CVE Feed

Server-side request forgery (ssrf) in Azure Data Factory allows an unauthorized attacker to disclose information over a network.

An issue in usememos through v0.30.0 allows a remote authenticated attacker to perform Server-Side Request Forgery (SSRF) via the Webhook validation mechanism in internal/webhook/validate.go, by setting a webhook target to an internal address.

A flaw was found in the multicloud-operators-channel component. This vulnerability allows a user with specific permissions to manipulate how the system handles sensitive information, known as Secrets, across different parts of the system (namespaces). By exploiting this, an attacker can modify these Secrets in unauthorized areas. This could lead to unauthorized access to information or elevated privileges within the system.

Inconsistent interpretation of http requests ('http request/response smuggling') in .NET allows an unauthorized attacker to bypass a security feature over a network.

An issue in usememos v0.27.1 allows a remote attacker to achieve account takeover via the ssoCredentials branch of the SignIn handler in server/router/api/v1/auth_service.go, because SSO identity is matched only on an attacker-controllable identifier without binding to the IdP's stable subject claim.

Heap-based buffer overflow in Windows Key Guard allows an authorized attacker to elevate privileges locally.

Uncaught exception in .NET allows an authorized attacker to elevate privileges locally.

Unchecked input for loop condition in .NET allows an unauthorized attacker to deny service over a network.

Improper removal of sensitive information before storage or transfer in .NET allows an unauthorized attacker to disclose information over a network.

The UpdateHub management subsystem (subsys/mgmt/updatehub/updatehub.c) drives every update operation through a single file-scope ctx structure that holds the CoAP block context, payload buffer, status code, socket, and a one-element poll-fd array fds[1]. Access to ctx was not serialized, and prepare_fds() wrote ctx.fds[ctx.nfds] and incremented ctx.nfds with no bounds check. Two independent paths mutate ctx concurrently: the background autohandler running on the system workqueue, and user-triggered operations reached through the updatehub run shell command, direct API calls, or — since the operations are exposed as syscalls — userspace threads. When a second flow enters prepare_fds() while ctx.nfds is already 1, the write lands one element past the array; by struct layout it overlaps the adjacent ctx.sock/ctx.nfds members. More broadly, the unsynchronized sharing lets two flows interleave connection setup and teardown, double-closing a socket descriptor or scribbling the shared buffers. The result is corruption of the update subsystem's internal state and denial of service of the firmware-update path; the out-of-bounds write is contained within the ctx structure and there is no demonstrated path to memory outside it or to code execution. Triggering requires a local actor able to invoke update operations (or, with CONFIG_USERSPACE, an unprivileged userspace thread) and to win a timing race against the background handler; remote peers cannot control the race timing. The fix serializes the entry points with a mutex and adds a bounds check to prepare_fds().

The UpdateHub over-the-air update client's start_coap_client() in subsys/mgmt/updatehub/updatehub.c leaks the CoAP/DTLS socket descriptor on its connection-setup failure paths. The shared error: cleanup gated socket closing on a ret > 0 flag, but ret was set to -1 immediately after the socket was created, so when zsock_setsockopt() (DTLS) or zsock_connect() subsequently failed the gate was false and cleanup_connection() was never called. The open descriptor in the global ctx.sock was then overwritten by the next attempt, permanently leaking it from the socket / net_context pool until reboot. The failing setup path is reached every time the OTA client tries to contact the UpdateHub server and the connection cannot be established — driven automatically by the periodic autohandler() poll (and on demand via the updatehub_probe()/updatehub_update() API or the updatehub run shell command). The DTLS handshake/connect outcome is influenceable by a network or on-path attacker who drops, resets, or otherwise disrupts traffic to the server, and also fails naturally whenever the server is unreachable. Each failed attempt permanently leaks one descriptor; once the shared socket pool is exhausted, networking degrades device-wide until the device is rebooted, a denial-of-service condition. Severity is low because the leak rate is bounded by the configured OTA poll interval (default once per 24 hours), the effect is gradual and recovered by reboot, and only builds with the UpdateHub client enabled are affected. There is no memory-corruption, information-disclosure, or authentication impact.

tls_opt_dtls_peer_connection_id_value_get() in subsys/net/lib/sockets/sockets_tls.c, which handles getsockopt(SOL_TLS, TLS_DTLS_PEER_CID_VALUE), passed the caller-supplied optval directly to mbedtls_ssl_get_peer_cid() without verifying the buffer was at least MBEDTLS_SSL_CID_OUT_LEN_MAX (default 32) bytes. mbedtls_ssl_get_peer_cid() copies the peer-negotiated DTLS Connection ID (length 1..MBEDTLS_SSL_CID_OUT_LEN_MAX) into that buffer without a destination-size parameter, so a caller-supplied optlen smaller than the CID causes a write of up to 31 bytes past the buffer end. In CONFIG_USERSPACE builds the getsockopt syscall verifier (z_vrfy_zsock_getsockopt) bounce-buffers the user's optval into a kernel allocation of exactly optlen bytes (k_usermode_alloc_from_copy -> z_thread_malloc), so an unprivileged user thread that passes a small optlen on a connected DTLS socket with Connection ID enabled induces a kernel-heap buffer overflow, with the overflowing content being the remote peer's CID. The defect requires CONFIG_MBEDTLS_SSL_DTLS_CONNECTION_ID, an established DTLS session with a negotiated peer CID, and (for the kernel-crossing case) CONFIG_USERSPACE. Introduced when the TLS_DTLS_CID option was added (v3.5.0). The fix rejects callers whose optlen is below MBEDTLS_SSL_CID_OUT_LEN_MAX with -EINVAL.

react18-use is a React 19 use hook shim. Between 2026-05-19 01:07:01 and 2026-05-19 15:20:43, the default branch contained malicious commits 7b79148d1495a2505f9277da295a98cf176f4496 through 7b79148d1495a2505f9277da295a98cf176f4496 that executed remote attacker-controlled code on developer machines during `npm install`. The commits were removed by force-push, but local clones, forks, and direct-SHA URLs may still contain them, and `npm install` against an affected checkout will still execute the code today. The package was not published to npm. `src/install.js` was added and wired into the `postinstall` script. It fetched a JavaScript payload from an attacker-controlled HTTPS endpoint (configurable via an environment variable), disabled TLS verification, and evaluated the response as code with `require` available. Execution was deliberately skipped on CI and cloud/serverless environments, targeting developer workstations. The second-stage payload was attacker-hosted and cannot be reconstructed. Assume full compromise of anything reachable from a Node process with the user's permissions. Those who ran `npm install` against an affected checkout on a developer machine on or after 2026-05-19 01:07:01 should treat the machine as compromised, rotate every credential the machine could reach, audit account activity since 2026-05-19 01:07:01, and clean local clones.

Certain web interface components in affected TP-Link Aginet devices do not validate and sanitize user-supplied input properly before passing it to system-level command execution functions.  An authenticated adjacent attacker may inject specially crafted input to execute arbitrary operation system commands with elevated privileges. Successful exploitation may allow execution of arbitrary system commands, potentially leading to full device compromise.

The affected TP-Link Aginet devices do not properly validate symbolic links created on external USB storage devices. By placing a crafted symbolic link on supported storage media, an attacker may cause the system to resolve the link. Successful exploitation may allow unauthorized read access to sensitive files within the device filesystem.

In affected TP-Link Aginet devices, use of hardcoded cryptographic keys embedded in the firmware to protect sensitive configuration data may allow an attacker who has access to device storage to recover the keys and decrypt stored data. Successful exploitation may allow access to decrypted sensitive configuration data, including credentials and service-related information.

In affected TP-Link Aginet devices, insufficient authorization validation allows authenticated low-privileged users to execute higher-privileged operations. An attacker may perform administrative actions such as creating privileged accounts or modifying critical configuration settings.

The affected TP-Link Aginet devices contain a flaw in the web management interface where authentication checks are not consistently enforced on certain endpoints. An attacker can send specially crafted requests to bypass authentication and directly invoke privileged functionality without valid credentials. This issue arises from improper enforcement of access control mechanisms on sensitive operations. Successful exploitation may allow an unauthenticated attacker to execute privileged operations and gain full control of the device.

Cross-Site Request Forgery weaknesses in the Administrative Console of PingFederate versions before version 13.1 may allow actors to perform unauthorized actions via specially-crafted links triggered by administrators with active sessions.

Off-by-one error in Microsoft Office Word allows an unauthorized attacker to disclose information locally.

Rocket.Chat is an open-source, secure, fully customizable communications platform. Prior to 7.10.14, 8.0.8, 8.1.7, 8.2.7, 8.3.7, 8.4.5, 8.5.2, and 8.6.1, the channels.convertToTeam REST endpoint allows an authenticated registered user with the create-team permission to convert an unrelated public channel by supplying channelName instead of channelId because the edit-room permission is checked only for channelId. This issue is fixed in versions 7.10.14, 8.0.8, 8.1.7, 8.2.7, 8.3.7, 8.4.5, 8.5.2, and 8.6.1.

Rocket.Chat is an open-source, secure, fully customizable communications platform. Prior to 7.10.14, 8.0.8, 8.1.7, 8.2.7, 8.3.7, 8.4.5, 8.5.2, and 8.6.1, the stream-notify-user stream in the WebSocket protocol allows an authenticated user to write arbitrary notification bodies because the sender is not checked, and the client-side UI can create an ephemeral fake message in another user's currently open chat. This issue is fixed in versions 7.10.14, 8.0.8, 8.1.7, 8.2.7, 8.3.7, 8.4.5, 8.5.2, and 8.6.1.

AnythingLLM is an application that turns pieces of content into context that any LLM can use as references during chatting. From 1.0.0 to 1.15.0, AnythingLLM's unauthenticated account-recovery flow in server/utils/PasswordRecovery/index.js uses recoverAccount() to deduplicate the raw recoveryCodes values before trimming them, so one valid code submitted twice with different surrounding whitespace can satisfy the two-code check. Each normalized value can also match the same stored hash instead of consuming a distinct hash. An attacker who knows the target username and one recovery code can call POST /api/system/recover-account in multi-user mode, receive a password-reset token, and use POST /api/system/reset-password to take over the account, including an administrator account.

Mastodon is a free, open-source social network server based on ActivityPub. Prior to 4.4.21, 4.5.14, 4.6.4, and 4.7.0-beta.1, PrivateAddressCheck.private_address? in app/lib/private_address_check.rb normalized IPv4-mapped IPv6 addresses but did not recognize IPv4-compatible IPv6 addresses with IPAddr#ipv4_compat?. An attacker could supply an address in the omitted range to bypass the ALLOWED_PRIVATE_ADDRESSES protection and make Mastodon send HTTP requests to loopback interfaces, potentially accessing private resources and services. Exploitation requires a system that supports the obsolete IPv4-compatible IPv6 mechanism. This issue is fixed in versions 4.4.21, 4.5.14, 4.6.4, and 4.7.0-beta.1.

Mastodon is a free, open-source social network server based on ActivityPub. From 4.6.0-beta.1 until 4.6.4 and 4.7.0-beta.1, any logged-in local user could use the show action in app/controllers/admin/collections_controller.rb to access personally identifying information about another local user in a collection because the controller used the general collection policy instead of the admin collection policy namespace. The exposed data included the other user's current email address and last-used IP address. This issue is fixed in versions 4.6.4 and 4.7.0-beta.1.

Mastodon is a free, open-source social network server based on ActivityPub. Prior to 4.4.21, 4.5.14, 4.6.4, and 4.7.0-beta.1, the administrative statistics endpoints handled by Api::V1::Admin::MeasuresController and Api::V1::Admin::RetentionController checked authorization only after beginning expensive calculations. Anonymous callers could submit keys, start_at, and end_at parameters that caused long-running SQL queries in Admin::Metrics::Measure, Admin::Metrics::Retention, and Admin::Metrics::Dimension::BaseDimension, allowing repeated requests to exhaust server resources. This issue is fixed in versions 4.4.21, 4.5.14, 4.6.4, and 4.7.0-beta.1.

react-tracked provides state usage tracking with Proxies. Between 2026-05-18 19:26:36 and 2026-05-19 15:22:45, the default branch contained malicious commits 6978272a7d6ca02225cb747ea69f427512e33699 through 949f1a3d6bb1ff7d1a0dec892afd773e742627e8 that executed remote attacker-controlled code on developer machines during `npm install`. The commits were removed by force-push, but local clones, forks, and direct-SHA URLs may still contain them, and `npm install` against an affected checkout will still execute the code today. The package was not published to npm. `src/install.js` was added and wired into the `postinstall` script. It fetched a JavaScript payload from an attacker-controlled HTTPS endpoint (configurable via an environment variable), disabled TLS verification, and evaluated the response as code with `require` available. Execution was deliberately skipped on CI and cloud/serverless environments, targeting developer workstations. The second-stage payload was attacker-hosted and cannot be reconstructed. Assume full compromise of anything reachable from a Node process with the user's permissions. Those who ran `npm install` against an affected checkout on a developer machine on or after 2026-05-18 19:26:36 should treat the machine as compromised, rotate every credential the machine could reach, audit account activity** since 2026-05-18 19:26:36, and clean local clones.

Kitty is a cross-platform GPU based terminal. Prior to 0.48.2, the @kitty-echo and @kitty-ssh DCS handlers in kitty/window.py write unauthenticated data to the child shell's stdin, where handle_remote_echo accepts printable shell command characters and handle_remote_ssh calls get_ssh_data in kittens/ssh/utils.py, which emits a newline; chaining the handlers can execute attacker-controlled commands when a user displays untrusted terminal data. This issue is fixed in version 0.48.2.

This CVE ID has been rejected or withdrawn by its CVE Numbering Authority.

This CVE ID has been rejected or withdrawn by its CVE Numbering Authority.

This CVE ID has been rejected or withdrawn by its CVE Numbering Authority.

This CVE ID has been rejected or withdrawn by its CVE Numbering Authority.

This CVE ID has been rejected or withdrawn by its CVE Numbering Authority.

This CVE ID has been rejected or withdrawn by its CVE Numbering Authority.

This CVE ID has been rejected or withdrawn by its CVE Numbering Authority.

CyberChef is a web app for encryption, encoding, compression, and data analysis. Prior to 11.3.0, CyberChef's pretty-recipe parser in src/core/Utils.mjs can exhaust client-side CPU when a malformed #recipe= URL fragment containing a large number of unmatched quote characters reaches Utils.parseRecipeConfig(). The function synchronously applies a complex global regular expression that may perform heavy backtracking before rejecting the input, causing the victim's browser tab to freeze during startup for seconds or longer. No code execution, data exfiltration, or privilege escalation occurs. This issue is fixed in version 11.3.0.

ERPNext is a free and open source Enterprise Resource Planning tool. Prior to 15.118.0 and 16.29.0, the validate_template and render_template calls in erpnext/accounts/doctype/process_statement_of_accounts/process_statement_of_accounts.py render subject, body, and pdf_name fields with unrestricted globals including frappe.utils, allowing an authenticated user with a common operational role to inject template expressions, execute arbitrary server-side code, and read data across the application. This issue is fixed in versions 15.118.0 and 16.29.0.

ERPNext is a free and open source Enterprise Resource Planning tool. Prior to 15.112.0 and 16.22.0, the merge_account, pause_job_for_doc, trigger_job_for_doc, change_release_date, and update_cost_center functions across erpnext/accounts/doctype/account/account.py, erpnext/accounts/doctype/process_payment_reconciliation/process_payment_reconciliation.py, erpnext/accounts/doctype/purchase_invoice/purchase_invoice.py, and erpnext/accounts/utils.py omit required write permission checks, allowing authenticated limited users to modify protected data beyond their roles. This issue is fixed in versions 15.112.0 and 16.22.0.

ERPNext is a free and open source Enterprise Resource Planning tool. Prior to 15.112.0 and 16.23.0, the ReceivablePayableReport prepare_conditions path in erpnext/accounts/report/accounts_receivable/accounts_receivable.py does not apply Customer and Supplier user permissions to the Payment Ledger Entry dynamic-link party field, allowing any authenticated user to read unauthorized cross-company financial data in Accounts Receivable and Accounts Payable reports. This issue is fixed in versions 15.112.0 and 16.23.0.

ERPNext is a free and open source Enterprise Resource Planning tool. Prior to 15.109.0 and 16.20.0, the get_tax_template function in erpnext/accounts/doctype/tax_rule/tax_rule.py constructs an SQL WHERE clause from request-influenced posting_date and args values, allowing an authenticated low-privilege user to inject SQL and extract sensitive information. This issue is fixed in versions 15.109.0 and 16.20.0.

ERPNext is a free and open source Enterprise Resource Planning tool. Prior to 15.111.0 and 16.22.0, the add_ac function in erpnext/accounts/utils.py accepts the ignore_permissions argument without enforcing Account create permission, allowing an authenticated limited user to create unauthorized accounting master records and affect financial data integrity and audit trails. This issue is fixed in versions 15.111.0 and 16.22.0.

ERPNext is a free and open source Enterprise Resource Planning tool. Prior to 15.111.0 and 16.22.0, the send_auto_email function in erpnext/accounts/doctype/process_statement_of_accounts/process_statement_of_accounts.py lacks a Process Statement Of Accounts permission check, allowing an authenticated low-privilege user to trigger automated emails outside the permitted role. This issue is fixed in versions 15.111.0 and 16.22.0.

Firecrawl turns entire websites into LLM-ready markdown or structured data. Prior to 2.11.32, a critical arbitrary file read vulnerability exists in Firecrawl's extraction functionality due to unsafe schema dereferencing of user-supplied JSON schemas in apps/api/src/lib/extract/helpers/dereference-schema.ts. The affected code invokes the json-schema-ref-parser dependency with default resolver settings, allowing external and local file references to be resolved during schema processing. An authenticated attacker can supply a malicious schema containing a $ref within default, const, or enum fields that are not traversed by AJV validation. By triggering a dereference error, file contents from the extract worker filesystem may be included in persisted error messages returned through the extraction API, enabling arbitrary file reads and SSRF against internal or external HTTP endpoints. This issue is fixed in version 2.11.32.

Tabby (formerly Terminus) is a highly configurable terminal emulator. Prior to 1.0.235, a malicious SFTP server can return a backslash traversal filename through entry.name. In tabby-ssh/src/session/sftp.ts, SFTPSession.readdir() and _makeFile() use POSIX path processing that preserves the backslashes as ordinary filename characters. In tabby-ssh/src/components/sftpPanel.component.ts, downloadFolderRecursive() propagates item.name into the local relative path. In tabby-electron/src/services/platform.service.ts, ElectronDirectoryDownload.createFile() passes that path to Windows-native path.join(), and in tabby-electron/src/sftpContextMenu.ts, EditSFTPContextMenu.edit() passes item.name to path.join() for the temporary edit path. Windows interprets the preserved backslashes and parent-directory components as traversal, allowing attacker-controlled content to be created or overwritten outside the selected download directory or temporary edit directory. This issue is fixed in version 1.0.235.

The UpdateHub OTA client in subsys/mgmt/updatehub/updatehub.c contains an out-of-bounds / uninitialized-memory read in z_impl_updatehub_probe(). The probe response from the UpdateHub server is copied into a heap buffer (metadata) that is correctly NUL-terminated, but a second buffer (metadata_copy) is allocated with k_malloc (unzeroed) and filled with memcpy(metadata_copy, metadata, strlen(metadata)), which omits the terminating NUL. Everything after the copied content remains uninitialized heap. When the first json_obj_parse() over the array descriptor fails, the code falls back to json_obj_parse(metadata_copy, strlen(metadata_copy), ...). The strlen() call scans past the copied bytes through uninitialized heap and, if no zero byte is found before the end of the allocation, reads beyond the buffer; the resulting over-long length is then parsed as JSON. The probe payload is fully controlled by the (malicious, compromised, or — without the optional CONFIG_UPDATEHUB_DTLS — on-path) UpdateHub server, which can craft a large payload that fails the first parse to drive this path. The consequence is a read of uninitialized heap, with a worst case of an out-of-bounds read past the metadata_copy allocation that can fault and crash the update thread/device, producing a network-triggerable denial of service. The over-read data is consumed only internally to evaluate the update and is not returned to the attacker, so there is no direct information disclosure and no out-of-bounds write. The fix zeroes metadata_copy with memset before the copy, guaranteeing NUL termination and bounding strlen() within the allocation.

The UpdateHub firmware-update agent's probe handler (z_impl_updatehub_probe() in subsys/mgmt/updatehub/updatehub.c) parses the JSON metadata returned by the update server into a fixed two-level nested-array struct. After parsing it validates only the outer array length (objects_len != 2) and then dereferences objects[1].objects[0].objects.sha256sum via strlen() without checking that the inner object array of element [1] is non-empty. The metadata is attacker-influenceable network input: the agent fetches it over CoAP from the configured UpdateHub server during its routine OTA probe. A malicious or compromised update server (or, when DTLS is disabled, a network man-in-the-middle) can return a response whose second outer object array is empty. Because the parse target is zero-initialised, the corresponding objects[1].objects[0].objects.sha256sum pointer is NULL, and the subsequent strlen() dereferences address zero. The same defect exists in both the 'any boards' and 'some boards' metadata layouts. The resulting CPU fault is fatal under Zephyr's default error handling, halting or resetting the device, so the flaw is a remotely triggerable denial of service. Impact is limited to availability; it is a read from NULL with no out-of-bounds write, memory corruption, or information disclosure. The fix rejects metadata whose inner object array is empty before any dereference, on both layouts.

npm-check-updates through 23.0.2, fixed in commit b554b84, contains a terminal escape sequence injection vulnerability that allows an attacker to embed arbitrary terminal control characters in a dependency's package.json homepage or repository URL fields. When a developer runs ncu with the --format homepage or --format repo option, unfiltered escape sequences are written directly to the terminal, enabling malicious output manipulation or other terminal-dependent effects.

Sucuri Security WordPress plugin through version 2.7.3 contains a path traversal vulnerability in the pageIntegritySubmission() method in src/integrity.lib.php that allows authenticated administrators to delete arbitrary files by supplying directory traversal sequences in the sucuriscan_integrity parameter. Attackers can manipulate the unsanitized file path concatenated with ABSPATH to traverse outside the WordPress installation directory and invoke unlink() on sensitive files such as wp-config.php and .htaccess, causing site outage or enabling malicious reinstallation.

unearth through 0.18.2, fixed in commit 6c78164, contains a path traversal vulnerability in the is_within_directory function that fails to normalize paths before validation, allowing ../ sequences to bypass directory containment checks. Attackers can supply malicious tar archives with symlink members or traversal sequences to write files to arbitrary filesystem locations accessible to the process.

SQLBot through 1.10.0, fixed in commit c3f40a5, contains a stored cross-site scripting vulnerability in the SQText dashboard component that renders TinyMCE output via v-html without sanitization. Attackers who can modify dashboard text widget content can inject arbitrary HTML and JavaScript that executes for all users viewing the dashboard.