🚨 CVE-2026-96890
A Server-Side Request Forgery (SSRF) vulnerability was identified in GitHub Enterprise Server that allowed a repository contributor to cause the appliance to issue requests to attacker-controlled internal hosts, which could be chained to achieve remote code execution on the appliance. The secret scanning validator for GCP service account credentials trusted the token endpoint embedded in a committed credential and issued a request to it without restricting the destination. Exploitation required an authenticated user with permission to push to a repository on an instance with GitHub Advanced Security and secret scanning validity checks enabled, a non-default configuration. This vulnerability affected GitHub Enterprise Server 3.20, 3.21, and 3.22 and was fixed in versions 3.20.9, 3.21.7, and 3.22.2. This vulnerability was reported through the GitHub Bug Bounty program.

🎖@cveNotify

🚨 CVE-2026-9226
Authentication bypass in the Azure AD external login flow in Devolutions Server 2026.3.7.0 and earlier allows a remote attacker to take over a user's account via replay of a captured login-session token exposed in a redirect URL.

🎖@cveNotify

🚨 CVE-2026-73076
Vim is an open source, command line text editor. Prior to 9.2.0847, runtime/autoload/vimball.vim allows a crafted vimball member named .VimballRecord to overwrite the installation record with attacker-chosen commands. When vimball#RmVimball() later processes the matching record entry, the stored Ex commands, including operating-system commands invoked through :!, execute with the privileges of the user running Vim. This issue is fixed in version 9.2.0847.

🎖@cveNotify

🚨 CVE-2026-73077
Vim is an open source, command line text editor. Prior to 9.2.0839, the runtime/ftplugin/sh.vim, runtime/ftplugin/zsh.vim, and runtime/ftplugin/ps1.vim filetype plugins pass attacker-controlled Visual-mode selections from K through keywordprg commands without safely separating shell arguments. fnameescape() and PATH_ESC_CHARS do not neutralize shell metacharacters before ShKeywordPrg, ZshKeywordPrg, or GetHelp invokes bash, zsh, or PowerShell, allowing arbitrary operating-system commands to execute with the privileges of the user running Vim. This issue is fixed in version 9.2.0839.

🎖@cveNotify

🚨 CVE-2026-73078
Vim is an open source, command line text editor. Prior to 9.2.0840, runtime/plugin/netrwPlugin.vim loads netrw and runtime/pack/dist/opt/netrw/autoload/netrw.vim constructs Bookmarks, History, and Targets menu entries by interpolating attacker-controlled directory paths into executed :menu commands. s:NetrwBookmarkMenu(), s:NetrwTgtMenu(), g:netrw_menu_escape, EX_TRLBAR, and netrw#MakeTgt() fail to neutralize the | command separator or single quotes at five construction sites, allowing a crafted path browsed or bookmarked in GUI Vim to execute arbitrary Ex and operating-system commands. This issue is fixed in version 9.2.0840.

🎖@cveNotify

🚨 CVE-2026-100647
vLLM versions before 0.29.0 contain a denial-of-service vulnerability in the cache_salt parameter accepted on OpenAI-compatible and Anthropic API endpoints, which lacks maximum length validation and is processed on the single EngineCore scheduler thread. Unauthenticated attackers can send HTTP requests with multi-hundred-megabyte salt values that trigger expensive pickle serialization and SHA-256 hashing, stalling the scheduler thread and denying service to all concurrent requests.

🎖@cveNotify

🚨 CVE-2026-100648
vllm before 0.29.0 fails to enforce VLLM_MAX_AUDIO_CLIP_FILESIZE_MB limit in multimodal chat audio decoding, allowing unauthenticated clients to bypass file size restrictions. Attackers can submit oversized audio files through chat endpoints to consume excessive memory and CPU resources during decoding.

🎖@cveNotify

🚨 CVE-2026-100649
vLLM before 0.29.0 contains a resource-limit bypass vulnerability in PyNvVideoCodec decoder allocation where sampler subclass shadowing allows independent counter increments. Unauthenticated attackers can select different sampler subclasses in video requests to exceed configured decoder limits and exhaust unaccounted GPU memory.

🎖@cveNotify

🚨 CVE-2026-100650
vLLM through 0.29.0 fetches and fully materializes remote or inline media before enforcing its documented media controls (the VLLM_MAX_AUDIO_CLIP_FILESIZE_MB compressed-audio size cap, default 25 MB, and the per-modality --limit-mm-per-prompt item limits). Across four ingress paths — the shared media-acquisition layer (HTTPConnection.get_bytes()/async_get_bytes()), the chat completions audio_url/base64 path, the batch speech runner, and the Rust frontend POST /tokenize route — the server reads the entire HTTP response body, base64-decodes the inline payload, or spawns one fetch/decode task per media part, and only then applies the limit (or, on some paths, never applies it). A remote attacker can therefore cause the API server or batch-runner process to allocate memory and consume outbound bandwidth proportional to an attacker-chosen body size or media item count before the request is rejected, resulting in pre-inference memory and bandwidth exhaustion (denial of service). The chat and batch surfaces require an API key when one is configured; the Rust frontend /tokenize route is unauthenticated by design. There is no code execution or data disclosure impact.

🎖@cveNotify

🚨 CVE-2026-100651
vLLM before 0.29.0 fails to enforce decoder prompt-length validation on the disaggregated serving endpoint /inference/v1/generate. When the request contains a 'features' (multimodal) payload, vllm/entrypoints/serve/disagg/serving.py builds a multimodal EngineInput directly from the caller-supplied token_ids, and GenerateRequest.token_ids (vllm/entrypoints/serve/disagg/protocol.py) is not checked against model_config.max_model_len. For multimodal processors that report skip_prompt_length_check=True (for example Nemotron Parse, Whisper, and FireRedLID), InputProcessor._validate_prompt_len() returns immediately for both encoder and decoder prompts, so an overlong prompt becomes an EngineCoreRequest and reaches the worker input-batch copy into a fixed max_model_len-wide NumPy row. A client able to reach the endpoint on an affected model configuration can therefore submit an overlong token_ids list to trigger a worker failure and denial of service. Fixed in 0.29.0.

🎖@cveNotify

🚨 CVE-2026-100652
vLLM versions 0.22.0 through 0.23.0 fail to validate stop_token_ids against vocabulary bounds in Rust HTTP and gRPC frontends, allowing out-of-vocabulary token IDs to reach MinTokensLogitsProcessor. Attackers can submit requests with min_tokens greater than zero and out-of-vocabulary stop_token_ids to trigger CUDA tensor indexing failures that leave EngineCore in a fatal state requiring service restart.

🎖@cveNotify

🚨 CVE-2026-73071
Vim is an open source, command line text editor. From 9.2.0511 until 9.2.0844, json_decode_item() in src/json.c can retain a stale pointer after json_decode_string() invokes channel_fill() to refill and free the current buffer, causing the error path to read freed memory instead of reader->js_buf + reader->js_used when an invalid JSON string spans buffers. This issue is fixed in version 9.2.0844.

🎖@cveNotify

🚨 CVE-2026-73072
Vim is an open source, command line text editor. Prior to 9.2.0846, set_sofo() in src/spellfile.c reuses sl_sal_first[] without resetting values left by set_sal_first(), so a crafted spell file containing an SN_SAL section before an SN_SOFO section causes under-counted mapping lists and attacker-influenced writes beyond a heap allocation. This issue is fixed in version 9.2.0846.

🎖@cveNotify

🚨 CVE-2026-73074
Vim is an open source, command line text editor. Prior to 9.2.0841, prop_add_one() in src/textprop.c uses the proplen value from get_text_props() to increment a uint16_t property count beyond 0xffff, wrapping the count to zero and copying existing text-property records into a heap allocation sized for none of them. This issue is fixed in version 9.2.0841.

🎖@cveNotify

🚨 CVE-2026-73075
Vim is an open source, command line text editor. From 9.2.0469 until 9.2.0843, popup_mark_opacity_zindex() in src/popupwin.c can use a negative w_winrow for a text-property-anchored popup with clipwindow and opacity, indexing before the screen array instead of accounting for w_popup_topoff and causing an out-of-bounds read and conditional write. This issue is fixed in version 9.2.0843.

🎖@cveNotify

🚨 CVE-2026-67421
RabbitMQ is a messaging and streaming broker. From 3.13.0 until 3.13.19, 4.0.24, 4.1.15, 4.2.10, and 4.3.5, RabbitMQ Management rendered an AMQP authorization-error reason containing an attacker-controlled queue name as HTML when the OAuth management UI was enabled. Exploitation requires an attacker with queue configure permission, a management administrator who can see but cannot read that queue, and the administrator clicking Get Message(s). A queue name containing a base element can then retarget the automatic relative refresh because the Content Security Policy omits base-uri and connect-src, and an attacker endpoint that permits the management origin through CORS can receive the victim's Authorization header. This issue is fixed in versions 3.13.19, 4.0.24, 4.1.15, 4.2.10, and 4.3.5.

🎖@cveNotify

🚨 CVE-2026-90441
A missing authorization vulnerability in the wgagent management daemon's session initialization function allows an authenticated, low-privileged user (including a read-only or guest administrator account) to crash the wgagent process and read arbitrary files accessible to the daemon by submitting a specially crafted management API request.

🎖@cveNotify

🚨 CVE-2026-103099
Pexip Infinity before 41.1 is affected by improper input validation in the media implementation that allows a remote attacker to trigger a software abort resulting in a denial of service.

🎖@cveNotify

🚨 CVE-2026-63576
Improper certificate validation in PkixNameConstraintValidator (ExtractHostFromURL) in Legion of the Bouncy Castle Inc. bc-csharp before 2.7.0 allows a name-constrained subordinate CA, or anyone able to obtain certificates with chosen subjectAltName URIs from such a CA, to bypass permitted or excluded uniformResourceIdentifier name constraints during certification path validation via a URI whose path, query, fragment or userinfo contains characters such as '@' or ':', because the host was extracted by string slicing without first isolating the RFC 3986 authority component, so the host compared against the constraints could differ from the URI's actual host.

🎖@cveNotify

🚨 CVE-2026-63577
Improper certificate validation in the directoryName name-constraint check (PkixNameConstraintValidator.WithinDNSubtree) in Legion of the Bouncy Castle Inc. bc-csharp before 2.7.0 allows an attacker who controls, or can have certificates issued by, a name-constrained intermediate CA to get certificates accepted by PKIX path validation whose subject distinguished name, or a directoryName subjectAltName, lies outside the CA's permitted subtrees, via a name that places other RDNs ahead of a copy of the permitted RDN sequence, because the check looks for the constraint's first RDN anywhere in the name and compares the remaining RDNs from that position, instead of requiring the constraint to be an initial prefix of the name as RFC 5280 sections 4.2.1.10 and 7.1 require.

🎖@cveNotify

🚨 CVE-2026-63578
Allocation of resources without limits in password-based private-key decryption (PbeUtilities.GenerateCipherParameters) in Legion of the Bouncy Castle Inc. bc-csharp before 2.7.0 allows an attacker who can supply an encrypted private key, such as a PKCS#8 EncryptedPrivateKeyInfo or "ENCRYPTED PRIVATE KEY" PEM file, to cause a denial of service through CPU exhaustion via an iteration count close to 2^31, because the count is taken from the unauthenticated algorithm parameters without an upper bound and the key derivation runs before the password or the data can be checked. PKCS#5 PBES1 and PBES2 (PBKDF2), the PKCS#12 PBE algorithms and CMS password recipients (CmsPbeKey) are affected. Loading PKCS#12 files with Pkcs12Store is covered by CVE-2026-63572, and a zero or negative count with the PKCS#12 algorithms by CVE-2026-63575.

🎖@cveNotify