38,614 results · Feed
-
Unscored —
Finished streaming calls keep reading a stalled request body indefinitely
Affected versions of `connectrpc` start a background task to read the request body of every client-streaming and bidirectional-streaming call, before the interceptors or the handler run. A call can finish before it has consumed its request: the handler can re…
-
Unscored —
`unzip` is unmaintained
The [unzip](https://crates.io/crates/unzip) crate has only ever released a single version, `0.1.0` (published 2017-12-23), and has seen no further activity. It appears to be unmaintained, and its known path-traversal ("zip-slip") vulnerability will not be fix…
-
Unscored —
Double free / use-after-free in `Consumer::skip` and `Consumer::clear` when an element's `Drop` panics
`Consumer::skip()` and `Consumer::clear()` are not panic-safe. They drop the consumed elements in place and only afterwards call `advance_read_index()` to move the ring buffer's read index past them. If an element's `Drop` panics mid-loop, `advance_read_index…
-
Unscored —
`SinglyLinkedList::remove` dereferences a null link
In versions before 0.2.1, `SinglyLinkedList::remove` is safe and walks the intrusive list with an unchecked dereference. On an empty list, or when `node` is not in the list, `(*current_elm).next` reads a null pointer. That is undefined behavior. The list head…
-
Unscored —
`hex_decode_unchecked` AVX2 path reads past `src`
`faster_hex::hex_decode_unchecked` is a safe function. On x86 and x86_64, when AVX2 is enabled, `hex_decode_avx2` keeps decoding while `dst.len() >= 32` and then loads 32 bytes from `src` and 32 bytes from `src[32..]`. Nothing in that loop requires `src` to c…
-
Low 3.1
vLLM through 0.29.0 Cross-Request Logits Corruption via bad_words
vLLM through 0.29.0 fails to properly validate bad_words token indices against the model's generation output width in SamplingParams.update_from_tokenizer(). Attackers can supply out-of-bounds token indices that corrupt logits memory of concurrent requests, c…
-
Medium 6.3
Unvalidated chunk-size line tail in Mint HTTP/1 client enables response smuggling against strict intermediaries on pooled connections
Inconsistent Interpretation of HTTP Requests ('HTTP Request/Response Smuggling') vulnerability in elixir-mint mint allows a malicious HTTP/1 server to desynchronize a strict intermediary and the Mint client on a pooled connection, enabling response-queue pois…
-
High 7.1
`decompress`: tar-family extractors write archive entries without a path-traversal check (tar-slip)
The tar-family extractors in `decompress` (`.tar`, `.tar.gz`, `.tar.xz`, `.tar.bz2`, `.tar.zst`) build each output path from the **raw archive entry path** and write to it with no traversal check, so a malicious archive can write files **outside** the destina…
-
Low 3.7
vLLM through 0.29.0 Adjacent-Request Sampler State Corruption via Unvalidated Prompt Token IDs
vLLM through 0.29.0 contains a memory corruption vulnerability in the Triton _bincount_kernel where prompt token IDs index the penalty prompt-presence bitset without bounds checking against vocabulary size. Attackers can submit multimodal audio requests with …
-
Low 3.7
vLLM before 0.29.0 Cross-Request Logits Corruption via allowed_token_ids
vLLM before 0.29.0 validates allowed_token_ids against tokenizer length instead of model output logits width in SamplingParams._validate_allowed_token_ids(). Attackers can supply token IDs above the output vocabulary that pass validation, causing LogitBiasSta…
-
High 7.5
source-map-js through 1.2.1 Event Loop Denial of Service
source-map-js through 1.2.1 fails to validate the per-section offset line value in indexed source maps, allowing attackers to specify arbitrary numeric values. Attackers can supply extremely large offset line values that cause synchronous event loop blocking …
-
High 7.6
Obot before v0.23.0 Server-Side Request Forgery via MCP
Obot before v0.23.0 contains a server-side request forgery vulnerability in remote MCP server registration that allows privileged users to specify arbitrary URLs without destination validation. Attackers with Power User or higher roles can coerce Obot to make…
-
Medium 5.3
Obot before v0.23.0 Authentication Bypass via Registry API
Obot versions before v0.23.0 fail to enforce authentication on MCP Registry endpoints under /v0.1/* when registry authentication is enabled. Unauthenticated attackers can read registry metadata including server names, descriptions, repository URLs, and connec…
-
High 8.8
Obot before v0.23.0 Authentication Bypass via OAuth Dynamic Client Registration
Obot before v0.23.0 (affected versions <= v0.22.1) running with OBOT_SERVER_ENABLE_AUTHENTICATION=true exposes OAuth dynamic client registration without authentication and without any restriction on the redirect URIs a client may register. Because the authori…
-
Medium 6.5
Paymenter: Credit-refund double-spend race condition in service downgrade (doUpgrade)
Paymenter is a free and open-source webshop solution for management of hosting services. Prior to 1.5.7, app/Livewire/Services/Upgrade.php::doUpgrade() relies on Service::upgradable to check for a pending service upgrade and later executes $credit->increment(…
-
High 7.5
Moquette: Missing Authorization in io.moquette:moquette-broker
Moquette is a lightweight Java MQTT broker. Prior to 0.18.1, PostOffice.publishWill publishes a client-controlled Last Will message through publish2Subscribers without invoking the authorizator.canWrite check used by normal PUBLISH paths. When anonymous acces…
-
Critical 9.1
Mnemosyne has JWT signature verification bypass sync server that allows authentication bypass
Mnemosyne is a memory layer for artificial intelligence agents. Prior to v3.10.1, the auth check in mnemosyne/core/sync_server.py parsed the JWT's header and payload using base64 decoding, then passed the token to a jwt library call with options that effectiv…
-
High 7.1
Perses: Unvalidated project parameter enables filesystem path traversal
Perses is an open-source dashboard and visualization project for observability data. Prior to 0.54.0-rc.0, list endpoints used with the file-system database bind the request-controlled project query parameter into the resource Query structure without validati…
-
High 8.3
Perses: Missing authorization in datasource proxy allows cross-scope secret disclosure
Perses is an open-source dashboard and visualization project for observability data. From 0.43.0 until 0.54.0-rc.0, the datasource creation and unsaved datasource proxy paths authorize the caller on a Datasource or GlobalDatasource scope but do not require re…
-
High 7.1
Perses project query parameter authorization bypass exposes cross-project resources
Perses is an open-source dashboard and visualization project for observability data. Prior to 0.54.0-beta.3, an authenticated user with viewer access to one project can supply another project through the project query parameter on project-scoped list endpoint…
-
High 8.2
File Viewer: DOM XSS via unsafe hyperlink schemes in the legacy DOC renderer
File Viewer is a browser-native viewer for Office, PDF, CAD, archive, and other files in private and internal web applications. Prior to @file-viewer/doc 2.3.1 and msdoc-viewer 0.2.2, the legacy DOC renderer emitted document-controlled hyperlink targets into …
-
High 7.5
adm-zip: Uncontrolled memory allocation via the declared uncompressed size (DoS)
adm-zip is a JavaScript library for creating and extracting ZIP archives in Node.js. Prior to 0.6.1, getData() in zipEntry.js trusts an entry's central-directory uncompressed size and allocates output memory before validating that value against the actual com…
-
Medium 5.1
Process Compose: Browser DNS rebinding lets websites control local process-compose MCP tools
Process Compose is a scheduler and orchestrator for non-containerized applications. Prior to 1.120.0, the MCP SSE listener in src/mcp/server.go accepts browser-origin requests to /sse and the returned message endpoint without validating the Host header, valid…
-
High 7.1
Convoy: Cross-Tenant Source IDOR Leaks Plaintext Message Broker Credentials
Convoy is a cloud native webhooks gateway. Prior to 26.6.8, Convoy's GET /api/v1/projects/{projectID}/sources/{sourceID} endpoint authorizes access to the project in the URL, but Handler.GetSource calls sources.Service.FindSourceByID() and fetches the Source …
-
Medium 6.1
md-editor-v3: XSS via fenced-code language rendering bypass
md-editor-v3 is a Markdown editor for Vue 3 developed in JSX and TypeScript. Prior to 6.5.4, MdPreview's useMarkdownIt() highlight callback in packages/MdEditor/layouts/Content/composition/useMarkdownIt.ts inserts a fenced-code language value into class and l…
-
Medium 5.9
AnyCable: Telemetry Subsystem Contains Hardcoded Authentication Token and Transmits CLI Arguments Including Secrets
AnyCable is a realtime server for reliable two-way communication that supports any backend. Prior to 1.6.15, the telemetry subsystem in telemetry/config.go enables tracking with a hardcoded public authToken, while clusterFingerprint in telemetry/telemetry.go …
-
High 7.0
AnyIO run_process/open_process ignores extra_groups and can retain parent supplementary groups
AnyIO is a high level asynchronous concurrency and networking framework that works on top of either Trio or asyncio. In 4.14.0, AnyIO accepts the POSIX extra_groups argument in anyio.run_process() and anyio.open_process(), but open_process() forwards the grou…
-
Critical 9.3
AnyIO: TLSStream IDNA 2003 host name encoding enables potential TLS certificate spoofing
AnyIO is a high level asynchronous concurrency and networking framework that works on top of either Trio or asyncio. Prior to 4.14.2, connect_tcp() and TLSStream.wrap() can validate internationalized host names after the standard library converts them with ID…
-
Medium 6.8
AnyIO process-pool workers can block indefinitely on undrained stderr
AnyIO is a high level asynchronous concurrency and networking framework that works on top of either Trio or asyncio. Prior to 4.14.2, AnyIO starts process-pool workers with standard error connected to a pipe that the parent never drains, even though the docum…
-
Medium 5.9
AnyCable: Pusher REST API Does Not Verify Request Body MD5 Enabling Signed-Request Replay with Arbitrary Body
AnyCable is a realtime server for reliable two-way communication that supports any backend. Prior to 1.6.15, the Pusher-compatible REST API in pusher/http.go includes the caller-supplied body_md5 value in the HMAC input but does not calculate the digest of th…
-
High 8.8
ToolHive: containerized MCP servers can reach host services via host.docker.internal, enabling lateral movement
ToolHive is a utility designed to simplify the deployment and management of Model Context Protocol servers. Prior to ToolHive CLI 0.30.1 and ToolHive Studio 0.38.0, locally run MCP server containers use the default network permission profile without network i…
-
Critical 9.9
kcp front-proxy does not strip inbound X-Remote-* identity headers, allowing any authenticated client to inject groups/warrants and impersonate system:masters in any workspace
kcp is a Kubernetes-like control plane for form-factors and use-cases beyond Kubernetes and container workloads. Prior to 0.31.4 and 0.32.2, the kcp front-proxy does not remove inbound X-Remote-User, X-Remote-Group, or X-Remote-Extra-* identity headers before…
-
High 8.1
zot: Bearer authentication maps DELETE to push scope, allowing unauthorized deletion
zot is a container image and artifact registry based on the Open Container Initiative Distribution Specification. Prior to 2.1.18, the bearer authentication handler in pkg/api/authn.go maps every HTTP method other than GET and HEAD to the push action, so DELE…
-
Medium 6.8
Capsule: hostnameRegexHandler.OnUpdate validates stale (old) Tenant regex, allowing invalid AllowedHostnames regex to bypass webhook validation
Capsule is a multi-tenancy and policy-based framework for Kubernetes. From 0.13.0 until 0.13.7, hostnameRegexHandler.OnUpdate in internal/webhook/tenant/validation/hostname_regex.go reverses the new and old Tenant parameters and validates the previous Allowed…
-
High 7.1
Capsule: Tenant owner bypasses Capsule's forbidden namespace/service/node label and annotation enforcement
Capsule is a multi-tenancy and policy-based framework for Kubernetes. Prior to 0.13.7, ForbiddenListSpec.ExactMatch in pkg/api/forbidden_list.go sorts denied metadata keys case-insensitively and then uses sort.SearchStrings, which assumes byte-order sorting. …
-
Medium 6.8
Capsule: Malformed ForbiddenAnnotations.Regex can bypass Tenant validation and trigger namespace admission panic
Capsule is a multi-tenancy and policy-based framework for Kubernetes. From 0.13.0 until 0.13.7, the Tenant update validation in internal/webhook/tenant/validation/forbidden_annotations_regex.go compiles ForbiddenLabels.Regex for both the labels and annotation…
-
High 7.5
LMDeploy has an SSRF bypass
### Summary The URL checking logic in lmdeploy has a logical flaw that could be bypassed by attackers, leading to SSRF attacks. ### Details The current lmdeploy project uses `_is_safe_url` to validate the input URL. The main logic is to perform security check…
-
High 8.8
LMDeploy vulnerable to arbitrary code execution via eval() of untrusted quant_dtype in model config loading
LMDeploy is a toolkit for compressing, deploying, and serving large language models. Versions 012.1 through 0.12.2 contain a code injection vulnerability in `lmdeploy/pytorch/config.py` line 620 that allows an attacker to execute arbitrary Python code by publ…
-
Critical 9.8
LMDeploy has Remote Code Execution by Pickle Deserialization via handle_zmq_recv in lmdeploy/lmdeploy/pytorch/disagg/conn/engine_conn.py
LMDeploy is a toolkit for compressing, deploying, and serving large language models. Starting in version 0.9.2 and prior to version 0.16.0, LMDeploy's PyTorch DistServe/PD-disaggregation control plane used `recv_pyobj()` to deserialize messages received throu…
-
High 7.3
Semantic MediaWiki 3.0.0 before 7.3.0 Missing Authorization via smwtask API
Semantic MediaWiki from 3.0.0 before 7.3.0 contains a missing authorization vulnerability in the api.php smwtask module that allows unauthenticated attackers to run admin-only maintenance tasks. Attackers can use the public anonymous CSRF token to read intern…