CVE-2026-92355
In affected versions of Octopus Server, a user with permission to modify non built-in external feeds could exploit a path traversal flaw to
CVSS
—
No CVSS
EPSS
—
KEV
—
Exploit Today
—
0-100
Published: Sep 16, 2026 · Last modified: Sep 16, 2026 · CWE-22
Not enough EPSS history yet.
In affected versions of Octopus Server, a user with permission to modify non built-in external feeds could exploit a path traversal flaw to overwrite arbitrary files on the server, which in some configurations could lead to remote code execution.
CVECVSSEPSSKEVRExploitTitleMod.
CVE-2026-926048.1 HIG—
———Scirius through 3.8.0 contains an arbitrary file write vulnerability in the PCAP filestore upload endpoint that allows default User role users to write attacker-controlled JSON content to filesystem paths. Attackers can supply path traversal sequences in the uploaded document's _id field to escape the intended directory and write files with .json extension to arbitrary locations as root.2hCVE-2026-857318.8 HIG—
———oras-go is a Go library for managing OCI artifacts. Prior to 2.6.2, content/file.Store extraction of OCI layers marked with io.deis.oras.content.unpack=true can write outside the store working directory. The pushDir path through extractTarDirectory and ensureLinkPath validates symlink targets lexically, resolveRelToBase skips its parent-symlink walk for root-level entries, and writeFile follows a terminal symlink when opening a regular file. A malicious archive can therefore create a symlink chain whose lexical target remains inside the extraction root but whose resolved target is an attacker-selected absolute path, then overwrite that target with a same-named regular-file entry even when AllowPathTraversalOnWrite is false. Pulling an attacker-controlled artifact can create or overwrite any file writable by the process and may lead to code execution. This issue is fixed in version 2.6.2.2hCVE-2026-599747.8 HIG—
———Stanza is a Stanford NLP Python library for tokenization, sentence segmentation, NER, and parsing of many human languages. Prior to 1.14.0, stanza.resources.common.unzip in stanza/resources/common.py passes downloaded model and resource archives to zipfile.ZipFile.extractall without validating member paths, and the vulnerable extraction path is reachable through stanza.download and stanza.install_corenlp. A malicious archive containing parent-directory traversal entries can write outside the intended model directory, allowing files writable by the Stanza process to be overwritten and potentially enabling code execution through modified shell configuration, SSH authorization data, Python packages, or executable scripts. This issue is fixed in version 1.14.0.2hCVE-2026-599446.1 MED—
———Composer is a dependency Manager for the PHP language. From 1.0 until 2.2.30 and from 2.3.0 until 2.10.3, a malicious or compromised dependency can bypass the earlier CVE-2026-59946 binary-path hardening because Composer validates literal parent-directory segments only during dependency resolution, while the symlink and installed-metadata paths described by the advisory skip that validation. A package can ship an in-package binary symlink that resolves outside its installation directory, or attacker-influenced vendor/composer/installed.json metadata can provide an escaping binary path during a reinstall or regeneration of missing vendor/bin entries. The installed-metadata path is reachable only when the vendor directory was not populated by the same validated install run, such as when it is restored from an untrusted cache, copied from an earlier build stage, carried over from an older Composer run, or writable by a lower-trust build step. Composer can follow the path, change the external target's permissions to make it world-readable and executable, and create a runnable vendor/bin proxy to that external file. The issue does not directly read or transmit data and does not by itself provide remote code execution. This issue is fixed in versions 2.2.30 and 2.10.3.2hCVE-2026-615609.8 CRI—
———`@zereight/mcp-gitlab` is a Model Context Protocol server for GitLab. Prior to version 2.1.27, the SSE transport mode (`SSE=true`) exposes all MCP tools without any authentication. The `upload_markdown` tool reads arbitrary files from the server's local filesystem via an unsanitized `file_path` parameter and uploads them to a GitLab project. Combined, any unauthenticated network-reachable attacker can read `/proc/self/environ` to steal the server's `GITLAB_PERSONAL_ACCESS_TOKEN` and achieve full GitLab account takeover. This is the default configuration for Docker deployments. Version 2.1.27 contains a patch.6hCVE-2026-890409.8 CRI—
———Tencent Mass Service Engine in Cluster (MSEC) allows a remote, unauthenticated attacker to send a crafted POST request including ../ and gain root access on the target device. An attacker who uploads a webshell can execute arbitrary code as root.1d