CE 2.1.0 and Consul Enterprise 2.1.0 and 1.21.19.
Local privilege escalation to root in boneIO Black affects versions prior to 1.6.0, where a low-privileged local attacker can abuse a time-of-check/time-of-use race in the timezone sudoers-file routine to inject arbitrary sudoers rules. Exploitation requires an existing local foothold on the device and either winning a non-deterministic race between validation and installation or pre-creating a symlink at the predictable temporary filename; it is not remotely reachable unauthenticated. No public exploit has been identified at time of analysis, and the vendor-released fix is version 1.6.0.
Server-side template injection in Mautic's Twig theme sandbox lets an authenticated user holding the core:themes:create permission upload a crafted theme that escapes sandbox restrictions and executes arbitrary code as the web server process. The flaw is an incomplete fix for GHSA-9fx4-7cmj-47vg and affects Mautic core branches before 5.2.12, 6.0.10, and 7.2.2. It is a high-priority issue, though exploitation requires privileged theme-creation access; no public exploit identified at time of analysis and no CISA KEV entry is present in the supplied data.
Memory corruption in rt-labs SOEM 2.0.0 and earlier allows a malicious or compromised EtherCAT slave device on the same fieldbus segment to overflow master-side buffers during multi-segment data transfers, writing attacker-controlled data beyond buffer boundaries with potential for code execution. Because SOEM validates only the declared total size at the start of a transfer and never tracks cumulative size as further segments arrive, a device that declares a small initial size can pass validation and then exceed the allocation. The flaw is scored 8.8 (AV:A/AC:L/PR:N/UI:N) - unauthenticated and requiring no user interaction, but gated by an adjacent-network precondition: the attacker must already sit on the EtherCAT segment and be able to act as or subvert a slave. No public exploit code was identified at time of analysis and there is no CISA KEV entry; priority is to address it this week, and fieldbus segmentation significantly bounds exposure.
Cross-site request forgery in the FunnelKit Automations WordPress plugin (wp-marketing-automations) affects every release through 3.8.5.2 and lets an attacker induce a logged-in privileged user - typically an administrator or plugin operator - to submit state-changing requests against the site without their intent. The issue carries CVSS 3.1 score 8.8 (AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:H) because successful forgery can compromise confidentiality and integrity, but it cannot succeed without that victim interaction on an active, authenticated session. Reported by Patchstack and assigned a vuln.today priority of 'this_week'; no public exploit code and no confirmed active exploitation (CISA KEV) are indicated in the available data.
Sandbox escape in OpenPanel's JavaScript webhook-template runtime (packages/js-runtime) allows an authenticated project member holding write or admin access on at least one project to execute arbitrary commands inside the worker process, which holds the database, ClickHouse, Redis and third-party integration credentials. The AST allowlist in validate() only inspects member-expression call targets and assignment left-hand sides, so a template can copy payload.constructor.constructor into a local and invoke it as a plain identifier, and execute() then compiled that template with new Function() in the worker itself. Every build since the JavaScript template feature landed in commit 286f8e16 (2026-01-23) through 61a92453 (2026-09-18) is affected; the trigger requires the attacker to attach a notification rule to the malicious webhook integration and have one matching event ingested via the public /track endpoint, with no user interaction and no special timing. No CISA KEV entry exists and EPSS is modest (0.49%, 42nd percentile), consistent with a flaw gated behind project write access rather than a drive-by remote attack; the vendor advisory nonetheless publishes a working proof-of-concept harness that demonstrates validator bypass and host command execution.
Regular expression denial of service in the Datadog dd-trace-py Python APM client allows unauthenticated remote attackers to exhaust CPU and stall request handling by sending crafted URL query strings. The flaw affects versions 1.5.0 through 3.19.9 and 4.0.0 through 4.15.6, and the vulnerable obfuscation processing is enabled by default for HTTP tracing integrations that record query strings in URL span tags, so no special configuration is required to be at risk. Assessment is availability-only (CVSS 3.1 AV:N/AC:L/PR:N/UI:N/A:H, vendor CVSS 4.0 score 8.7); there is no CISA KEV listing and no public exploit identified at time of analysis, and EPSS sits at 0.47% (40th percentile).
Untrusted zstd input can crash any Go application that decompresses it with klauspost/compress 1.20.1 or earlier: a crafted two-block frame of roughly 27 KB drives the FSE decoder used by huff0.ReadTable into an out-of-range slice index, producing a runtime panic instead of a decode error. Both default decode paths - DecodeAll and the streaming decoder - are affected, and the flaw requires no authentication or user interaction (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H), so impact is availability only; there is no memory corruption, code execution or data exposure. Exploitation is not confirmed (CISA KEV not listed, SSVC exploitation: none), EPSS is a low 0.33%, and a public PoC reference exists in the GitHub advisory as a Go test reproducer; it is not verified as working exploit code beyond that panic reproduction. The vendor ships a fix in 1.20.2, and the issue is assessed as a bounded but genuine denial-of-service, prioritised 'this week'.
Pre-authentication denial of service in Open5GS 1.3.0 through 2.8.0 lets a peer that can reach a network function's SBI listener crash that NF with a single crafted multipart/related request. Because the multipart boundary is accepted when empty, the parser's index never reaches its terminating boundary length and writes linearly past a fixed heap lookbehind buffer, producing a SIGSEGV and an NF crash/restart; the official open5gs:2.8.0 container reproduces it with roughly a 30-byte body under default configuration. Impact is availability only - no data disclosure or code execution is demonstrated - and the exposure is bounded to peers on the internal core SBI plane (SEPP N32 inter-operator traffic is HTTPS with a client CA), so this is a this_week-priority fix rather than an Internet-facing emergency; no CISA KEV listing or public working exploit code was identified, and the advisory describes reproduction steps while stating that a self-contained PoC is held privately on request.
Missing authentication in boneIO Black firmware before 1.6.0 leaves the management web API bound to all network interfaces and accessible without credentials. A network-adjacent, unauthenticated attacker can fully control the device, including causing denial of service, changing configuration, and operating connected physical circuits; no public exploit was identified at time of analysis. Version 1.6.0 adds first-run administrator setup and role-restricted API routes, while devices older than 1.2.0 require re-flashing because they are based on unsupported Debian 10.
boneIO Black home automation controllers running software before version 1.6.0 expose a Node-RED instance through their Caddy reverse proxy at the /nodered/ path without any authentication, and because the Node-RED Admin API permits deploying flows that can invoke system commands, an unauthenticated attacker who can reach the device from the adjacent network can deploy a malicious flow and obtain remote code execution with full control of the home automation system. Exploitation requires local-network adjacency (CVSS AV:A) rather than internet exposure, and no credentials, tokens or user interaction are needed against the default configuration; devices updated to 1.6.0 or later are not affected. No public exploit code has been identified at time of analysis and the issue is not in CISA KEV; it was reported by CERT-PL and assigned a priority of this_week, reflecting a genuine high-priority issue affecting a small-audience embedded product.
Unauthenticated remote CPU exhaustion in mtrudel Bandit's HTTP/2 server (versions before 1.14.0) is achievable by holding thousands of response streams blocked on the connection send window and then sending tiny SETTINGS or stream-0 WINDOW_UPDATE frames. Each such frame makes the server retry every blocked send via do_pending_sends/2, producing O(n²) list operations; the advisory's verified PoC measured 133 ms for a single 9-byte SETTINGS frame with 3,000 blocked streams. The issue is remotely exploitable without authentication or user interaction and is prioritized this_week; the supplied data does not confirm active exploitation, and a public proof-of-concept is included in the advisory.
Unbounded per-connection memory and CPU growth in the Elixir Bandit HTTP/2 server (0.3.6 up to 1.14.0) lets an unauthenticated remote client advertise a SETTINGS_HEADER_TABLE_SIZE of up to 2^32-1, which Bandit applies to its HPACK encoder table with no server-side cap; every response header is then stored into that oversized table and never evicted, so the encoding cost of each response grows linearly and the connection's total cost grows quadratically. No authentication or user interaction is required (PR:N/UI:N), the effect is purely availability (C:N/I:N/A:H, scored 8.7 under CVSS 4.0 and assessed as AV:N/AC:L/PR:N/UI:N/A:H under CVSS 3.1), and it only bites while a connection is kept alive and busy with cheap requests to endpoints whose response headers vary per request. It is not confirmed actively exploited (CISA KEV), but the vendor advisory documents a public proof-of-concept procedure; treat it as a this-week priority rather than an emergency, since capping requests per connection or upgrading to 1.14.0 removes the growth.
Unauthenticated remote attackers can exhaust BEAM node memory on default-configured mtrudel Bandit HTTP/2 servers (0.3.3 before 1.14.0) by sending a small HPACK HEADERS block that expands into a huge joined cookie value before any plug code executes. Concurrent streams multiply the allocation, and the vendor advisory includes a public proof-of-concept description; it is not verified as working exploit code. Bandit 1.14.0 is the vendor-released fix, while setting a finite max_header_list_size or disabling HTTP/2 removes the vulnerable path.
Unauthenticated remote denial of service in the Elixir HTTP server Bandit (mtrudel/bandit) from 0.1.0 before 1.14.0 lets a single remote attacker exhaust server memory and CPU with one HTTP/1.1 request whose chunked body is composed of a very large number of 1-byte chunks. Because the chunked reader keeps every chunk as its own cons cell and binary term until IO.iodata_to_binary/1 joins them, and the :length option of read_body/2 counts only data bytes, the vendor's reproduction shows 48 MB of 1-byte chunks growing one handler process to roughly 500 MB and consuming about 6 seconds of CPU - a ~48x amplification over the payload sent - and the amplification also occurs against routes that never call read_body, since ensure_completed/1 drains unread bodies on keep-alive connections. HTTP/1.1 is enabled by default, so any default deployment is exposed without authentication or user interaction; the vendor shipped a fix in 1.14.0, there are no reports of exploitation in CISA KEV and no independently confirmed public exploit code, though the advisory does publish detailed reproduction steps.
Unauthenticated remote attackers can consume disproportionate server CPU on Bandit (mtrudel/bandit, Elixir/Erlang) by sending a Content-Length value or a Host/:authority port composed of thousands of digits, which the header integer parser processes in quadratic time before any range check. Over HTTP/2 the cost is amplified because a 4,000-digit value can be stored once in the HPACK dynamic table and replayed cheaply across thousands of new streams; the advisory's proof-of-concept sent roughly 124 KB that kept the server busy for 27.9 seconds. Both HTTP/1 and HTTP/2 deployments are affected at default configuration, versions 1.0.0-pre.7 up to but not including 1.14.0. The advisory includes a public PoC reference that is not independently verified as working exploit code; there is no CISA KEV listing or other indication of active exploitation, and the vendor has released 1.14.0 as the fix.
Memory exhaustion in the Erlang/Elixir Bandit web server (mtrudel/bandit 1.11.0 up to but not including 1.14.0) lets an unauthenticated remote client drive a single WebSocket connection to roughly 150 times the configured max_fragmented_message_size, about 1.2-1.4 GB at the 8 MB default. Bandit counts only payload bytes against the fragment limit while storing every non-final 1-byte continuation frame as its own list cell costing about 150-170 bytes of BEAM heap, and ping frames injected between fragments keep the connection alive past the idle timeout. Any deployment that accepts WebSocket upgrades with default settings is exposed, notably Phoenix LiveView and Phoenix Channels applications; the vendor advisory includes reproduction steps (a public PoC reference exists; it is not verified as working exploit code), no public exploit tooling or active exploitation is indicated by the available data, and the vendor has released a fix in 1.14.0.
Unauthenticated remote attackers can hold Bandit HTTP/2 stream processes and the request-body reads they are servicing open indefinitely by sending small WINDOW_UPDATE or empty SETTINGS frames before the read timeout expires, affecting Bandit versions from 1.4.0 before 1.14.0 with HTTP/2 enabled and routes that read request bodies. Impact is availability only: request handlers and associated resources can be occupied across many streams and connections, creating denial-of-service conditions; confidentiality and integrity are not affected. Vendor-released patch: 1.14.0; no public exploit identified at time of analysis, and this is a genuine availability risk that warrants prompt attention, though not an emergency.
Unauthenticated remote memory exhaustion in the Elixir Bandit HTTP/2 server lets a single client consume all memory on the BEAM node by sending DATA frames that the server never refuses. Affected are bandit releases from 0.3.1 up to, but not including, 1.14.0 when HTTP/2 is enabled, which is the default. The attacker stalls a stream by requesting a response larger than the 65,535-byte initial window and withholding WINDOW_UPDATE, then floods small DATA frames that queue unread in the stream process mailbox; a public proof-of-concept procedure is described in the vendor advisory, but it is not confirmed as working exploit code, and no CISA KEV entry or other active-exploitation signal is present in the supplied data.
CreditNirvana's Agentic AI Call Centre ships a hardcoded static API key inside its client-side JavaScript bundle, so anyone who can reach the application can download the bundle, extract the key and use it to call the application API. The result is remote, unauthenticated read-only exposure of call transcripts and recordings (CWE-321); integrity and availability are unaffected. The reporting source is CERT-In and the priority signal places it in the 'this week' handling window; no public exploit code or confirmed active exploitation was identified at time of analysis.
Access control bypass in Syncplify Server! before 8.8.0 allows authenticated SFTP users to evade per-path permission overrides by addressing a protected directory with alternate capitalization or its 8.3 short name, enabling file uploads and downloads the override is meant to deny. The flaw is conditional on an existing deny override and on filesystem behavior that accepts alternate case or short names, such as Windows/NTFS; on strictly case-sensitive filesystems without short-name generation the technique is limited. No public exploit was identified at time of analysis, and the vendor has released 8.8.0 as the fix.
Unauthenticated file disclosure and arbitrary ONNX computation in rembg prior to 2.0.82: the public /api/remove endpoint forwards an attacker-supplied extras.sam_model value into SAM model download logic, where it is interpolated into local filenames and GitHub download URLs passed to pooch.retrieve() without a pinned hash. By combining path traversal with URL-fragment syntax - the HTTP request discards the '#fragment' while the local POSIX path keeps it - an attacker who can reach the HTTP server mode endpoint primes a directory tree with a first request (HTTP 500) and then, with an identical second request, plants malicious ONNX encoder/decoder files beside a target file; those graphs read the target via ONNX external data and encode its bytes into the returned PNG mask. No authentication is required (CVSS 3.1 AV:N/AC:L/PR:N/UI:N, 8.6), impact is bounded by what the rembg process can read, and the vendor advisory includes proof-of-concept scripts validated against a loopback lab deployment; no CISA KEV entry exists and the vendor fix ships in v2.0.82.
Cross-tenant data exposure in OpenPanel before 3.0.0 lets any authenticated user read other projects' profiles, events and sessions out of a shared ClickHouse cluster by injecting a SELECT expression into the 'property' parameter of the profile.values tRPC endpoint. The injected expression carries its own FROM/WHERE clause and comments out the trailing project_id scope, and results are returned directly in the response 'values' field, so exfiltration is non-blind (emails, IPs, session and revenue data from any tenant). Registration and project creation are self-service, so the authentication prerequisite is trivial; a reproducible proof-of-concept with full request steps is published in the vendor's GitHub advisory, but there is no KEV listing and no confirmed active exploitation, and EPSS is low at 0.39% (32nd percentile) with a vuln.today decision of this_week. The vendor CVSS 4.0 base score is 8.6, while our assessment treats it as a scope-changed confidentiality-only issue.