| CVE |
Vendors |
Products |
Updated |
CVSS v3.1 |
| vLLM from 0.8.5 before 0.31.0 has a denial of service flaw. LogprobsTensors.empty_cpu() in vllm/v1/outputs.py allocates an uninitialized int32 tensor, and the V1 model runner leaves prompt positions reported as computed by a KV connector unwritten. A client sharing the engine can send prompt_logprobs (or echo with logprobs) against a connector-cached prefix, making tokenizer decode raise OverflowError, which crashes EngineCore and drops all co-tenant requests. |
| vLLM from 0.18.0 before 0.31.0 contains a resource exhaustion vulnerability that allows API clients to bypass the video/jpeg frame-count cap by overriding num_frames through media_io_kwargs. Attackers can set num_frames to -1 with a large inline video so VideoMediaIO.load_base64 decodes every frame, exhausting memory and crashing the API frontend. |
| vLLM from 0.20.0 before 0.30.0 contains a denial of service vulnerability that allows remote attackers to crash the engine by sending inconsistent multimodal metadata to the disaggregated /inference/v1/generate endpoint. Attackers can replay a valid multimodal item with repeated or overlapping mm_placeholders ranges or mismatched kwargs_data and mm_hashes lists, producing a negative text span in Qwen2.5-VL position construction that terminates the engine. |
| vLLM from 0.10.2 before 0.25.0 contains an uncontrolled resource consumption vulnerability that allows unauthenticated attackers to exhaust memory because the Nemotron-VL processors globally set Image.MAX_IMAGE_PIXELS to None. Attackers can submit a pathological decompression-bomb image to /v1/chat/completions to OOM-kill the worker and deny service to co-tenants. |
| vLLM from 0.9.2 before 0.24.0 contains a server-side request forgery vulnerability that allows API callers to force outbound connections by embedding a decode address in the X-Request-Id header. Attackers can make a P2pNcclConnector prefill worker ZMQ-connect to an attacker host, leaking the NCCL unique_id and KV-cache tensors or blocking the worker. |
| vLLM from 0.19.0 before 0.24.0 contains a decompression bomb vulnerability because the AudioMediaIO multimodal path calls load_audio without the max_duration_s decoded-duration limit. Authenticated attackers can send small compressed audio via input_audio or audio_url in /v1/chat/completions requests that expands to huge decoded PCM, exhausting server memory. |
| vLLM from 0.22.0 before 0.24.0 contains an improper input validation vulnerability in the Rust gRPC Generate and GenerateStream service that forwards unchecked prompt token IDs to EngineCore. Attackers can submit requests with out-of-vocabulary prompt.token_ids values to trigger a fatal embedding lookup failure that terminates EngineCore and leaves the server unhealthy. |
| vLLM from 0.22.0 before 0.24.0 contains a denial of service vulnerability that allows remote attackers to terminate EngineCore by bypassing logprob_token_ids limits in the Rust gRPC frontend. Attackers can send Generate or GenerateStream requests with output_logprobs enabled and more than 128 output_candidates.token_ids, triggering a fatal sampler exception that crashes the engine. |
| vLLM from 0.22.0 before 0.24.0 contains an improper input validation vulnerability in the Rust gRPC converter that allows remote attackers to bypass max_logprobs limits via CandidateTokens.top_n. Attackers can send Generate or GenerateStream requests with oversized or integer-wrapping top_n values to trigger torch.topk failures that terminate EngineCore. |
| SiYuan 3.5.4 before 3.8.7 on Windows contains a sensitive-path denylist bypass in IsSensitivePath because the Windows extended-length prefix (\\?\) survives filepath.Clean and matches no forbidden prefix. Authenticated API callers can prepend the prefix to localPath in /api/import/importStdMd to copy workspace conf files, including the API token and TLS private keys, into assets served to unauthenticated clients. |
| SiYuan before v3.8.7 contains an sql injection vulnerability that allows admin-token holders to read history.db by injecting unescaped op and type=3 query values. Attackers can send crafted UNION SELECT payloads to searchHistory or getHistoryItems endpoints to exfiltrate deleted document content, titles, and paths. |
| SiYuan from 2.10.0 before 3.8.7 contains a stored cross-site scripting vulnerability that allows users with edit permission to inject script by placing javascript: URLs in math block content rendered with KaTeX trust enabled. Attackers can store \href, \url, or \includegraphics payloads that mathRender assigns via innerHTML, executing script in the application origin when a reader clicks the rendered link. |
| SiYuan before 3.8.7 contains a path traversal vulnerability in the MCP asset tool because validateAssetUploadPaths relies on an incomplete sensitive-path blacklist that omits .env and other secret files. Authenticated attackers driving the MCP asset upload action can copy such local files into the workspace and read them back via /assets/ over HTTP. |
| SiYuan from 2.9.0 before 3.8.7 contains a cross-site scripting vulnerability in the unauthenticated /check-auth page that assigns the to parameter to window.location.href without scheme validation. Attackers can send a crafted link with a javascript: URI that executes after victim login, invoking kernel APIs to read or modify workspace content and settings. |
| SiYuan 3.6.0 before 3.8.7 accepts and persists the networkServeTLS setting but never enforces it, so the kernel port keeps serving full API access over plain HTTP alongside HTTPS. Network-adjacent attackers can sniff the cleartext login flow and session cookie, which lacks the Secure attribute while responses omit HSTS, to hijack kernel API sessions. |
| Parse Server before 8.6.101 and 9.0.0 before 9.10.4-alpha.4 contains a missing authorization vulnerability on MongoDB deployments that lets clients without the master key delete the stored schema of any class. Unauthenticated attackers can send REST API delete requests targeting the internal _SCHEMA collection to remove field definitions, class-level permissions, and protected fields, exposing data to unauthorized reads and modifications. |
| Parse Server before 8.6.100 and 9.0.0 before 9.10.4-alpha.3 does not reject reserved internal field names in client object writes, allowing clients to store or clear internal fields. Attackers can write a reserved field that makes reads of an object fail on MongoDB, disrupting queries on that class for all users, or clear their own account lockout state and password history to bypass lockout and password policies. |
| Parse Server before 8.6.102 and 9.0.0 before 9.10.5-alpha.2 contains an authentication bypass because the validateSetUp, validateLogin, and validateUpdate methods of custom adapters subclassing AuthAdapter are never called, and an inherited default that accepts any authData is used instead. Attackers can supply arbitrary authData to sign up, link identities, log in as existing users, or skip custom additional authentication factors. |
| Parse Server before 8.6.104 and 9.0.0 before 9.10.5-alpha.5 saves the entire login request with master privileges when issuing an SMS multi-factor authentication verification code. An SMS MFA-enrolled user can include extra fields in that request to modify master-key-restricted fields on their own account, such as the email-verification flag or beforeSave-protected values. |
| Parse Server before 8.6.103 and 9.0.0 before 9.10.5-alpha.4 contains an information disclosure vulnerability in LiveQuery. Its subscription hash omits constraints next to a $or operator, including the _Session owner filter and beforeSubscribe constraints, so distinct subscriptions are grouped together. Authenticated clients subscribing with $or queries can receive other users' events, including session tokens usable for impersonation. |
