Claude Code is an agentic coding tool. Prior to version 2.1.64, Claude Code's sandbox did not prevent sandboxed processes from creating symlinks pointing to locations outside the workspace. When Claude Code subsequently wrote to a path within such a symlink, its unsandboxed process followed the symlink and wrote to the target location outside the workspace without prompting the user for confirmation. This allowed a sandbox escape where neither the sandboxed command nor the unsandboxed app could independently write outside the workspace, but their combination could write to arbitrary locations, potentially leading to code execution outside the sandbox. Reliably exploiting this required the ability to add untrusted content into a Claude Code context window to trigger sandboxed code execution via prompt injection. Users on standard Claude Code auto-update have received this fix automatically. Users performing manual updates are advised to update to version 2.1.64 or later.

Claude Code is an agentic coding tool. In versions below 0.2.111, a path validation flaw using prefix matching instead of canonical path comparison, makes it possible to bypass directory restrictions and access files outside the CWD. Successful exploitation depends on the presence of (or ability to create) a directory with the same prefix as the CWD and the ability to add untrusted content into a Claude Code context window. This is fixed in version 0.2.111.

Claude Code is an agentic coding tool. Prior to Claude Code version 1.0.39, when using Claude Code with Yarn versions 2.0+, Yarn plugins are auto-executed when running yarn --version. This could lead to a bypass of the directory trust dialog in Claude Code, as plugins would be executed prior to the user accepting the risks of working in an untrusted directory. Users running Yarn Classic were unaffected by this issue. This issue has been fixed in version 1.0.39. Users on standard Claude Code auto-update will have received this fix automatically. Users performing manual updates are advised to update to the latest version.

In versions 2.1.63 through 2.1.83 of Claude Code, the folder trust determination logic used the git worktree commondir file without validating its contents. An attacker could craft a malicious repository with a commondir file pointing to a path the victim had previously trusted, causing Claude Code to bypass its trust confirmation dialog and immediately execute hooks defined in `.claude/settings.json`. Exploitation requires the victim to clone the malicious repository and run Claude Code within it, and the attacker must know or guess a path the victim had already trusted. This issue has been fixed in version 2.1.84.

Claude Code is an agentic coding tool. Versions prior to 2.1.53 resolved the permission mode from settings files, including the repo-controlled .claude/settings.json, before determining whether to display the workspace trust confirmation dialog. A malicious repository could set permissions.defaultMode to bypassPermissions in its committed .claude/settings.json, causing the trust dialog to be silently skipped on first open. This allowed a user to be placed into a permissive mode without seeing the trust confirmation prompt, making it easier for an attacker-controlled repository to gain tool execution without explicit user consent. This issue has been patched in version 2.1.53.

Claude Code is an agentic coding tool. Prior to version 2.0.57, Claude Code failed to properly validate directory changes when combined with write operations to protected folders. By using the cd command to navigate into sensitive directories like .claude, it was possible to bypass write protection and create or modify files without user confirmation. Reliably exploiting this required the ability to add untrusted content into a Claude Code context window. This issue has been patched in version 2.0.57.

Claude Code is an agentic coding tool. Prior to version 2.1.2, Claude Code's bubblewrap sandboxing mechanism failed to properly protect the .claude/settings.json configuration file when it did not exist at startup. While the parent directory was mounted as writable and .claude/settings.local.json was explicitly protected with read-only constraints, settings.json was not protected if it was missing. This allowed malicious code running inside the sandbox to create this file and inject persistent hooks (such as SessionStart commands) that would execute with host privileges when Claude Code was restarted. This issue has been patched in version 2.1.2.

Claude Code is an agentic coding tool. Prior to version 2.0.55, Claude Code failed to properly validate commands using piped sed operations with the echo command, allowing attackers to bypass file write restrictions. This vulnerability enabled writing to sensitive directories like the .claude folder and paths outside the project scope. Exploiting this required the ability to execute commands through Claude Code with the "accept edits" feature enabled. This issue has been patched in version 2.0.55.

Claude Code is an agentic coding tool. Prior to version 2.0.72, due to an error in command parsing, it was possible to bypass the Claude Code confirmation prompt to trigger execution of untrusted commands through the find command. Reliably exploiting this required the ability to add untrusted content into a Claude Code context window. This issue has been patched in version 2.0.72.

Claude Code is an agentic coding tool. Prior to version 1.0.39, when running on a machine with Yarn 3.0 or above, Claude Code could have been tricked to execute code contained in a project via yarn plugins before the user accepted the startup trust dialog. Exploiting this would have required a user to start Claude Code in an untrusted directory and to be using Yarn 3.0 or above. This issue has been patched in version 1.0.39.

Claude SDK for TypeScript provides access to the Claude API from server-side TypeScript or JavaScript applications. From version 0.79.0 to before version 0.81.0, the local filesystem memory tool in the Anthropic TypeScript SDK validated model-supplied paths using a string prefix check that did not append a trailing path separator. A model steered by prompt injection could supply a crafted path that resolved to a sibling directory sharing the memory root's name as a prefix, allowing reads and writes outside the sandboxed memory directory. This issue has been patched in version 0.81.0.

in-toto-golang is a go implementation of the in-toto framework to protect software supply chain integrity. In affected versions authenticated attackers posing as functionaries (i.e., within a trusted set of users for a layout) are able to create attestations that may bypass DISALLOW rules in the same layout. An attacker with access to trusted private keys, may issue an attestation that contains a disallowed artifact by including path traversal semantics (e.g., foo vs dir/../foo). Exploiting this vulnerability is dependent on the specific policy applied. The problem has been fixed in version 0.3.0.

Directory traversal vulnerability in Pleasanter (Community Edition and Enterprise Edition) 1.3.39.2 and earlier versions allows a remote authenticated attacker to alter an arbitrary file on the server.

The Betheme theme for WordPress is vulnerable to Arbitrary File Deletion in versions up to, and including, 28.4. This is due to the upload_icons() function workflow using a user-controlled upload path (`mfn-icon-upload`) in a filesystem move operation without constraining it to the uploads directory. This makes it possible for authenticated attackers, with contributor-level access and above, to move/delete arbitrary local files via path traversal.

The Brizy Page Builder plugin <= 2.3.11 for WordPress used an incorrect authorization check that allowed any logged-in user accessing any endpoint in the wp-admin directory to modify the content of any existing post or page created with the Brizy editor. An identical issue was found by another researcher in Brizy <= 1.0.125 and fixed in version 1.0.126, but the vulnerability was reintroduced in version 1.0.127.

Pimcore is an open source data and experience management platform. Versions of Pimcore prior to 10.5.18 are vulnerable to path traversal. The impact of this path traversal and arbitrary extension is limited to creation of arbitrary files and appending data to existing files. When combined with the SQL Injection, the exported data `RESTRICTED DIFFUSION 9 / 9` can be controlled and a webshell can be uploaded. Attackers can use that to execute arbitrary PHP code on the server with the permissions of the webserver. Users may upgrade to version 10.5.18 to receive a patch or, as a workaround, apply the patch manually.

The McFeeder server (distributed as part of SSW package), is susceptible to an arbitrary file write vulnerability on the MAIN computer system. This vulnerability stems from the use of an outdated version of a third-party library, which is used to extract archives uploaded to McFeeder server. An authenticated malicious client can exploit this vulnerability by uploading a crafted ZIP archive via the network to McFeeder’s service endpoint.

A vulnerability in the web-based UI of Cisco Prime Collaboration Assurance could allow an authenticated, remote attacker to overwrite files on the file system. The vulnerability is due to insufficient input validation. An attacker could exploit this vulnerability by using a specific UI input field to provide a custom path location. A successful exploit could allow the attacker to overwrite files on the file system.

The Gluster file system through version 4.1.4 is vulnerable to abuse of the 'features/index' translator. A remote attacker with access to mount volumes could exploit this via the 'GF_XATTROP_ENTRY_IN_KEY' xattrop to create arbitrary, empty files on the target server.

IBM Langflow Desktop 1.2.0 through 1.8.4 Langflow could allow an authenticated attacker to traverse directories on the system. An attacker could send a specially crafted URL request containing "dot dot" sequences (/../) to write arbitrary files on the system.

A Directory Traversal vulnerability in the Unzip feature in Elements-IT HTTP Commander 5.3.3 allows remote authenticated users to write files to arbitrary directories via relative paths in ZIP archives.

A vulnerability in the application data endpoints of Cisco Catalyst SD-WAN Manager, formerly Cisco SD-WAN vManage, could allow an authenticated, remote attacker to write arbitrary files to an affected system. This vulnerability is due to improper validation of requests to APIs. An attacker could exploit this vulnerability by sending malicious requests to an API within the affected system. A successful exploit could allow the attacker to conduct directory traversal attacks and write files to an arbitrary location on the affected system.

In Eclipse Ditto versions 3.0.0 to 3.5.5, the user input of several input fields of the Eclipse Ditto Explorer User Interface https://eclipse.dev/ditto/user-interface.html was not properly neutralized and thus vulnerable to both Reflected and Stored XSS (Cross Site Scripting). Several inputs were not persisted at the backend of Eclipse Ditto, but only in local browser storage to save settings of "environments" of the UI and e.g. the last performed "search queries", resulting in a "Reflected XSS" vulnerability. However, several other inputs were persisted at the backend of Eclipse Ditto, leading to a "Stored XSS" vulnerability. Those mean that authenticated and authorized users at Eclipse Ditto can persist Things in Ditto which can - when being displayed by other users also being authorized to see those Things in the Eclipse Ditto UI - cause scripts to be executed in the browser of other users.

OX App Suite through 7.10.5 allows Directory Traversal via ../ in an OOXML or ODF ZIP archive, because of the mishandling of relative paths in mail addresses in conjunction with auto-configuration DNS records.

The Eleganzo theme for WordPress is vulnerable to arbitrary directory deletion due to insufficient path validation in the akd_required_plugin_callback function in all versions up to, and including, 1.2. This makes it possible for authenticated attackers, with Subscriber-level access and above, to delete arbitrary directories on the server, including the WordPress root directory.

IBM OpenPages with Watson 8.3 and 9.0 IBM OpenPages could allow a remote attacker to traverse directories on the system. An attacker with privileges to perform Import Configuration could send a specially crafted http request containing "dot dot" sequences (/../) in the file name parameter used in Import Configuration to write files to arbitrary locations outside of the specified directory and possibly overwrite arbitrary files.

DIAEnergie Version 1.7.5 and prior is vulnerable to stored cross-site scripting when an unauthenticated user injects arbitrary code into the parameter “descr” of the script “DIAE_hierarchyHandler.ashx”.

Improper limitation of a pathname to a restricted directory ('path traversal') in OneDrive for Android allows an authorized attacker to elevate privileges over a network.

A vulnerability was identified in Scada-LTS up to 2.7.8.1. Affected is the function Common.getHomeDir of the file br/org/scadabr/vo/exporter/ZIPProjectManager.java of the component Project Import. Such manipulation leads to path traversal. The attack may be launched remotely. The exploit is publicly available and might be used. The vendor was contacted early about this disclosure but did not respond in any way.

Arbitrary file write as the OSV-SCALIBR user on the host system via a path traversal vulnerability when using OSV-SCALIBR's unpack() function for container images. Particularly, when using the CLI flag --remote-image on untrusted container images.

E-cology has a directory traversal vulnerability. An attacker can exploit this vulnerability to delete the server directory, causing the server to permanently deny service.

QTIWorks is a software suite for standards-based assessment delivery. Prior to version 1.0-beta15, the QTIWorks Engine allows users to upload QTI content packages as ZIP files. The ZIP handling code does not sufficiently check the paths of files contained within ZIP files, so can insert files into other locations in the filesystem if they are writable by the process running the QTIWorks Engine. In extreme cases, this could allow anonymous users to change files in arbitrary locations in the filesystem. In normal QTIWorks Engine deployments, the impact is somewhat reduced because the default QTIWorks configuration does not enable the public demo functionality, so ZIP files can only be uploaded by users with "instructor" privileges. This vulnerability is fixed in version 1.0-beta15. There are no database configuration changes required when upgrading to this version. No known workarounds for this issue exist.

Multiple relative path traversal vulnerabilities [CWE-23] in Fortinet FortiSOAR before 7.2.1 allows an authenticated attacker to write to the underlying filesystem with nginx permissions via crafted HTTP requests.

NATS nats-server before 2.7.4 allows Directory Traversal (with write access) via an element in a ZIP archive for JetStream streams. nats-streaming-server before 0.24.3 is also affected.

ssr-pages is an HTML page builder for the purpose of server-side rendering (SSR). In versions prior to 0.1.4, a path traversal issue can occur when providing untrusted input to the `svg` property as an argument to the `build(MessagePageOptions)` function. While there is no known workaround at this time, there is a patch in version 0.1.4.

APOC (Awesome Procedures on Cypher) is an add-on library for Neo4j that provides hundreds of procedures and functions. A path traversal vulnerability found in the apoc.export.* procedures of apoc plugins in Neo4j Graph database. The issue allows a malicious actor to potentially break out of the expected directory. The vulnerability is such that files could only be created but not overwritten. For the vulnerability to be exploited, an attacker would need access to execute an arbitrary query, either by having access to an authenticated Neo4j client, or a Cypher injection vulnerability in an application. The minimum versions containing patch for this vulnerability are 4.4.0.12 and 4.3.0.12 and 5.3.1. As a workaround, you can control the allowlist of the procedures that can be used in your system, and/or turn off local file access by setting apoc.export.file.enabled=false.

CoreFTP Server before 727 allows directory traversal (for file creation) by an authenticated attacker via ../ in an HTTP PUT request.

An issue was discovered in Vaultize 21.07.27. When uploading files, there is no check that the filename parameter is correct. As a result, a temporary file will be created outside the specified directory when the file is downloaded. To exploit this, an authenticated user would upload a file with an incorrect file name, and then download it.

Oceanic is a NodeJS library for interfacing with Discord. Prior to version 1.10.4, input to functions such as `Client.rest.channels.removeBan` is not url-encoded, resulting in specially crafted input such as `../../../channels/{id}` being normalized into the url `/api/v10/channels/{id}`, and deleting a channel rather than removing a ban. Version 1.10.4 fixes this issue. Some workarounds are available. One may sanitize user input, ensuring strings are valid for the purpose they are being used for. One may also encode input with `encodeURIComponent` before providing it to the library.

A malicious, but authorised and authenticated user can construct an HTTP request using their existing CSRF token and session cookie to manually upload files to any location that the operating system user account under which pgAdmin is running has permission to write.

An issue in the emqx_sn plugin of EMQX v4.3.8 allows attackers to execute a directory traversal via uploading a crafted .txt file.

A vulnerability in the web-based management interface of Cisco Unified Communications Manager (Unified CM) and Cisco Unified Communications Manager Session Management Edition (Unified CM SME) could allow an authenticated, remote attacker to delete arbitrary files from an affected system. This vulnerability exists because the affected software does not properly validate HTTP requests. An attacker could exploit this vulnerability by sending a crafted HTTP request to the affected software. A successful exploit could allow the attacker to delete arbitrary files from the affected system.

Directory traversal vulnerability in the save_config function in ntpd in ntp_control.c in NTP before 4.2.8p4, when used on systems that do not use '\' or '/' characters for directory separation such as OpenVMS, allows remote authenticated users to overwrite arbitrary files.

A directory traversal issue was discovered in Gradle gradle-enterprise-test-distribution-agent before 1.3.2, test-distribution-gradle-plugin before 1.3.2, and gradle-enterprise-maven-extension before 1.8.2. A malicious actor (with certain credentials) can perform a registration step such that crafted TAR archives lead to extraction of files into arbitrary filesystem locations.

Thruk is a multibackend monitoring webinterface which currently supports Naemon, Icinga, Shinken and Nagios as backends. In versions 3.06 and prior, the file `panorama.pm` is vulnerable to a Path Traversal vulnerability which allows an attacker to upload a file to any folder which has write permissions on the affected system. The parameter location is not filtered, validated or sanitized and it accepts any kind of characters. For a path traversal attack, the only characters required were the dot (`.`) and the slash (`/`). A fix is available in version 3.06.2.

Path Traversal in create template function in EasyUse MailHunter Ultimate 2023 and earlier allow remote authenticated users to extract files into arbitrary directories via a crafted ZIP archive.

Incomplete path traversal fixes in awslabs/tough before tough-v0.22.0 allow remote authenticated users with delegated signing authority to write files outside intended output directories via absolute target names in copy_target/link_target, symlinked parent directories in save_target, or symlinked metadata filenames in SignedRole::write, because write paths trust the joined destination path without post-resolution containment verification. We recommend you upgrade to tough-v0.22.0 / tuftool-v0.15.0.

pyLoad is a free and open-source download manager written in Python. Prior to 0.5.0b3.dev100, package folder names are sanitized using insufficient string replacement. The pattern ....// becomes .._ after replacement (partial removal), leaving .. which can be exploited when the path is later resolved by the OS. This vulnerability is fixed in 0.5.0b3.dev100.

pyLoad is a free and open-source download manager written in Python. Prior to 0.5.0b3.dev100, when passing a folder name in the set_package_data() API function call inside the data object with key "_folder", there is no sanitization at all, allowing a user with Perms.MODIFY to specify arbitrary directories as download locations for a package. This vulnerability is fixed in 0.5.0b3.dev100.

Spring AI's support for Anthropic's Skills API used LLM-influenced filenames unsanitized in Path.resolve before writing files to disk. This could allow a malicious user to write files outside the intended target directory, including restricted directories. Affected versions: Spring AI: 1.1.0 through 1.1.x