CVE-20XX-?
agent?

CVEs of 2026.

Every CVE published in 2026 that we hold, ranked by exploit odds.

2,349 CVEs · updated 2026-11-21 22:04 UTC

Reset all
  1. CVE-2026-92215

    A vulnerability has been found in a2ui-project a2ui up to 0.10.7. Affected by this vulnerability is the function httpx.get of the file agent_sdks/python/a2ui_agent/src/a2ui/extensions/file_resolve/file_resolver.py of the component FileResolver. The manipulation leads to server-side request forgery. It is possible to initiate the attack remotely. The identifier of the patch is 2bb8423060308bbdea8ba468dabed4fc256d18ea. To fix this issue, it is recommended to deploy a patch.

    Disclosure Server

    6.9 MEDIUM 0.5% exploit odds Fix listed
  2. CVE-2026-18973

    A vulnerability has been found in heshengtao super-agent-party up to 0.4.1. The impacted element is the function sanitize_proxy_url of the file server.py of the component extension_proxy Route. The manipulation of the argument url leads to server-side request forgery. The attack can be initiated remotely. The exploit has been disclosed to the public and may be used. The vendor was contacted early about this disclosure but did not respond in any way.

    Disclosure Server

    5.5 MEDIUM 0.5% exploit odds No fix listed
  3. CVE-2026-63764

    LMDeploy through 0.14.0, fixed in commit 03c3130, contains a server-side request forgery (SSRF) vulnerability in the _load_http_url function within the connection.py media handler, where the private-IP guard validates only the original URL without re-validating hosts after HTTP redirects. An unauthenticated attacker can submit a crafted image_url to the chat completions endpoint pointing to an attacker-controlled host that returns a redirect to a private IP or cloud-metadata endpoint, causing the server to follow the redirect and expose internal service content through the model pipeline.

    Disclosure Server Internlm Lmdeploy

    7.7 HIGH 0.5% exploit odds Fix listed
  4. CVE-2026-1616

    The $uri$args concatenation in nginx configuration file present in Open Security Issue Management (OSIM) prior v2025.9.0 allows path traversal attacks via query parameters.

    Disclosure Server Redhat Open Security Issue Management

    7.5 HIGH 0.5% exploit odds Fix listed
  5. CVE-2026-129870

    Rocboss paopao-ce 0.2.0 through 0.6.0-alpha.3 contains a path traversal vulnerability in LocalOSS DeleteObject that allows authenticated users to delete arbitrary files by submitting crafted media values. Attackers can send dot-dot-slash sequences in the avatar or media content to POST /v1/user/avatar or POST /v1/post/comment, deleting server-writable files including configuration and data.

    Disclosure Server

    7.2 HIGH 0.5% exploit odds No fix listed
  6. CVE-2026-62316

    Microsoft UFO open-source framework for intelligent automation across devices and platforms. Prior to 3.0.8, ufo/client/mcp/http_servers/linux_mcp_server.py binds a FastMCP streamable HTTP server to localhost:8010 but does not validate the Host, Origin, or Sec-Fetch-Site headers. An attacker-controlled web page can use DNS rebinding to reach the local /mcp endpoint, enumerate tool schemas through tools/list, and invoke execute_command with a valid UFO_MCP_API_KEY to read files or execute allowed operating system commands as the victim's user. This issue is fixed in version 3.0.8.

    Disclosure Server

    8.8 HIGH 0.5% exploit odds Fix listed
  7. CVE-2026-107331

    IBM DataStage on Cloud Pak for Data 5.4.0.0 could allow a remote authenticated attacker to obtain sensitive information due to an absolute-path traversal vulnerability.

    Disclosure Server IBM Datastage On Cloud Pak For Data

    9.9 CRITICAL 0.5% exploit odds Fix listed
  8. CVE-2026-53872

    picklescan before 0.0.35 contains an unsafe pickle deserialization vulnerability allowing unauthenticated attackers to read arbitrary server files by chaining io.FileIO and urllib.request.urlopen. Attackers can bypass RCE-focused blocklists to exfiltrate sensitive data like /etc/passwd to external servers.

    Disclosure Server

    8.7 HIGH 0.5% exploit odds Fix listed
  9. CVE-2026-73659

    Trigger.dev is the open-source platform for building AI workflows in TypeScript. From 4.4.2 until 4.5.0, the packet presign routes in apps/webapp/app/routes/api.v1.packets.$.ts pass a caller-controlled filename through resolveStoreProtocolForPacketPresign to generatePresignedUrl and generatePresignedRequest in apps/webapp/app/v3/objectStore.server.ts, allowing .. traversal to escape the packets/<projectRef>/<env>/ object-store prefix and enabling a project API key to read or overwrite another organization's offloaded task payloads and outputs on multi-organization self-hosted instances. This issue is fixed in version 4.5.0.

    Disclosure Server

    8.1 HIGH 0.5% exploit odds Fix listed
  10. CVE-2026-67343

    ArcadeDB versions before 26.7.2 fail to properly redact the cluster token in the GET /api/v1/server endpoint, allowing authenticated users to retrieve the arcadedb.ha.clusterToken value in cleartext. Attackers can use the leaked token with X-ArcadeDB-Cluster-Token and X-ArcadeDB-Forwarded-User headers to impersonate root and execute administrative actions including user creation, database operations, and server shutdown.

    Disclosure Server

    8.7 HIGH 0.5% exploit odds Fix listed
  11. CVE-2026-26202

    Penpot is an open-source design tool for design and code collaboration. Prior to version 2.13.2, an authenticated user can read arbitrary files from the server by supplying a local file path (e.g. `/etc/passwd`) as a font data chunk in the `create-font-variant` RPC endpoint, resulting in the file contents being stored and retrievable as a "font" asset. This is an arbitrary file read vulnerability. Any authenticated user with team edit permissions can read arbitrary files accessible to the Penpot backend process on the host filesystem. This can lead to exposure of sensitive system files, application secrets, database credentials, and private keys, potentially enabling further compromise of the server. In containerized deployments, the blast radius may be limited to the container filesystem, but environment variables, mounted secrets, and application configuration are still at risk. Version 2.13.2 contains a patch for the issue.

    Disclosure Server Kaleidos Penpot

    7.5 HIGH 0.5% exploit odds Fix listed
  12. CVE-2026-113728

    Bulwark Webmail is a self-hosted webmail client for Stalwart Mail Server. Prior to 1.9.2, the unauthenticated POST /api/fetch-ical endpoint validates a caller-supplied hostname with one DNS lookup and then passes the URL to fetch, which performs a separate lookup for the socket connection. An attacker controlling authoritative DNS can return a public address during validation and then rebind the hostname to a loopback, private, link-local, or cloud metadata address for the connection, causing the server to return up to 10 MB of an internal HTTP or HTTPS response. The same check-then-fetch pattern also existed in authenticated JMAP login and TOTP token-exchange flows. This issue is fixed in version 1.9.2.

    Disclosure Server

    7.5 HIGH 0.5% exploit odds Fix listed
  13. CVE-2026-76728

    A vulnerability in the API endpoint of HPE Networking Instant ON APs could allow an authenticated remote attacker with high privileges to conduct a server-side request forgery (SSRF) attack. Successful exploitation could allow an attacker to execute arbitrary commands as a privileged user on the underlying operating system.

    Disclosure Server

    7.2 HIGH 0.5% exploit odds No fix listed
  14. CVE-2026-91081

    Docs through 5.6.1 contains a server-side request forgery vulnerability in the cors-proxy endpoint that allows anonymous attackers to make outbound requests by providing a public document UUID. Attackers can exploit DNS time-of-check-time-of-use race conditions and shared address space bypasses to access internal network resources and exfiltrate image content.

    Disclosure Server

    6.9 MEDIUM 0.5% exploit odds No fix listed
  15. CVE-2026-85673

    LLaMA-Factory contains a server-side request forgery vulnerability in the OpenAI-compatible API multimodal media URL handler that allows unauthenticated attackers to bypass SSRF validation. The check_ssrf_url guard validates URLs once but requests.get follows redirects and re-resolves DNS without re-validation, enabling attackers to use HTTP redirects or DNS rebinding to access internal addresses and cloud metadata endpoints.

    Disclosure Server

    8.7 HIGH 0.5% exploit odds No fix listed
  16. CVE-2026-52855

    Wings is the server control plane for Pterodactyl, a free, open-source game server management panel. Prior to 1.12.3, {{config.}} placeholders in egg configuration-file templates allow a low-privileged user to read {{config.token}}, {{config.token_id}}, and {{config.docker.registries}} from the full daemon configuration. This issue is fixed in version 1.12.3.

    Disclosure Server

    9.9 CRITICAL 0.5% exploit odds Fix listed
  17. CVE-2026-67346

    Swarms through 6.8.1, fixed in commit 8b0fc9e, contains a server-side request forgery vulnerability in the _is_safe_url function that fails to validate hostnames through DNS resolution, allowing attackers to bypass the blocklist. Attackers can supply user-controlled image or audio URLs that resolve to private, loopback, or metadata addresses to reach internal services and exfiltrate credentials.

    Disclosure Server

    7.7 HIGH 0.5% exploit odds Fix listed
  18. CVE-2026-59101

    AutoBangumi before 3.2.8 contains a server-side request forgery (SSRF) vulnerability that allows unauthenticated remote attackers to probe internal network services by supplying arbitrary host values to an unprotected setup endpoint. Attackers can send requests to the POST /api/v1/setup/test-downloader endpoint during the initial setup window, causing the server to issue HTTP GET requests to internal or reserved addresses and leak information through echoed connection-error messages.

    Disclosure Server

    6.9 MEDIUM 0.5% exploit odds Fix listed
  19. CVE-2026-49238

    An issue was discovered in Canonical Multipass before version 1.16.3. The host-side SFTP server component (sshfs_server), which executes with root privileges on the host, contains a path containment bypass vulnerability within its validate_path function in src/sshfs_mount/sftp_server.cpp. The function performs a plain string prefix comparison on requested paths without path separator validation or dot-dot (..) normalization. A local attacker with root privileges inside a guest virtual machine can bypass the FUSE layer by injecting raw SFTP frames (such as an SSH_FXP_OPEN request) directly into the sshfs_server process stdin/stdout pipes via procfs. By supplying a path containing directory traversal sequences that match the allowed mount prefix, the attacker can force the host-side root process to resolve the traversal and open files outside the designated mount boundary. This allows a guest-side user to read arbitrary files on the host filesystem, resulting in a virtual machine escape.

    Disclosure Server Canonical Multipass

    8.4 HIGH 0.5% exploit odds Fix listed
  20. CVE-2026-90466

    Path traversal of 'trusted_jar_paths' in Impala 4.5.2 allows an attacker-controlled JAR to be loaded via a relative path where the prefix matches a path specified in 'trusted_jar_paths'. The startup flag 'trusted_jar_paths' references URIs for loading files from local or remote filesystems. Path traversal can't override the schema, but can result in loading a JAR that has been uploaded to a different location in that filesystem via Impala DDLs such as CREATE DATA SOURCE and CREATE TABLE. Path traversal can only be used if a trusted path exists, so this attack requires 'trusted_jar_paths' have a non-empty value configured by the Impala admin. Users are recommended to upgrade to version 4.5.3, which fixes this issue.

    Disclosure Server Apache Impala

    6.5 MEDIUM 0.5% exploit odds Fix listed
  21. CVE-2026-63044

    Server-Side Request Forgery (SSRF) vulnerability in Apache InLong.  Any authenticated user (no admin role required) can cause the InLong Manager server to make outbound HTTP requests or TCP connections to arbitrary internal hosts and ports. This issue affects Apache InLong: from 2.0.0 before 2.4.0. Users are advised to upgrade to Apache InLong's 2.4.0 or cherry-pick [1] to solve it. [1]  https://github.com/apache/inlong/pull/12130 .

    Disclosure Server Apache Inlong

    5.4 MEDIUM 0.5% exploit odds Fix listed
  22. CVE-2026-41184

    In Calico, the install-cni init container logs the rendered CNI configuration to standard output. When the configuration template uses the __SERVICEACCOUNT_TOKEN__ placeholder (Canal/Flannel-Calico deployments), the installer substitutes the live Kubernetes ServiceAccount bearer token before logging, exposing the token to any authenticated user with pods/log permission in the namespace with calico-node. The token holds patch privileges on pods/status, enabling annotation-based attacks against cluster workloads. The default kubeconfig-based authentication path is not affected. This is a direct regression of TTA-2018-001.

    Disclosure Server Tigera Calico

    6.0 MEDIUM 0.5% exploit odds Fix listed
  23. CVE-2026-44829

    Gotenberg is a Docker-powered stateless API for PDF files. In 8.32.0 and earlier, filename handling in pkg/modules/api/context.go uses filepath.Base on Linux, which does not treat backslashes as path separators, so a multipart filename containing Windows-style parent directory components survives sanitization. The original filename flows through ctx.diskToOriginal and the multi-output PDF routes into archives.FilesFromDisk and archives.Zip.Archive as the generated zip entry name. A remote attacker can submit a name such as ........\Windows\System32\evil.pdf through an upload or an upstream downloadFrom Content-Disposition header, and a Windows archive extractor can write the resulting file outside the intended extraction directory. The affected paths include /forms/pdfengines/split and other multi-output PDF, LibreOffice, and conversion routes, and exploitation can cause arbitrary file writes on a downstream Windows system when a user or process extracts the returned archive. This issue is fixed in version 8.33.0.

    Disclosure Server

    8.8 HIGH 0.5% exploit odds Fix listed
  24. CVE-2026-15280

    IBM WebSphere Application Server - Liberty 17.0.0.3 through 26.0.0.8 ND Collective Controller is affected by a path-segment injection vulnerability in the collective routing mechanism.

    Disclosure Server IBM Websphere Application Server

    7.5 HIGH 0.5% exploit odds Fix listed
  25. CVE-2026-129286

    N-able N-central before 2026.4.1 HF3 does not validate version values and download locations in software update metadata. An attacker able to supply or modify the update metadata received by the server can write files outside the download directory and execute code as the N-central service account.

    Disclosure Server N-able N-central

    9.2 CRITICAL 0.5% exploit odds No fix listed
  26. CVE-2026-124495

    When workers run as the topology owner, the Logviewer serves files that a worker has placed in its own log directory as links to other files. A user who is allowed to submit topologies can read the worker logs of other users' topologies, and other files the Logviewer process can read. Users are recommended to upgrade to version 3.2.0, which fixes the issue.

    Disclosure Server

    6.5 MEDIUM 0.5% exploit odds Fix listed
  27. CVE-2026-71845

    A flaw was found in insights-client. The setDefault() function logs the value of every environment variable it processes, including CCX_TOKEN, a bearer credential used in disconnected cluster deployments. When glog verbosity is set to level 2 or higher, the token is written in clear text to the pod log on every startup. An attacker with access to pod logs or centralized logging could obtain the credential, leading to unauthorized access to the CCX API.

    Disclosure Server Redhat Advanced Cluster Management For Kubernetes

    6.3 MEDIUM 0.5% exploit odds Fix listed
  28. CVE-2026-32750

    SiYuan is a personal knowledge management system. In versions 3.6.0 and below, POST /api/import/importStdMd passes the localPath parameter directly to model.ImportFromLocalPath with zero path validation. The function recursively reads every file under the given path and permanently stores their content as SiYuan note documents in the workspace database, making them searchable and accessible to all workspace users. Data persists in the workspace database across restarts and is accessible to Publish Service Reader accounts. Combined with the renderSprig SQL injection ( separate advisory ), a non-admin user can then read all imported secrets without any additional privileges. This issue has been fixed in version 3.6.1.

    Disclosure Server b3log Siyuan

    6.8 MEDIUM 0.5% exploit odds Fix listed
  29. CVE-2026-68903

    LLMKube is a Kubernetes operator for self-hosted large language model inference across heterogeneous hardware. Prior to 0.9.0, the ModelReconciler accepts arbitrary absolute paths and file:// URIs in Model.spec.source without restricting them to approved roots. A namespace user who can create or update Model and InferenceService resources can cause copyLocalModel to open an attacker-selected file in the controller pod or cause a HostPathVolumeSource to mount an attacker-selected node file when modelCache.enabled is enabled, which is the default. The selected file is copied into model storage and can expose controller service-account credentials or other node secrets outside the attacker's namespace. This issue is fixed in version 0.9.0.

    Disclosure Server Defilantech LLMKube

    7.7 HIGH 0.5% exploit odds Fix listed
  30. CVE-2026-79880

    A vulnerability in the command line interface of AOS-8 could allow an authenticated remote attacker with high privileges to conduct a server-side request forgery (SSRF) attack. Successful exploitation could allow an attacker to execute arbitrary commands on the underlying operating system.

    Disclosure Server

    8.7 HIGH 0.5% exploit odds No fix listed
  31. CVE-2026-77266

    MCP Atlassian is a Model Context Protocol (MCP) server for Atlassian products (Confluence and Jira). Prior to 0.22.0, upload_attachment accepts absolute paths and traversal sequences without constraining the resolved path to the server workspace. An MCP caller with attachment access can read a chosen server-local file and exfiltrate it through Jira or Confluence. The advisory traces the vulnerable input and processing flow through upload_attachment, file_path, and path traversal, which identify the affected entry points, controls, and code paths. This issue is fixed in version 0.22.0.

    Disclosure Server mcp-atlassian Mcp Atlassian

    6.5 MEDIUM 0.5% exploit odds Fix listed
  32. CVE-2026-77270

    MCP Atlassian is a Model Context Protocol (MCP) server for Atlassian products (Confluence and Jira). Prior to 0.22.0, the Jira and Confluence attachment upload tools treat caller-controlled file_path values as trusted server-local paths. The server opens the selected file and uploads it to an Atlassian issue or page, allowing an MCP caller with upload access to disclose any file readable by the server process. The advisory traces the vulnerable input and processing flow through confluence_upload_attachment, jira_upload_attachment, file_path, and open(file_path, "rb"), which identify the affected entry points, controls, and code paths. This issue is fixed in version 0.22.0.

    Disclosure Server mcp-atlassian Mcp Atlassian

    6.5 MEDIUM 0.5% exploit odds Fix listed
  33. CVE-2026-76900

    CordysCRM is an open source AI-powered customer relationship management system that supports private deployment. In version 1.7.3, ApprovalResourceService.sendWebHook reads WebHookConfig.webHookUrl from stored approval-node configuration and passes it through ApprovalFlowService.updateApprovalPostField to HttpClientUtils without the SSRF validation used by the optional testConnect path. A user with PROCESS_SETTING_ADD can configure an internal URL through POST /approval-flow/add and cause the server to request it when POST /approval-action/approve executes the approval action, enabling cloud metadata access, internal network reconnaissance, and interaction with reachable internal services. This issue is fixed in version 1.7.4.

    Disclosure Server

    6.8 MEDIUM 0.5% exploit odds Fix listed
  34. CVE-2026-85609

    Openpanel before 2.3.0 contains an unauthenticated full-read server-side request forgery (SSRF) vulnerability in the GET /tools/site-checker endpoint (apps/api/src/controllers/tools.controller.ts). The endpoint passes a user-supplied url query parameter to fetchWithRedirects() and performs server-side HTTP requests to arbitrary URLs without any SSRF/IP validation. An unauthenticated remote attacker can access cloud instance metadata endpoints, probe internal services, scan internal network ports, and read returned content (status code, page size, timing, and parsed HTML metadata), and leak internal IP addresses (via getIPInfo() to a third party).

    Disclosure Server

    6.9 MEDIUM 0.5% exploit odds Fix listed
  35. CVE-2026-25199

    Instances deployed via the Proxmox extension allow unauthorized access to instances belonging to other tenants. This issue affects Apache CloudStack: from 4.21.0.0 through 4.22.0.0. The Proxmox extension for CloudStack improperly uses a user-editable instance setting, proxmox_vmid, to associate CloudStack instances with Proxmox virtual machines. Because this value is not restricted or validated against tenant ownership and Proxmox VM IDs are predictable, a non-privileged attacker can modify the setting to reference a VM belonging to another account. This allows unauthorized cross-tenant access and enables full control over the targeted VM, including starting, stopping, and destroying the virtual machine. Users are recommended to upgrade to version 4.22.0.1, which fixes this issue. As a workaround for the existing installations, editing of the proxmox_vmid instance detail by users can be prevented by adding this detail name to the global configuration parameter - user.vm.denied.details.

    Disclosure Server Apache Cloudstack

    9.1 CRITICAL 0.5% exploit odds Fix listed
  36. CVE-2026-107045

    Insecure permissions in the CitrineContentProvider component of Citrine v3.1.0 (versionCode 62) allows unauthorized attackers to arbitrarily read another's Nostr event database.

    Disclosure Server

    9.1 CRITICAL 0.5% exploit odds No fix listed
  37. CVE-2026-107021

    Insecure permissions in the CardProvider content provider component of helloworld1 AnyMemo versionCode 20260730 allows unauthorized attackers to arbitrarily read flashcards across the user's entire memorization database.

    Disclosure Server

    9.1 CRITICAL 0.5% exploit odds No fix listed
  38. CVE-2026-85961

    On Azure deployments, Progress MarkLogic Server 12.x releases before 12.1.0 and 11.x releases before 11.3.7 may record security-sensitive authentication information in diagnostic logs under certain configurations. An attacker with access to the affected logs could use this information to obtain unauthorized access to protected cloud resources.

    Disclosure Server

    7.5 HIGH 0.5% exploit odds No fix listed
  39. CVE-2026-54688

    mcp-searxng is a Model Context Protocol server that gives AI assistants web search and URL-reading capabilities through SearXNG. Prior to 1.2.0, web_url_read passes a caller-supplied URL to the server-side fetch path while assertUrlAllowed() in src/url-reader.ts runs only when MCP_HTTP_HARDEN is enabled, even though MCP_HTTP_HARDEN is disabled by default in src/http-security.ts. In the default configuration, an attacker who influences the URL selected by a user or AI agent can make the server fetch loopback, private-network, or cloud metadata endpoint resources and return their contents into the model context. file:// URLs remain rejected, and the separate DNS-resolution and redirect-validation bypasses are outside this record. This issue is fixed in version 1.2.0.

    Disclosure Server

    6.5 MEDIUM 0.5% exploit odds Fix listed
  40. CVE-2026-47394

    PraisonAI is a multi-agent teams system. Prior to version 4.6.40, the fix for GHSA-9mqq-jqxf-grvw / CVE-2026-44336 is incomplete. The original advisory description named four vulnerable handlers in `mcp_server/adapters/cli_tools.py`. Commit `68cc9427` ("fix(security): harden MCP rules path handling…") added a `_resolve_rule_path()` helper and applied it to `rules.create`, `rules.show`, and `rules.delete`. `workflow.show` was left unchanged. Two adjacent handlers in the same file have the same pattern, `workflow.validate` and `deploy.validate`. Neither was mentioned in the original advisory. Both remained unchanged. The original advisory also identified the dispatcher (`server.py:281-298`) as a root cause. It accepts unvalidated `**kwargs` from `params["arguments"]` with no enforcement against the tool's declared `input_schema`. That code is unchanged prior to version 4.6.40. A single unauthenticated MCP `tools/call` to `praisonai.workflow.show` returns the contents of any file the host user can read: `/etc/passwd`, `~/.ssh/id_rsa`, `~/.aws/credentials`, or any project `.env`. Version 4.6.40 contains an updated fix.

    Disclosure Server

    8.7 HIGH 0.5% exploit odds Fix listed
  41. CVE-2026-47356

    Terrascan v1.18.3 and prior are vulnerable to Server-Side Request Forgery (SSRF) via the webhook_url parameter in the file scan endpoint (POST /v1/{iac}/{iacVersion}/{cloud}/local/file/scan) when running in server mode. An unauthenticated remote attacker can supply an arbitrary URL as the webhook_url multipart form parameter. After scanning the uploaded file, Terrascan sends an HTTP POST request to the attacker-controlled URL containing the full scan results as a JSON body, with the attacker-supplied webhook_token forwarded as a Bearer token in the Authorization header. The retryable HTTP client retries up to 10 times on failure. This affects deployments running terrascan in server mode (terrascan server), which binds to 0.0.0.0 with no authentication. Note: Terrascan was archived in August 2023 and no patch will be released.

    Disclosure Server Tenable Terrascan

    8.7 HIGH 0.5% exploit odds No fix listed
  42. CVE-2026-67613

    CyberPanel before 3.0.0 contains a path traversal vulnerability that allows authenticated administrators to read arbitrary files from the server filesystem by supplying unsanitized file paths to the cloudAPI ReadReport endpoint. Attackers can manipulate the reportFile parameter in the JSON request body, which is passed directly to open() in cloudManager.py without validation or allowlisting, enabling traversal to any file readable by the root-privileged CyberPanel process including credential files, SSL and SSH private keys, and JWT secret files.

    Disclosure Server

    6.9 MEDIUM 0.5% exploit odds No fix listed
  43. CVE-2026-72581

    A server-side request forgery (SSRF) vulnerability in duhow/xiaoai-patch through commit fb07049 allows a remote attacker to make the Xiaomi smart speaker perform HTTP requests to arbitrary internal or external URLs. The /auth endpoint in api/main.py uses the user-supplied url POST parameter to redirect to a Home Assistant instance without validating the destination URL, enabling internal network scanning and access to internal services.

    Disclosure Server

    8.6 HIGH 0.5% exploit odds No fix listed
  44. CVE-2026-53446

    Wekan is open source kanban built with Meteor. Prior to 9.32, Wekan webhook integration URLs in models/integrations.js are stored from user input and later fetched by server/notifications/outgoing.js without applying the existing validateAttachmentUrl() private-network checks from models/lib/attachmentUrlValidation.js. A board administrator can configure webhook URLs that cause server-side requests to internal or metadata services. This issue is fixed in version 9.32.

    Disclosure Server

    6.2 MEDIUM 0.5% exploit odds Fix listed
  45. CVE-2026-24147

    NVIDIA Triton Inference Server contains a vulnerability in triton server where an attacker may cause an information disclosure by uploading a model configuration. A successful exploit of this vulnerability may lead to information disclosure or denial of service.

    Disclosure Server Nvidia Triton Inference Server

    4.8 MEDIUM 0.5% exploit odds Fix listed
  46. CVE-2026-85608

    Douyin_TikTok_Download_API through 4.1.2 contains a server-side request forgery vulnerability in the /api/download and /api/hybrid/video_data endpoints that allows unauthenticated attackers to fetch arbitrary URLs by supplying a url query parameter. Attackers can request internal services including cloud metadata endpoints and retrieve response bodies containing sensitive credentials through error messages.

    Disclosure Server

    8.7 HIGH 0.5% exploit odds No fix listed
  47. CVE-2026-85180

    Ollama fails to validate redirect destinations when pulling tensor-layer models, allowing unauthenticated attackers to redirect blob downloads to arbitrary hosts. An attacker can control a registry, serve a malicious tensor-layer manifest, and cause the server to issue GET requests to internal hosts including cloud metadata endpoints.

    Disclosure Server

    8.7 HIGH 0.5% exploit odds No fix listed
  48. CVE-2026-82638

    jina-ai reader disables its private-address guard outside Google Cloud deployments, allowing unauthenticated attackers to perform server-side request forgery. Attackers can supply publicly resolvable hostnames mapping to private addresses to retrieve cloud metadata and internal service content.

    Disclosure Server

    8.7 HIGH 0.5% exploit odds No fix listed
  49. CVE-2026-77680

    An algorithmic complexity flaw exists in libsoup's HTTP Range header processing that persists after the CVE-2025-32907 fix. CVE-2025-32907 addressed memory amplification when a client repeated the same range many times in a single Range header. Commit 9bb92f7a corrected merge correctness in soup_message_headers_get_ranges_internal() in libsoup/soup-message-headers.c, but the coalescing loop still removes merged ranges using g_array_remove_index() for each coalesced element. Because GArray is contiguous, each mid-array removal performs an O(N) memmove. When many identical satisfiable ranges are supplied (for example bytes=0-0 repeated thousands of times), the loop performs O(N²) work coalescing them into a single range. The vulnerable path is reachable server-side from handle_partial_get() in libsoup/server/http1/soup-server-message-io-http1.c when a SoupServer handler returns HTTP 200 with a non-empty body. No authentication is required. The number of ranges is bounded only by the maximum request header size (~100 KiB), allowing roughly 25,000 ranges per request. Reporter measurements on libsoup HEAD containing the CVE-2025-32907 fix show ~90 ms single-core CPU per such request at the wire maximum, blocking the server's event loop for that duration. This is a CPU exhaustion / availability issue only. No memory corruption or information disclosure occurs. Affected: libsoup versions containing the CVE-2025-32907 fix but not merge request !550. Fixed upstream: MR !550 merged 2026-08-20, replacing per-element removal with O(N) in-place compaction and rejecting Range headers requesting more than 200 ranges. Upstream report: https://gitlab.gnome.org/GNOME/libsoup/-/issues/538 Related: CVE-2025-32907

    Disclosure Server

    5.3 MEDIUM 0.5% exploit odds No fix listed
  50. CVE-2026-67435

    linuxfabrik-lib provides Python modules for database access, caching, shell execution, and API integrations. Prior to version 6.0.0, lib.url.fetch() followed cross-origin redirects while forwarding caller-supplied credential headers other than Authorization and Cookie, allowing a malicious redirect-capable server to receive headers such as X-Auth-Token from authenticated monitoring requests. This issue is fixed in version 6.0.0.

    Disclosure Server

    6.0 MEDIUM 0.5% exploit odds Fix listed

CVE records published in 2026 (ids starting CVE-2026-), sorted by how likely they are to be exploited.