CVE Watch401,677 in full archive

Vulnerabilities exploitable today

16,781in current view

Single score combining CVSS, KEV membership and EPSS. Every CVE with its own record — timeline from publication to active exploitation.

In KEV catalog1,754

New KEV · 24H0

Exploit Today ≥ 701,666

Distribution · last window

Critical
3,464
High
15,842
Medium
14,999
Low
1,476

Filters

Window

Severity

Flags

Vulnerabilities12,161–12,200 · 16,781

CVECVSSEPSSKEVRExploitTitleMod.

CVE-2026-104873—

15.8%

——5LangGraph Python SDK is used to connect to running LangGraph API servers, manage assistants, threads and stream runs from Python applications. From 0.1.45 until 0.4.4, the langgraph-sdk resource-scoped authorization decorators @auth.on.threads, @auth.on.assistants, and @auth.on.crons ignore the actions argument and register the selected handler for every action on the resource. Because that wildcard resource handler is selected before broader fallback handlers, an authenticated user may bypass fallback action, ownership, or permission checks and read, update, or delete another user's resource. Only Python deployments using actions on the affected decorators are vulnerable, and a deployment remains protected when the selected handler independently enforces all required checks for every action it receives. This issue is fixed in version 0.4.4.16d CVE-2026-1048725.8 MED

14.7%

35.1%

——11The Angular SSR is a server-rise rendering tool for Angular applications. Prior to versions 20.3.36, 21.2.23, and 22.1.7, the CommonEngine retrieveSSGPage prerendered-page retrieval logic in @angular/ssr/node, and in @angular/ssr for versions 17 through 18, accepts a relative request URL containing a backslash parent-traversal segment on Windows. The non-special resolve:// URL base preserves the backslash, path.join interprets it as a Windows separator, and the pagePath.startsWith(normalize(publicPath)) check incorrectly accepts a sibling output directory whose name shares the configured public-directory prefix. An unauthenticated requester can therefore retrieve a sibling prerendered HTML page when that page contains the Angular SSG marker. The issue is limited to Windows deployments that pass relative request URLs to CommonEngine.render, have a prefix-sharing sibling output directory, and contain qualifying prerendered Angular HTML; it does not provide arbitrary file read. This issue is fixed in versions 20.3.36, 21.2.23, and 22.1.7.16d CVE-2026-1040199.0 CRI

72.6%

——22OS command injection in the Studio Space startup validation script in Amazon SageMaker Distribution 2.x before 2.14.12, 3.x before 3.9.12, 4.0.x before 4.0.11, 4.1.x before 4.1.11, 4.2.x before 4.2.8, 4.3.x before 4.3.5, and 4.4.x before 4.4.3, as used by Amazon SageMaker Unified Studio, might allow an authenticated remote user with project contributor permissions to execute arbitrary commands in another project member's Studio Space and obtain that member's temporary execution role credentials via a crafted connection resource property that is interpolated into a shell invocation without neutralization. To remediate this issue, users should upgrade to version 2.14.12, 3.9.12, 4.0.11, 4.1.11, 4.2.8, 4.3.5, or 4.4.3, as applicable to the minor line in use. Users on minor lines that have reached end of support must move to a supported minor line, because no patched version will be released for those lines. In Amazon SageMaker Unified Studio, Studio Spaces adopt the latest patch of their minor line on restart once the patched images are deployed, so no version selection is required.16d CVE-2026-1039186.5 MED

13.5%

——4oRPC is a tool that helps build APIs that are end-to-end type-safe and adhere to OpenAPI standards. Prior to 1.14.10, the @orpc/zod ZodSmartCoercionPlugin and experimental_ZodSmartCoercionPlugin collect object and record properties in plain objects and resolve shape keys through the prototype chain. A remote client that can reach a procedure with an object or record input can supply __proto__ to replace the prototype of the returned request object, allowing attacker-controlled inherited values to reach application lookups. For object schemas, keys such as constructor, toString, and __proto__ can instead resolve inherited members as Zod schemas and cause an unhandled TypeError before validation. The global Object.prototype, unrelated objects, other requests, and other users are not modified, and the availability effect is limited to crafted requests rather than persistent process-wide state. This issue is fixed in version 1.14.10.16d CVE-2026-1030366.5 MED

13.5%

——4oRPC is a tool that helps build APIs that are end-to-end type-safe and adhere to OpenAPI standards. Prior to 1.14.9, the @orpc/json-schema SmartCoercionPlugin uses JsonSchemaCoercer to collect object properties in a plain object and to resolve schema.properties entries through the prototype chain. A remote client that can reach a procedure with an object input schema can supply __proto__ to replace the prototype of the single coerced request object, or supply Object.prototype member names such as constructor and toString so inherited values are treated as sub-schemas and pass the coercer's satisfaction check. Attacker-controlled inherited properties can consequently affect handler, Object.assign, or configuration lookups, while legitimate __proto__ properties are dropped. The global Object.prototype, unrelated objects, other requests, and other users are not modified, and downstream schema validation still runs. This issue is fixed in version 1.14.9.16d CVE-2026-969408.8 HIG

40.7%

——12Weak authentication in Microsoft Exchange Server allows an authorized attacker to elevate privileges over a network.8d CVE-2026-198566.5 MED

7.5%

——2The All in One SEO WordPress plugin before 5.0.2.1 does not correctly determine which shortcodes are present in content derived from user input before deciding which ones to strip, allowing unauthenticated users to execute arbitrary shortcodes registered on the site. On sites upgraded from older versions the protection is disabled outright, making the issue reachable without any crafted input.16d CVE-2026-1039587.6 HIG

25.0%

——8Server-side request forgery in the tool server and remote agent connection handling in Loom for AWS before 1.7.0 might allow an authenticated remote user to obtain the credentials of the application's own container role and to read responses from arbitrary internal network locations, via a crafted connection address supplied when registering, updating or testing a tool server or remote agent. To remediate this issue, users should upgrade to version 1.7.0 or later.16d CVE-2026-1039576.2 MED

30.6%

——9Server-side request forgery in the OAuth2 discovery handling in Loom for AWS before 1.7.0 might allow an authenticated remote user to obtain the access token of another user of the deployment and to cause the application to issue requests to arbitrary internal network locations, via a crafted discovery document address supplied when registering a tool server or remote agent configured for delegated authentication. To remediate this issue, users should upgrade to version 1.7.0 or later.16d CVE-2026-10395610.0 CRI

39.5%

——12Missing authentication for critical function in the authentication dependency in Loom for AWS before 1.6.1 allowed remote actors to obtain super-admin authority over the agent control plane, including registering tool servers, reading stored integration credentials, and rewriting the IAM role policies attached to managed agent roles, via any request to the application API in a deployment where no identity provider is configured. To remediate this issue, users should upgrade to version 1.6.1 or later.16d CVE-2023-544059.8 CRI

47.4%

——14H3C CVM, the Cloud Virtualization Management component of the H3C CAS cloud platform, contains an unauthenticated arbitrary file upload vulnerability in the /cas/fileUpload/upload endpoint that allows remote attackers to write arbitrary files by manipulating the caller-supplied token parameter without restricting path traversal or file type. Attackers can exploit the path traversal in the token parameter to upload a malicious JSP file into a web-accessible directory and then request it to achieve remote code execution as the web-server user. Exploitation evidence was first observed by the Shadowserver Foundation on 2023-10-14.16d CVE-2020-372787.5 HIG

28.0%

——8Weaver e-Bridge contains an unauthenticated arbitrary file read vulnerability that allows remote attackers to access arbitrary files on the host system by supplying a file: URL to the downloadUrl parameter of the saveYZJFile endpoint. Attackers can exploit this flaw to read sensitive files such as /etc/passwd or configuration and credential files, and the same endpoint's support for http(s) URLs also enables server-side request forgery against internal network resources. Exploitation evidence was first observed by the Shadowserver Foundation on 2023-10-17.16d CVE-2014-1251307.5 HIG

44.1%

——13CodeArt Google MP3 Audio Player plugin (google-mp3-audio-player) for WordPress through 1.0.11 contains an unauthenticated arbitrary file read vulnerability that allows remote attackers to retrieve sensitive files by supplying a path-traversal payload in the file parameter of direct_download.php. Attackers can request paths ../../wp-config.php without authentication to download configuration files containing database credentials and secret keys, leading to full site compromise. Exploitation evidence was first observed by the Shadowserver Foundation on 2023-10-19.16d CVE-2026-64818—

—

——0Rejected reason: This CVE ID has been rejected or withdrawn by its CVE Numbering Authority.6d CVE-2026-592658.8 HIG

36.0%

——11A code execution issue in the Java integration in Apache OpenOffice v4.1.16 and earlier allows a crafted untrusted document to trigger executing arbitrary (even remote) code when opened by the user. This issue is expected to be fixed in version 4.1.17, which is in the release candidate phase. Until then, users can mitigate this issue by disabling Java runtime integration in the Preferences dialog. This prevents the attack. If this is not possible, or as an extra precaution, you can avoid opening open untrusted files entirely. Once 4.1.17 is released, upgrade to that version to fix the issue.15d CVE-2026-1048617.5 HIG

35.7%

——11probe-image-size gets image dimensions without downloading the entire file. Prior to 7.4.0, lib/parse_sync/svg.js and lib/parse_stream/svg.js use the searching regular expression /<[-_.:a-zA-Z0-9][^>]*>/, which repeatedly scans to the end of input when attacker-controlled data contains many less-than characters without a closing greater-than character. The synchronous parser converts and scans the full supplied buffer without an input cap, while the streaming parser reparses the complete accumulated SVG prefix for every received chunk. The probe.sync(), probe(stream), and probe(url) entry points can therefore block the Node.js event loop at full CPU, and attacker-controlled chunking can amplify the streaming cost. This issue is fixed in version 7.4.0.20d CVE-2026-104859—

3.8%

——1Nx is a monorepo solution for TypeScript and polyglot codebases. From 21.4.0 until 22.7.8 and from 23.0.0 until 23.1.1, the @nx/docker release pipeline builds docker tag, image lookup, and docker push invocations as shell command strings. The release.docker.repositoryName and registryUrl configuration values are interpolated into those strings and passed to /bin/sh -c, allowing shell syntax in untrusted Nx configuration to execute during nx release version or nx release publish. A pull request or repository configuration change can therefore execute commands with the release job's privileges and expose registry credentials or cloud tokens, and dry-run publishing does not prevent the vulnerable pre-check command from executing. This issue is fixed in versions 22.7.8 and 23.1.1.17d CVE-2026-104855—

19.0%

——6Wasmtime is a runtime for WebAssembly. From 46.0.0 until 46.0.2 and 47.0.3, fuel and epoch preemption checks inside bulk operations including memory.copy, table.grow, and array.copy can expose invalid intermediate state when an embedder mutates a Store in Store::epoch_deadline_callback or continues using a Store after cancellation or a trap. A cancelled non-nullable table growth can leave null elements, linear-memory growth during memory.copy can invalidate retained raw pointers, and callback-triggered garbage collection during array.copy can invalidate GC pointers, resulting in a crash, invalid memory access, or GC heap corruption. Embeddings whose callbacks only access the host data in Store<T>, and embeddings that discard a Store after timeout or epoch deadline, are not affected. This issue is fixed in versions 46.0.2 and 47.0.3.16d CVE-2026-1027959.3 CRI

19.7%

——6Improper Access Control vulnerability in Apache Traffic Server. This issue affects Apache Traffic Server: from 9.0.0 through 9.2.14, from 10.0.0 through 10.1.3. Users are recommended to upgrade to version 9.2.15 or 10.1.4, which fixes the issue. This CVE supersedes CVE-2026-41920, whose record listed the affected 9.x versions as 9.0.0 through 9.1.14 and the fixed version as 9.1.15. All 9.2.x releases before 9.2.15 are affected.20d CVE-2026-102626—

18.0%

——5An authenticated LimeSurvey Community Edition 7.4.0 user with the global Surveys: create permission can store a JavaScript-breaking value in the date_min attribute of a Date/Time question. When another user renders the affected question, LimeSurvey inserts the stored value into a single-quoted inline JavaScript literal without JavaScript-context encoding.20d CVE-2026-104854—

3.1%

——1Nx is a monorepo solution for TypeScript and polyglot codebases. From 14.6.0 until 22.7.9 and 23.1.2, Nx creates Unix domain sockets for its daemon and isolated plugin workers in shared temporary locations without owner-only directory and socket permissions. Another unprivileged local account on a shared build server, developer host, or multi-user container can discover and connect to a running socket because the transport performs no authentication and relies on filesystem containment. The daemon's PROCESS_IN_BACKGROUND request accepts a module path and invokes its default export, allowing a caller that controls a file to execute code as the account running Nx; other handlers can expose workspace file contents, project graphs, and task hashes. Disabling the daemon alone does not remove the vulnerable plugin-worker sockets, while single-user machines without another local account are not exposed. This issue is fixed in versions 22.7.9 and 23.1.2.20d CVE-2026-104853—

6.5%

——2Nx is a monorepo solution for TypeScript and polyglot codebases. From 13.10.0 until 22.7.10 and 23.2.1, Nx migration planning reads the nx-migrations.migrations value from a target package manifest without validating that it is a contained relative path. A hostile direct dependency or a package introduced through a trusted package's packageGroup can supply .. segments or an absolute path, causing nx migrate to join an escaping destination onto its temporary directory. The migration archive can then write attacker-controlled bytes outside the temporary directory, while opening the destination stream can truncate an existing writable file even when no archive entry matches. This occurs during migration planning before review of the migration list or use of --run-migrations; the vulnerable installed Nx copy is reached when the normal nx@latest handoff is bypassed with NX_USE_LOCAL, NX_MIGRATE_USE_LOCAL, NX_MIGRATE_CLI_VERSION, --run-id, or fallback after a temporary-install failure. This issue is fixed in versions 22.7.10 and 23.2.1.20d CVE-2026-1048518.8 HIG

23.4%

——7fsspec is a specification and Python implementation framework for filesystem interfaces. From 0.9.0 until 2026.6.0, fsspec.implementations.reference.ReferenceFileSystem evaluates fields from Kerchunk reference JSON documents through unrestricted jinja2.Template(...).render(...) calls in _process_references1._render_jinja, _process_templates, and _process_gen in fsspec/implementations/reference.py. A document supplied inline or fetched from an attacker-controlled URL can provide template expressions that execute Python code when the reference filesystem is opened, including through consumers such as xarray, before referenced data is read. The _process_gen path is reached whenever a document includes a gen array, while the other paths depend on template-related options and values. This issue is fixed in version 2026.6.0.17d CVE-2026-104849—

31.8%

——10Tinypool is a minimal Node.js worker thread pool implementation. Prior to 2.1.2, Tinypool reads filename from a caller-supplied options object in pool.run(task, options) without requiring an own property, so a polluted Object.prototype.filename can replace the intended worker module. Applications are affected only when they pass their own second-argument options object to pool.run(); calls without that argument use the trusted default options object. An attacker who can first pollute the prototype can cause the worker pool to load attacker-selected JavaScript and can read or modify task data with the host process's privileges. This issue is fixed in version 2.1.2.16d CVE-2026-104848—

41.6%

——12Tinypool is a minimal Node.js worker thread pool implementation. Prior to 2.1.1, Tinypool constructs ThreadPool.options from a normal options object and reads the execArgv and env worker options in dist/index.js, allowing values inherited from a polluted Object.prototype to be copied into own properties and passed to worker_threads.Worker. An attacker who can first pollute either property can cause each newly spawned worker to load attacker-selected JavaScript through command-line preload arguments or NODE_OPTIONS, resulting in code execution with the host process's privileges and possible access to CI secrets, signing material, or build artifacts. This issue is fixed in version 2.1.1.20d CVE-2026-966136.5 MED

13.7%

——4The Meari IoT Cloud Platform OpenAPI Service is vulnerable to an authorization flaw that allows authenticated users to access the complete device shadow of any device by specifying its device ID. This vulnerability exposes sensitive information, such as device credentials, owner details, network data, and telemetry, without verifying any relationship between the requester and the target device.19d CVE-2026-945444.2 MED

7.1%

——2Next.js is a React framework for building full-stack web applications. From 16.3.0 until 16.3.8, pending use cache fills for the same key are shared without separating Draft Mode requests from regular requests. An overlapping regular request can receive unauthenticated unpublished content from an editor's Draft Mode fill, while an overlapping Draft Mode request can receive published content from a regular fill. When the regular request prerenders a page, the draft-dependent content can persist in the generated page and be served to later visitors until revalidation. Sites are affected when Cache Components or experimental.useCache is enabled and cached functions return draft-dependent content. This issue is fixed in version 16.3.8.17d CVE-2026-945435.3 MED

20.1%

——6Next.js is a React framework for building full-stack web applications. From 15.0.0 until 15.5.27 and 16.3.8, self-hosted applications using the Pages Router with statically generated or Incremental Static Regeneration pages can key a response cache entry without sufficiently binding it to the source route. A request can replace one page's cache entry with content from a different route, causing the affected page to serve incorrect content to every visitor until revalidation. Applications deployed on Vercel are not affected. This issue is fixed in versions 15.5.27 and 16.3.8.16d CVE-2026-944865.4 MED

1.4%

——0Next.js is a React framework for building full-stack web applications. From 16.0.0 until 16.3.8, the next dev development server exposes a Model Context Protocol endpoint without reliably restricting cross-site requests. A malicious website visited by a developer can reach the endpoint and read the project's disk location, source code snippets from error reports, route inventory, and development logs. Production deployments do not serve this endpoint. This issue is fixed in version 16.3.8.17d CVE-2026-944855.3 MED

6.4%

——2Next.js is a React framework for building full-stack web applications. From 16.0.0 until 16.3.8, the `next dev` development server exposes a Model Context Protocol endpoint without reliably restricting cross-site requests. A malicious website visited by a developer can reach the endpoint and read the project's disk location, source code snippets from error reports, route inventory, and development logs. Production deployments do not serve this endpoint. This issue is fixed in version 16.3.8.17d CVE-2026-944844.8 MED

14.8%

——4Next.js is a React framework for building full-stack web applications. From 15.0.0 until 15.5.27 and 16.3.8, applications with a root-level catch-all page and statically generated or Incremental Static Regeneration routes can use a shared response cache key that is insufficiently scoped to the source route. A single unauthenticated crafted request can poison that cache, causing cross-user content substitution or persistent denial of service until the poisoned entry is revalidated or replaced. This issue is fixed in versions 15.5.27 and 16.3.8.17d CVE-2026-944836.5 MED

14.7%

——4Next.js is a React framework for building full-stack web applications. From 16.0.0 until 16.3.8, Image Optimization can follow attacker-controlled DNS resolution for a remote URL that matches images.remotePatterns, allowing the optimized image fetch to reach private IP addresses after the URL passes the allow-list check. Applications without images.remotePatterns are not affected. Administrators unable to upgrade should audit allow-listed hosts and avoid entries whose DNS records are not trusted. This issue is fixed in version 16.3.8.16d CVE-2026-679897.5 HIG

26.0%

——8crmne/ruby_llm at commit fa6f279847d6d7027814539d9c0dfc3bbdfd2a83 contains a polynomial-time regular expression denial-of-service condition in Mistral model capability matching on Ruby 3.1.x20d CVE-2026-519229.8 CRI

60.5%

——18agentscope v1.0.20 contains code injection in execute_shell_command (src/agentscope/tool/_coding/_shell.py). Depending on the exposed entry, an attacker can trigger attacker-controlled code or command execution.17d CVE-2026-519189.8 CRI

47.9%

——14FinRobot 1.0.0 contains code injection in CodingUtils.create_file_with_code ().17d CVE-2026-519177.3 HIG

26.4%

——8FinRobot v1.0.0 is vulnerable to Code Injection in CodingUtils.modify_code.17d CVE-2026-519167.5 HIG

35.8%

——11TransformerOptimus SuperAGI v0.0.14 contains an incorrect access control vulnerability in delete_user_knowledge in superagi/controllers/knowledges.py. In affected source snapshots, POST /knowledges/delete/{knowledge_id} deletes the selected knowledge object without requiring authentication in the route and without verifying organization ownership of the supplied knowledge_id.20d CVE-2026-519159.8 CRI

46.7%

——14TransformerOptimus SuperAGI v0.0.14 is vulnerable to Incorrect Access Control in the tool controller. In affected source snapshots, get_tool and update_tool in superagi/controllers/tool.py accept a caller-supplied tool_id and fail to verify organization ownership through the associated toolkit. A remote authenticated attacker from one organization can read or modify another organization's tool metadata through /tools/get/{tool_id} and /tools/update/{tool_id}.17d CVE-2026-519148.8 HIG

42.9%

——13TransformerOptimus SuperAGI v0.0.14 is vulnerable to Incorrect Access Control in the agent template controller. In affected source snapshots, save_agent_as_template and publish_template in superagi/controllers/agent_template.py accept caller-supplied agent_id or agent_execution_id values and do not verify that the referenced agent or execution belongs to the authenticated user's organization.17d