Threat database

Flarum Extension Manager 2.0.0-beta.1 before 2.0.0-beta.8 lacks authentication and admin checks on the TaskResource index endpoint, allowing unauthenticated information disclosure. Remote attackers can send GET requests to /api/extension-manager-tasks to read task output revealing filesystem paths, installed extensions, versions, and PHP version.

Intelligence Community UI Kit @ukic/canary-web-components 3.0.0-canary.12 before 3.0.0-canary.48 contains a DOM-based cross-site scripting vulnerability in ic-tree-item that re-parses router-item slot text as HTML. When tree item truncation is enabled, attackers who influence a tree item label can have escaped markup decoded by truncateTreeItemLabel() and executed as JavaScript in visitors' browsers.

Koalixcrm through 1.14.0 contains a stored cross-site scripting vulnerability that allows staff users to inject script by storing malicious task titles rendered unescaped by Task.link_to_task(). Attackers can embed HTML or script that executes when administrators open the task changelist or change form, enabling actions like creating a superuser.

Krayin CRM 2.2.5 before 2.2.7 contains a server-side request forgery filter bypass that allows users with webhook permissions to reach internal addresses via DNS rebinding. Attackers can return a public address during WebhookService::isSafeEndpoint() validation and an internal address when Guzzle re-resolves, sending blind requests to internal services and cloud metadata endpoints.

Krayin CRM before 2.2.7 contains a broken object level authorization vulnerability in EmailController write actions that allows scope-restricted authenticated users to manipulate other users' emails by supplying their numeric IDs. Attackers can call update, destroy, mass-update or mass-destroy endpoints to edit, re-send via is_draft=0, trash or permanently delete emails and disclose their subject, body and linked lead and person data.

Krayin CRM before 2.2.7 contains an insecure direct object reference vulnerability that allows scope-restricted authenticated users to read other users' quotes by bypassing per-user data scope checks in QuoteController. Attackers can query the quotes search and lead-products endpoints to retrieve quote addresses, totals, contact details, and product pricing belonging to any user or lead.

libaribcaption 1.1.0 through 1.1.2 contains a global out-of-bounds read in DecoderImpl::HandleGLGR() that reads past the 78-entry kLatinSpecialTable without an upper bound check. Attackers can supply crafted captions designating the Latin special set with ESC 28 4C followed by bytes 0x6F-0x7E to leak adjacent read-only data or crash decoding.

VideoLAN libbluray through 1.5.1 contains an uncontrolled memory allocation vulnerability in _parse_extent_start_points() in clpi_parse.c that trusts the num_point field without bounds checking. Attackers can supply a crafted .clpi file on a disc, image, or directory with a huge num_point to exhaust memory and CPU, causing denial of service.

VideoLAN libbluray through 1.5.1 contains an out-of-bounds read vulnerability in clpi_access_point() that allows attackers to crash applications by supplying a ClipInfo file with an empty EP-map fine table. Attackers can place a crafted .clpi file in BDMV/CLIPINF with num_ep_fine set to zero and an unclamped ref_ep_fine_id, triggering reads during seeking for denial of service.

A stack-based buffer overflow vulnerability exists in libaribb24 (aribb24) through version 1.0.3 in the decoder_handle_csi() function. This flaw allows attackers to write beyond a fixed 256-entry array by supplying unbounded CSI parameter bytes embedded in media files or broadcast streams. Exploitation can cause crashes in applications like FFmpeg or VLC or corrupt the return address, potentially leading to denial of service or other impacts.

libaribb24 (aribb24) through 1.0.3 contains a heap-based buffer overflow vulnerability in src/parser.c that allows attackers to corrupt heap memory via an oversized caption data unit. Attackers can supply a crafted caption PES packet in a media file or broadcast stream, processed through arib_parse_pes(), to crash FFmpeg or VLC and possibly corrupt further memory.

Libdisplay-info 0.4.0 contains a heap out-of-bounds read vulnerability in the DisplayID v2 section parser in displayid2.c that allows attackers to read past the EDID buffer via oversized data block lengths. Attackers can supply a crafted EDID from a malicious monitor, KVM, capture device, or file to over-advance the parse cursor and crash applications.

libdisplay-info through 0.4.0 contains a heap out-of-bounds read vulnerability in the DisplayID v1 section parser in displayid.c that occurs when a data block's declared length exceeds the section. Attackers can supply a crafted EDID from a malicious monitor, KVM, capture device, or EDID file to make is_all_zeroes read past the buffer, causing a possible crash.

SGLang through 0.5.22 with the Mooncake disaggregation backend contains an uncontrolled memory allocation vulnerability in handle_staging_rsp() that lets unauthenticated network-adjacent attackers exhaust Prefill memory. Attackers can obtain the rank address from the /route endpoint and send STAGING_RSP messages with huge chunk_idx values under fresh room IDs, causing the scheduler to be OOM-killed.

SGLang through 0.5.22 with the Mooncake disaggregation backend accepts unauthenticated decode registration messages on the Prefill rank ZMQ PULL socket, allowing denial of service. Network-adjacent attackers can send many registrations with unique mooncake_session_id values and large nested-buffer frames to exhaust Prefill memory until the scheduler is OOM-killed.

sglang-router (sgl-model-gateway) through 0.3.2 contains an authorization bypass vulnerability that allows holders of the data-plane API key to access mesh HA management routes because /ha/* endpoints skip control-plane Admin role checks. Attackers can read cluster topology, write gossiped configuration, change global rate limits, and call POST /ha/shutdown to silently dissolve HA coordination.

SGLang through 0.5.22 in prefill/decode disaggregation mode contains a server-side request forgery vulnerability that allows unauthenticated attackers to trigger outbound requests via client-supplied bootstrap_host and bootstrap_port fields. Attackers sending crafted requests to the Decode worker's /generate endpoint can probe internal hosts and ports and block the shared scheduler loop, delaying other inference requests.

SGLang through 0.5.22 contains a denial of service vulnerability that allows API clients to abort other users' in-flight generation requests by supplying a short client-controlled rid. Attackers can submit streaming /generate requests with single hex character rids and disconnect, triggering prefix-matching abort logic that terminates all other clients' requests.

LangBot through 4.11.0 contains a missing authenticity verification vulnerability that allows unauthenticated attackers to inject forged Lark/Feishu events by skipping signature checks. Attackers knowing a bot UUID can POST fabricated im.message.receive_v1 or card.action.trigger events to drive the LLM pipeline, tools, and approvals as other users.

Television (tv) 0.15.0 through 0.15.9 contains an OS command injection vulnerability that allows attackers to run shell commands through improperly escaped single quotes in format_command. Attackers can plant a file with a crafted name containing an apostrophe so that triggering an action like the default edit action executes commands with victim privileges.