CVEs

In the Linux kernel, the following vulnerability has been resolved: tcp: secure_seq: add back ports to TS offset This reverts 28ee1b746f49 ("secure_seq: downgrade to per-host timestamp offsets") tcp_tw_recycle went away in 2017. Zhouyan Deng reported off-path TCP source port leakage via SYN cookie side-channel that can be fixed in multiple ways. One of them is to bring back TCP ports in TS offset randomization. As a bonus, we perform a single siphash() computation to provide both an ISN and a TS offset.

In the Linux kernel, the following vulnerability has been resolved: wifi: mac80211: bounds-check link_id in ieee80211_ml_reconfiguration link_id is taken from the ML Reconfiguration element (control & 0x000f), so it can be 0..15. link_removal_timeout[] has IEEE80211_MLD_MAX_NUM_LINKS (15) elements, so index 15 is out-of-bounds. Skip subelements with link_id >= IEEE80211_MLD_MAX_NUM_LINKS to avoid a stack out-of-bounds write.

In the Linux kernel, the following vulnerability has been resolved: net/sched: act_gate: snapshot parameters with RCU on replace The gate action can be replaced while the hrtimer callback or dump path is walking the schedule list. Convert the parameters to an RCU-protected snapshot and swap updates under tcf_lock, freeing the previous snapshot via call_rcu(). When REPLACE omits the entry list, preserve the existing schedule so the effective state is unchanged.

In the Linux kernel, the following vulnerability has been resolved: nvme: fix memory allocation in nvme_pr_read_keys() nvme_pr_read_keys() takes num_keys from userspace and uses it to calculate the allocation size for rse via struct_size(). The upper limit is PR_KEYS_MAX (64K). A malicious or buggy userspace can pass a large num_keys value that results in a 4MB allocation attempt at most, causing a warning in the page allocator when the order exceeds MAX_PAGE_ORDER. To fix this, use kvzalloc() instead of kzalloc(). This bug has the same reasoning and fix with the patch below: https://lore.kernel.org/linux-block/20251212013510.3576091-1-kartikey406@gmail.com/ Warning log: WARNING: mm/page_alloc.c:5216 at __alloc_frozen_pages_noprof+0x5aa/0x2300 mm/page_alloc.c:5216, CPU#1: syz-executor117/272 Modules linked in: CPU: 1 UID: 0 PID: 272 Comm: syz-executor117 Not tainted 6.19.0 #1 PREEMPT(voluntary) Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014 RIP: 0010:__alloc_frozen_pages_noprof+0x5aa/0x2300 mm/page_alloc.c:5216 Code: ff 83 bd a8 fe ff ff 0a 0f 86 69 fb ff ff 0f b6 1d f9 f9 c4 04 80 fb 01 0f 87 3b 76 30 ff 83 e3 01 75 09 c6 05 e4 f9 c4 04 01 <0f> 0b 48 c7 85 70 fe ff ff 00 00 00 00 e9 8f fd ff ff 31 c0 e9 0d RSP: 0018:ffffc90000fcf450 EFLAGS: 00010246 RAX: 0000000000000000 RBX: 0000000000000000 RCX: 1ffff920001f9ea0 RDX: 0000000000000000 RSI: 000000000000000b RDI: 0000000000040dc0 RBP: ffffc90000fcf648 R08: ffff88800b6c3380 R09: 0000000000000001 R10: ffffc90000fcf840 R11: ffff88807ffad280 R12: 0000000000000000 R13: 0000000000040dc0 R14: 0000000000000001 R15: ffffc90000fcf620 FS: 0000555565db33c0(0000) GS:ffff8880be26c000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 000000002000000c CR3: 0000000003b72000 CR4: 00000000000006f0 Call Trace: <TASK> alloc_pages_mpol+0x236/0x4d0 mm/mempolicy.c:2486 alloc_frozen_pages_noprof+0x149/0x180 mm/mempolicy.c:2557 _

In the Linux kernel, the following vulnerability has been resolved: RDMA/umad: Reject negative data_len in ib_umad_write ib_umad_write computes data_len from user-controlled count and the MAD header sizes. With a mismatched user MAD header size and RMPP header length, data_len can become negative and reach ib_create_send_mad(). This can make the padding calculation exceed the segment size and trigger an out-of-bounds memset in alloc_send_rmpp_list(). Add an explicit check to reject negative data_len before creating the send buffer. KASAN splat: [ 211.363464] BUG: KASAN: slab-out-of-bounds in ib_create_send_mad+0xa01/0x11b0 [ 211.364077] Write of size 220 at addr ffff88800c3fa1f8 by task spray_thread/102 [ 211.365867] ib_create_send_mad+0xa01/0x11b0 [ 211.365887] ib_umad_write+0x853/0x1c80

In the Linux kernel, the following vulnerability has been resolved: RDMA/siw: Fix potential NULL pointer dereference in header processing If siw_get_hdr() returns -EINVAL before set_rx_fpdu_context(), qp->rx_fpdu can be NULL. The error path in siw_tcp_rx_data() dereferences qp->rx_fpdu->more_ddp_segs without checking, which may lead to a NULL pointer deref. Only check more_ddp_segs when rx_fpdu is present. KASAN splat: [ 101.384271] KASAN: null-ptr-deref in range [0x00000000000000c0-0x00000000000000c7] [ 101.385869] RIP: 0010:siw_tcp_rx_data+0x13ad/0x1e50

In the Linux kernel, the following vulnerability has been resolved: fs: ntfs3: fix infinite loop triggered by zero-sized ATTR_LIST We found an infinite loop bug in the ntfs3 file system that can lead to a Denial-of-Service (DoS) condition. A malformed NTFS image can cause an infinite loop when an ATTR_LIST attribute indicates a zero data size while the driver allocates memory for it. When ntfs_load_attr_list() processes a resident ATTR_LIST with data_size set to zero, it still allocates memory because of al_aligned(0). This creates an inconsistent state where ni->attr_list.size is zero, but ni->attr_list.le is non-null. This causes ni_enum_attr_ex to incorrectly assume that no attribute list exists and enumerates only the primary MFT record. When it finds ATTR_LIST, the code reloads it and restarts the enumeration, repeating indefinitely. The mount operation never completes, hanging the kernel thread. This patch adds validation to ensure that data_size is non-zero before memory allocation. When a zero-sized ATTR_LIST is detected, the function returns -EINVAL, preventing a DoS vulnerability.

In the Linux kernel, the following vulnerability has been resolved: fs: ntfs3: check return value of indx_find to avoid infinite loop We found an infinite loop bug in the ntfs3 file system that can lead to a Denial-of-Service (DoS) condition. A malformed dentry in the ntfs3 filesystem can cause the kernel to hang during the lookup operations. By setting the HAS_SUB_NODE flag in an INDEX_ENTRY within a directory's INDEX_ALLOCATION block and manipulating the VCN pointer, an attacker can cause the indx_find() function to repeatedly read the same block, allocating 4 KB of memory each time. The kernel lacks VCN loop detection and depth limits, causing memory exhaustion and an OOM crash. This patch adds a return value check for fnd_push() to prevent a memory exhaustion vulnerability caused by infinite loops. When the index exceeds the size of the fnd->nodes array, fnd_push() returns -EINVAL. The indx_find() function checks this return value and stops processing, preventing further memory allocation.

In the Linux kernel, the following vulnerability has been resolved: fs: ntfs3: fix infinite loop in attr_load_runs_range on inconsistent metadata We found an infinite loop bug in the ntfs3 file system that can lead to a Denial-of-Service (DoS) condition. A malformed NTFS image can cause an infinite loop when an attribute header indicates an empty run list, while directory entries reference it as containing actual data. In NTFS, setting evcn=-1 with svcn=0 is a valid way to represent an empty run list, and run_unpack() correctly handles this by checking if evcn + 1 equals svcn and returning early without parsing any run data. However, this creates a problem when there is metadata inconsistency, where the attribute header claims to be empty (evcn=-1) but the caller expects to read actual data. When run_unpack() immediately returns success upon seeing this condition, it leaves the runs_tree uninitialized with run->runs as a NULL. The calling function attr_load_runs_range() assumes that a successful return means that the runs were loaded and sets clen to 0, expecting the next run_lookup_entry() call to succeed. Because runs_tree remains uninitialized, run_lookup_entry() continues to fail, and the loop increments vcn by zero (vcn += 0), leading to an infinite loop. This patch adds a retry counter to detect when run_lookup_entry() fails consecutively after attr_load_runs_vcn(). If the run is still not found on the second attempt, it indicates corrupted metadata and returns -EINVAL, preventing the Denial-of-Service (DoS) vulnerability.

Missing Authorization vulnerability in Ajay Contextual Related Posts contextual-related-posts allows Exploiting Incorrectly Configured Access Control Security Levels.This issue affects Contextual Related Posts: from n/a through < 4.2.2.

The Yoast Duplicate Post plugin for WordPress is vulnerable to unauthorized modification of data due to a missing capability check on the clone_bulk_action_handler() and republish_request() functions in all versions up to, and including, 4.5. This makes it possible for authenticated attackers, with Contributor-level access and above, to duplicate any post on the site including private, draft, and trashed posts they shouldn't have access to. Additionally, attackers with Author-level access and above can use the Rewrite & Republish feature to overwrite any published post with their own content.

A JSONPath injection vulnerability in Spring AI's AbstractFilterExpressionConverter allows authenticated users to bypass metadata-based access controls through crafted filter expressions. User-controlled input passed to FilterExpressionBuilder is concatenated into JSONPath queries without proper escaping, enabling attackers to inject arbitrary JSONPath logic and access unauthorized documents. This vulnerability affects applications using vector stores that extend AbstractFilterExpressionConverter for multi-tenant isolation, role-based access control, or document filtering based on metadata. The vulnerability occurs when user-supplied values in filter expressions are not escaped before being inserted into JSONPath queries. Special characters like ", ||, and && are passed through unescaped, allowing injection of arbitrary JSONPath logic that can alter the intended query semantics.

A critical SQL injection vulnerability in Spring AI's MariaDBFilterExpressionConverter allows attackers to bypass metadata-based access controls and execute arbitrary SQL commands. The vulnerability exists due to missing input sanitization.

A CSRF vulnerability in the Link Aggregation configuration interface allows an unauthenticated remote attacker to trick authenticated users into sending unauthorized POST requests to the device by luring them to a malicious webpage. This can silently alter the device’s configuration without the victim’s knowledge or consent. Availability impact was set to low because after a successful attack the device will automatically recover without external intervention.

A stored cross‑site scripting (XSS) vulnerability in the Link Aggregation configuration interface allows an unauthenticated remote attacker to create a trunk entry containing malicious HTML/JavaScript code. When the affected page is viewed, the injected script executes in the context of the victim’s browser, enabling unauthorized actions such as interface manipulation. The session cookie is secured by the httpOnly Flag. Therefore an attacker is not able to take over the session of an authenticated user.

A stack-based buffer overflow in the device's Telnet/SSH CLI login routine occurs when a unauthenticated attacker send an oversized or unexpected username input. An overflow condition crashes the thread handling the login attempt, forcing the session to close. Because other CLI sessions remain unaffected, the impact is limited to a low‑severity availability disruption.

A stack-based buffer overflow in the CLI's TFTP file‑transfer command handling allows a low-privileged attacker with Telnet/SSH access to trigger memory corruption by supplying unexpected or oversized filename input. Exploitation results in the corruption of the internal buffer, causing the CLI and web dashboard to become unavailable and leading to a denial of service.

A stack-based buffer overflow in the device's file installation workflow allows a high-privileged attacker to send oversized POST parameters that overflow a fixed-size stack buffer within an internal process, resulting in a DoS attack.

A stack-based buffer overflow vulnerability in the device's file transfer parameter workflow allows a high-privileged attacker to send oversized POST parameters, causing memory corruption in an internal process, resulting in a DoS attack.

A command injection vulnerability in the device’s Root CA certificate transfer workflow allows a high-privileged attacker to send crafted HTTP POST requests that result in arbitrary command execution on the underlying Linux OS with root privileges.

A remote attacker with user privileges for the webUI can use the setting of the TFTP Filename with a POST Request to trigger a stack-based Buffer Overflow, resulting in a DoS attack.

A vulnerability found in Dahua NVR/XVR device. A third-party malicious attacker with physical access to the device may gain access to a restricted shell via the serial port, and bypasses the shell's authentication mechanism to escalate privileges.

The Writeprint Stylometry plugin for WordPress is vulnerable to Reflected Cross-Site Scripting via the 'p' GET parameter in all versions up to and including 0.1. This is due to insufficient input sanitization and output escaping in the bjl_wprintstylo_comments_nav() function. The function directly outputs the $_GET['p'] parameter into an HTML href attribute without any escaping. This makes it possible for authenticated attackers with Contributor-level permissions or higher to inject arbitrary web scripts in pages that execute if they can successfully trick another user into performing an action such as clicking on a link.

Glances is an open-source system cross-platform monitoring tool. The Glances action system allows administrators to configure shell commands that execute when monitoring thresholds are exceeded. These commands support Mustache template variables (e.g., `{{name}}`, `{{key}}`) that are populated with runtime monitoring data. The `secure_popen()` function, which executes these commands, implements its own pipe, redirect, and chain operator handling by splitting the command string before passing each segment to `subprocess.Popen(shell=False)`. Prior to 4.5.2, when a Mustache-rendered value (such as a process name, filesystem mount point, or container name) contains pipe, redirect, or chain metacharacters, the rendered command is split in unintended ways, allowing an attacker who controls a process name or container name to inject arbitrary commands. Version 4.5.2 fixes the issue.

The Get Use APIs WordPress plugin before 2.0.10 executes imported JSON, which could allow users with a role as low as contributor to perform Cross-Site Scripting attacks under certain server configurations.

Glances is an open-source system cross-platform monitoring tool. Prior to 4.5.2, Glances web server runs without authentication by default when started with `glances -w`, exposing REST API with sensitive system information including process command-lines containing credentials (passwords, API keys, tokens) to any network client. Version 4.5.2 fixes the issue.

IncusOS is an immutable OS image dedicated to running Incus. Prior to 202603142010, the default configuration of systemd-cryptenroll as used by IncusOS through mkosi allows for an attacker with physical access to the machine to access the encrypted data without requiring any interaction by the system's owner or any tampering of Secure Boot state or kernel (UKI) boot image. That's because in this configuration, the LUKS key is made available by the TPM so long as the system has the expected PCR7 value and the PCR11 policy matches. That default PCR11 policy importantly allows for the TPM to release the key to the booted system rather than just from the initrd part of the signed kernel image (UKI). The attack relies on the attacker being able to substitute the original encrypted root partition for one that they control. By doing so, the system will prompt for a recovery key on boot, which the attacker has defined and can provide, before booting the system using the attacker's root partition rather than the system's original one. The attacker only needs to put a systemd unit starting on system boot within their root partition to have the system run that logic on boot. That unit will then run in an environment where the TPM will allow for the retrieval of the encryption key of the real root disk, allowing the attacker to steal the LUKS volume key (immutable master key) and then use it against the real root disk, altering it or getting data out before putting the disk back the way it was and returning the system without a trace of this attack having happened. This is all possible because the system will have still booted with Secure Boot enabled, will have measured and ran the expected bootloader and kernel image (UKI). The initrd selects the root disk based on GPT partition identifiers making it possible to easily substitute the real root disk for an attacker controlled one. This doesn't lead to any change in the TPM state and therefore allows for retrieval of the LUKS k

The Azure Blob Storage for Craft CMS plugin provides an Azure Blob Storage integration for Craft CMS. In versions on the 2.x branch prior to 2.1.1, unauthenticated users can view a list of buckets the plugin has access to. The `DefaultController->actionLoadContainerData()` endpoint allows unauthenticated users with a valid CSRF token to view a list of buckets that the plugin is allowed to see. Because Azure can return sensitive data in error messages, additional attack vectors are also exposed. Users should update to version 2.1.1 of the plugin to mitigate the issue.

A flaw was identified in Keycloak, an identity and access management solution, where it improperly follows HTTP redirects when processing certain client configuration requests. This behavior allows an attacker to trick the server into making unintended requests to internal or restricted resources. As a result, sensitive internal services such as cloud metadata endpoints could be accessed. This issue may lead to information disclosure and enable attackers to map internal network infrastructure.

The Google Cloud Storage for Craft CMS plugin provides a Google Cloud Storage integration for Craft CMS. In versions on the 2.x branch prior to 2.2.1, the `DefaultController->actionLoadBucketData()` endpoint allows unauthenticated users with a valid CSRF token to view a list of buckets that the plugin is allowed to see. Users should update to version 2.2.1 of the plugin to mitigate the issue.

The Subscriptions for WooCommerce plugin for WordPress is vulnerable to unauthorized modification of data due to a missing capability check on the `wps_sfw_admin_cancel_susbcription()` function in all versions up to, and including, 1.9.2. This is due to the function being hooked to the `init` action without any authentication or authorization checks, and only performing a non-empty check on the nonce parameter without actually validating it via `wp_verify_nonce()`. This makes it possible for unauthenticated attackers to cancel any active WooCommerce subscription by sending a crafted GET request with an arbitrary nonce value via the `wps_subscription_id` parameter.

The [CR]Paid Link Manager plugin for WordPress is vulnerable to Reflected Cross-Site Scripting via the URL path in all versions up to, and including, 0.5 due to insufficient input sanitization and output escaping. This makes it possible for unauthenticated attackers to inject arbitrary web scripts in pages that execute if they can successfully trick a user into performing an action such as clicking on a link.

The Amazon S3 for Craft CMS plugin provides an Amazon S3 integration for Craft CMS. In versions 2.0.2 through 2.2.4, unauthenticated users can view a list of buckets the plugin has access to. The `BucketsController->actionLoadBucketData()` endpoint allows unauthenticated users with a valid CSRF token to view a list of buckets that the plugin is allowed to see. Users should update to version 2.2.5 of the plugin to mitigate the issue.

music-metadata is a metadata parser for audio and video media files. Prior to version 11.12.3, music-metadata's ASF parser (`parseExtensionObject()` in `lib/asf/AsfParser.ts:112-158`) enters an infinite loop when a sub-object inside the ASF Header Extension Object has `objectSize = 0`. Version 11.12.3 fixes the issue.

A flaw was found in Keycloak. An unauthenticated remote attacker can trigger an application level Denial of Service (DoS) by sending a highly compressed SAMLRequest through the SAML Redirect Binding. The server fails to enforce size limits during DEFLATE decompression, leading to an OutOfMemoryError (OOM) and subsequent process termination. This vulnerability allows an attacker to disrupt the availability of the service.

Kube-router is a turnkey solution for Kubernetes networking. Prior to version 2.8.0, Kube-router's proxy module does not validate externalIPs or loadBalancer IPs before programming them into the node's network configuration. Version 2.8.0 contains a patch for the issue. Available workarounds include enabling DenyServiceExternalIPs feature gate, deploying admission policy, restricting service creation RBAC, monitoring service changes, and applying BGP prefix filtering.

jsPDF is a library to generate PDFs in JavaScript. Prior to version 4.2.1, user control of the `options` argument of the `output` function allows attackers to inject arbitrary HTML (such as scripts) into the browser context the created PDF is opened in. The vulnerability can be exploited in the following scenario: the attacker provides values for the output options, for example via a web interface. These values are then passed unsanitized (automatically or semi-automatically) to the attack victim. The victim creates and opens a PDF with the attack vector using one of the vulnerable method overloads inside their browser. The attacker can thus inject scripts that run in the victims browser context and can extract or modify secrets from this context. The vulnerability has been fixed in jspdf@4.2.1. As a workaround, sanitize user input before passing it to the output method.

jsPDF is a library to generate PDFs in JavaScript. Prior to version 4.2.1, user control of arguments of the `createAnnotation` method allows users to inject arbitrary PDF objects, such as JavaScript actions. If given the possibility to pass unsanitized input to the following method, a user can inject arbitrary PDF objects, such as JavaScript actions, which might trigger when the PDF is opened or interacted with the `createAnnotation`: `color` parameter. The vulnerability has been fixed in jsPDF@4.2.1. As a workaround, sanitize user input before passing it to the vulnerable API members.

Cockpit is a headless content management system. Any Cockpit CMS instance running version 2.13.4 or earlier with API access enabled is potentially affected by a a SQL Injection vulnerability in the MongoLite Aggregation Optimizer. Any deployment where the `/api/content/aggregate/{model}` endpoint is publicly accessible or reachable by untrusted users may be vulnerable, and attackers in possession of a valid read-only API key (the lowest privilege level) can exploit this vulnerability — no admin access is required. An attacker can inject arbitrary SQL via unsanitized field names in aggregation queries, bypass the `_state=1` published-content filter to access unpublished or restricted content, and extract unauthorized data from the underlying SQLite content database. This vulnerability has been patched in version 2.13.5. The fix applies the same field-name sanitization introduced in v2.13.3 for `toJsonPath()` to the `toJsonExtractRaw()` method in `lib/MongoLite/Aggregation/Optimizer.php`, closing the injection vector in the Aggregation Optimizer.

Elysia is a Typescript framework for request validation, type inference, OpenAPI documentation, and client-server communication. Prior to version 1.4.27, an Elysia cookie can be overridden by prototype pollution , eg. `__proto__`. This issue is patched in 1.4.27. As a workaround, use t.Cookie validation to enforce validation value and/or prevent iterable over cookie if possible.

pyasn1 is a generic ASN.1 library for Python. Prior to 0.6.3, the `pyasn1` library is vulnerable to a Denial of Service (DoS) attack caused by uncontrolled recursion when decoding ASN.1 data with deeply nested structures. An attacker can supply a crafted payload containing thousands of nested `SEQUENCE` (`0x30`) or `SET` (`0x31`) tags with "Indefinite Length" (`0x80`) markers. This forces the decoder to recursively call itself until the Python interpreter crashes with a `RecursionError` or consumes all available memory (OOM), crashing the host application. This is a distinct vulnerability from CVE-2026-23490 (which addressed integer overflows in OID decoding). The fix for CVE-2026-23490 (`MAX_OID_ARC_CONTINUATION_OCTETS`) does not mitigate this recursion issue. Version 0.6.3 fixes this specific issue.

mdjnelson/moodle-mod_customcert is a Moodle plugin for creating dynamically generated certificates with complete customization via the web browser. Prior to versions 4.4.9 and 5.0.3, a teacher who holds `mod/customcert:manage` in any single course can read and silently overwrite certificate elements belonging to any other course in the Moodle installation. The `core_get_fragment` callback `editelement` and the `mod_customcert_save_element` web service both fail to verify that the supplied `elementid` belongs to the authorized context, enabling cross-course information disclosure and data tampering. Versions 4.4.9 and 5.0.3 fix the issue.

DiceBear is an avatar library for designers and developers. Prior to version 9.4.0, the `ensureSize()` function in `@dicebear/converter` read the `width` and `height` attributes from the input SVG to determine the output canvas size for rasterization (PNG, JPEG, WebP, AVIF). An attacker who can supply a crafted SVG with extremely large dimensions (e.g. `width="999999999"`) could force the server to allocate excessive memory, leading to denial of service. This primarily affects server-side applications that pass untrusted or user-supplied SVGs to the converter's `toPng()`, `toJpeg()`, `toWebp()`, or `toAvif()` functions. Applications that only convert self-generated DiceBear avatars are not practically exploitable, but are still recommended to upgrade. This is fixed in version 9.4.0. The `ensureSize()` function no longer reads SVG attributes to determine output size. Instead, a new `size` option (default: 512, max: 2048) controls the output dimensions. Invalid values (NaN, negative, zero, Infinity) fall back to the default. If upgrading is not immediately possible, validate and sanitize the `width` and `height` attributes of any untrusted SVG input before passing it to the converter.

Kanboard is project management software focused on Kanban methodology. Versions prior to 1.2.51 have an authenticated SQL injection vulnerability. Attackers with the permission to add users to a project can leverage this vulnerability to dump the entirety of the kanboard database. Version 1.2.51 fixes the issue.

Kanboard is project management software focused on Kanban methodology. Prior to 1.2.51, Kanboard's user invite registration endpoint (`UserInviteController::register()`) accepts all POST parameters and passes them to `UserModel::create()` without filtering out the `role` field. An attacker who receives an invite link can inject `role=app-admin` in the registration form to create an administrator account. Version 1.2.51 fixes the issue.

OpenClaw versions prior to 2026.2.26 contain an approval bypass vulnerability in system.run execution that allows attackers to execute commands from unintended filesystem locations by rebinding writable parent symlinks in the current working directory after approval. An attacker can modify mutable parent symlink path components between approval and execution time to redirect command execution to a different location while preserving the visible working directory string.

OpenClaw versions prior to 2026.2.21 accept prototype-reserved keys in runtime /debug set override object values, allowing prototype pollution attacks. Authorized /debug set callers can inject __proto__, constructor, or prototype keys to manipulate object prototypes and bypass command gate restrictions.

OpenClaw versions prior to 2026.2.24 contain a sandbox bind validation vulnerability allowing attackers to bypass allowed-root and blocked-path checks via symlinked parent directories with non-existent leaf paths. Attackers can craft bind source paths that appear within allowed roots but resolve outside sandbox boundaries once missing leaf components are created, weakening bind-source isolation enforcement.

OpenClaw versions prior to 2026.2.24 contain a local media root bypass vulnerability in sendAttachment and setGroupIcon message actions when sandboxRoot is unset. Attackers can hydrate media from local absolute paths to read arbitrary host files accessible by the runtime user.

OpenClaw version 2026.2.22 prior to 2026.2.23 contain an arbitrary code execution vulnerability in shell-env that allows attackers to execute attacker-controlled binaries by exploiting trusted-prefix fallback logic for the $SHELL variable. An attacker can influence the $SHELL environment variable on systems with writable trusted-prefix directories such as /opt/homebrew/bin to execute arbitrary binaries in the OpenClaw process context.