In Eclipse Theia versions prior to 1.71.0, the AI chat agent processed workspace file and directory names as part of its prompt context without distinguishing them from system instructions. An attacker could craft a malicious repository with adversarial directory or file names that, when analyzed by the AI agent, would cause the agent to follow attacker-controlled instructions (indirect prompt injection). Combined with other AI chat features available in untrusted workspaces, this enabled attack chains leading to data exfiltration via Markdown image rendering or arbitrary command execution via task definitions.

In Eclipse Theia versions prior to 1.69.0, custom task definitions in workspace files (e.g. .theia/tasks.json, .vscode/tasks.json) could be executed without requiring workspace trust. An attacker could craft a malicious repository that, when cloned and opened in Theia, leads to execution of arbitrary commands with the user's privileges. In combination with AI chat features and a workspace .theia/settings.json that disabled tool confirmation, this could be triggered automatically by sending a message in the AI chat.

For Eclipse Che versions 6.16 to 7.3.0, with both authentication and TLS disabled, visiting a malicious web site could trigger the start of an arbitrary Che workspace. Che with no authentication and no TLS is not usually deployed on a public network but is often used for local installations (e.g. on personal laptops). In that case, even if the Che API is not exposed externally, some javascript running in the local browser is able to send requests to it.

In affected versions of Eclipse Theia (1.8.1 and later), the browser backend exposes privileged terminal RPC over WebSocket (/services/shell-terminal, /services/terminals/:id) without service-level authentication. WebSocket origin validation in @theia/core is fail-open: connections are accepted when the Origin header is missing or when no THEIA_HOSTS allowlist is configured (the default). The Socket.IO integration additionally replaces the real Origin header with a client-supplied fix-origin header that an attacker can control or omit. As a result, a foreign-origin web page visited by a user with a running Theia instance can open the /services WebSocket namespace, invoke terminal creation, attach to the resulting terminal data channel, execute arbitrary OS commands, and read their output. This affects both local developer setups (drive-by attack) and hosted or tunneled deployments without strong external authentication. A fix is in development that enforces same-origin validation by default, removes trust in the fix-origin header, gates HTTP and WebSocket access on a SameSite=Strict; HttpOnly connection-token cookie, and sanitizes shell terminal creation options.

Vert.x-Web framework v4.0 milestone 1-4 does not perform a correct CSRF verification. Instead of comparing the CSRF token in the request with the CSRF token in the cookie, it compares the CSRF token in the cookie against a CSRF token that is stored in the session. An attacker does not even need to provide a CSRF token in the request because the framework does not consider it. The cookies are automatically sent by the browser and the verification will always succeed, leading to a successful CSRF attack.

In Eclipse Theia 0.3.9 to 1.8.1, the "mini-browser" extension allows a user to preview HTML files in an iframe inside the IDE. But with the way it is made it is possible for a previewed HTML file to trigger an RCE. This exploit only happens if a user previews a malicious file..

Arbitrary File Overwrite in Eclipse JGit <= 6.6.0 In Eclipse JGit, all versions <= 6.6.0.202305301015-r, a symbolic link present in a specially crafted git repository can be used to write a file to locations outside the working tree when this repository is cloned with JGit to a case-insensitive filesystem, or when a checkout from a clone of such a repository is performed on a case-insensitive filesystem. This can happen on checkout (DirCacheCheckout), merge (ResolveMerger via its WorkingTreeUpdater), pull (PullCommand using merge), and when applying a patch (PatchApplier). This can be exploited for remote code execution (RCE), for instance if the file written outside the working tree is a git filter that gets executed on a subsequent git command. The issue occurs only on case-insensitive filesystems, like the default filesystems on Windows and macOS. The user performing the clone or checkout must have the rights to create symbolic links for the problem to occur, and symbolic links must be enabled in the git configuration. Setting git configuration option core.symlinks = false before checking out avoids the problem. The issue was fixed in Eclipse JGit version 6.6.1.202309021850-r and 6.7.0.202309050840-r, available via Maven Central https://repo1.maven.org/maven2/org/eclipse/jgit/  and repo.eclipse.org https://repo.eclipse.org/content/repositories/jgit-releases/ . A backport is available in 5.13.3 starting from 5.13.3.202401111512-r. The JGit maintainers would like to thank RyotaK for finding and reporting this issue.

In Eclipse p2, installable units are able to alter the Eclipse Platform installation and the local machine via touchpoints during installation. Those touchpoints can, for example, alter the command-line used to start the application, injecting things like agent or other settings that usually require particular attention in term of security. Although p2 has built-in strategies to ensure artifacts are signed and then to help establish trust, there is no such strategy for the metadata part that does configure such touchpoints. As a result, it's possible to install a unit that will run malicious code during installation without user receiving any warning about this installation step being risky when coming from untrusted source.

In Eclipse Theia versions prior to 1.71.0, the AI chat rendered Markdown image tags from AI responses, triggering HTTP requests to arbitrary external URLs without restriction. Combined with prompt injection in a malicious workspace, an attacker could induce the AI agent to construct image URLs encoding sensitive information from the workspace or conversation context, exfiltrating it to attacker-controlled servers. The workspace trust enforcement introduced in v1.71.0 mitigates the documented attack chain by disabling AI features in untrusted workspaces.

In the Eclipse Theia Website repository, the GitHub Actions workflow .github/workflows/preview.yml used pull_request_target trigger while checking out and executing untrusted pull request code. This allowed any GitHub user to execute arbitrary code in the repository's CI environment with access to repository secrets and a GITHUB_TOKEN with extensive write permissions (contents:write, packages:write, pages:write, actions:write). An attacker could exfiltrate secrets, publish malicious packages to the eclipse-theia organization, modify the official Theia website, and push malicious code to the repository.

In Eclipse Buildship versions prior to 3.1.1, the build files indicate that this project is resolving dependencies over HTTP instead of HTTPS. Any of these artifacts could have been MITM to maliciously compromise them and infect the build artifacts that were produced. Additionally, if any of these JARs or other dependencies were compromised, any developers using these could continue to be infected past updating to fix this.

All Xtext & Xtend versions prior to 2.18.0 were built using HTTP instead of HTTPS file transfer and thus the built artifacts may have been compromised.

Eclipse hawkBit versions prior to 0.3.0M2 resolved Maven build artifacts for the Vaadin based UI over HTTP instead of HTTPS. Any of these dependent artifacts could have been maliciously compromised by a MITM attack. Hence produced build artifacts of hawkBit might be infected.

Eclipse Vorto versions prior to 0.11 resolved Maven build artifacts for the Xtext project over HTTP instead of HTTPS. Any of these dependent artifacts could have been maliciously compromised by a MITM attack. Hence produced build artifacts of Vorto might be infected.

In Eclipse Theia versions up to and including 0.16.0, in the notification messages there is no HTML escaping, so Javascript code can run.

Concrete CMS 9.5.0 and below does not validate a CSRF token before processing requests to /dashboard/extend/update/prepare_remote_upgrade/<remoteMPID>. An attacker who controls the remote package returned for a known marketplace item ID can overwrite the package PHP on disk and force its upgrade() method to execute in a single browser navigation. This results in remote code execution as the web server user.   In order to be vulnerable, the victim must be passing canInstallPackages, victim site must be connected to the Concrete marketplace; and the attacker controls the package returned for a marketplace item ID already installed on the victim site. The Concrete CMS security team gave this vulnerability a CVSS v.4.0 score of 7.5 with vector CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:A/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N. Thanks https://github.com/maru1009 for reporting.

pnpm is a package manager. Prior to 10.34.2 and 11.5.3, pnpm can persist package-manager bootstrap metadata in the first YAML document of pnpm-lock.yaml. Before the patch, direct pnpm execution trusted an already resolved packageManagerDependencies entry when the committed env lockfile contained matching pnpm and @pnpm/exe versions. A malicious repository could therefore commit package-manager lockfile package records and snapshots that bypassed fresh package-manager resolution, then cause pnpm to install and execute bytes selected by that committed lockfile state during automatic version switching. This vulnerability is fixed in 10.34.2 and 11.5.3.

OpenClaw before 2026.5.18 contains a code execution vulnerability where marketplace runtime extension metadata can redirect loading toward unscanned package payloads. Attackers with trusted operator access can manipulate extension metadata to load plugin code outside reviewed package entry points, bypassing security scanning.

Concrete CMS 9.5.0 and below emits a CSRF token in the local_available_update.php view ($token->output('do_update')) but the corresponding do_update() method in concrete/controllers/single_page/dashboard/system/update/update.php never calls $this->token->validate('do_update'). The form is rendered as a POST form, meaning the token reaches the browser, but because the controller discards it without verification, an attacker can craft a cross-site POST that triggers a core CMS update to an attacker-specified version string.  In order to be vulnerable, theictim must be passing canUpgrade()anda valid update version must be present under DIR_CORE_UPDATES. The Concrete CMS security team gave this vulnerability a CVSS v.4.0 score of 7.5 with vector CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:A/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N. Thanks https://github.com/maru1009 for reporting.

Atril Document Viewer is the default document reader of the MATE desktop environment for Linux. A single-click remote code execution vulnerability in versions prior to 1.26.3 and 1.28.4 allows an attacker to achieve arbitrary code execution as the user by tricking them into clicking a link inside a malicious PDF document. The PDF can be packaged as a polyglot file that is simultaneously a valid PDF and a valid ELF shared library, making the attack a single-file, single-click, configuration-independent RCE on stock atril installations. The root cause is `shell/ev-application.c:ev_spawn`, which builds a command line from attacker-controlled PDF link-destination fields without applying `g_shell_quote`. The cmdline is then handed to `g_app_info_create_from_commandline`, which shell-parses it back into argv — splitting any embedded `--gtk-module=PATH` into a separate argv element. GTK then `dlopen()`s the path during init, running any `__attribute__((constructor))` it finds. Versions 1.26.3 and 1.28.4 contain a patch for the issue. This is the same defect class as CVE-2023-51698 (CBT `--checkpoint-action` injection in `comics-document.c`, fixed in 1.6.2) but in a different code path (`shell/ev-application.c`) that the original patch did not touch.

A flaw was found in InstructLab. The `linux_train.py` script hardcodes `trust_remote_code=True` when loading models from HuggingFace. This allows a remote attacker to achieve arbitrary Python code execution by convincing a user to run `ilab train/download/generate` with a specially crafted malicious model from the HuggingFace Hub. This vulnerability can lead to complete system compromise.

Inclusion of Functionality from Untrusted Control Sphere vulnerability in Simplehelp.This issue affects Simplehelp: before 5.5.12.

OpenClaw before 2026.4.9 contains an authentication bypass vulnerability allowing untrusted workspace plugins to be auto-enabled during non-interactive onboarding when provider auth choices are shadowed. Attackers can exploit this by crafting malicious workspace plugins that are automatically selected and enabled during authentication setup without explicit user consent.

A DLL search order hijacking vulnerability in Thermalright TR-VISION HOME on Windows (64-bit) allows a local attacker to escalate privileges via DLL side-loading. The application loads certain dynamic-link library (DLL) dependencies using the default Windows search order, which includes directories that may be writable by non-privileged users.\n\n\n\nBecause these directories can be modified by unprivileged users, an attacker can place a malicious DLL with the same name as a legitimate dependency in a directory that is searched before trusted system locations. When the application is executed, which is always with administrative privileges, the malicious DLL is loaded instead of the legitimate library.\n\n\n\nThe application does not enforce restrictions on DLL loading locations and does not verify the integrity or digital signature of loaded libraries. As a result, attacker-controlled code may be executed within the security context of the application, allowing arbitrary code execution with elevated privileges.\n\n\n\nSuccessful exploitation requires that an attacker place a crafted malicious DLL in a user-writable directory that is included in the application's DLL search path and then cause the affected application to be executed. Once loaded, the malicious DLL runs with the same privileges as the application.\n\n\n\nThis issue affects \nTR-VISION HOME  versions up to and including 2.0.5.

Python JSON Logger is a JSON Formatter for Python Logging. Between 30 December 2024 and 4 March 2025 Python JSON Logger was vulnerable to RCE through a missing dependency. This occurred because msgspec-python313-pre was deleted by the owner leaving the name open to being claimed by a third party. If the package was claimed, it would allow them RCE on any Python JSON Logger user who installed the development dependencies on Python 3.13 (e.g. pip install python-json-logger[dev]). This issue has been resolved with 3.3.0.

A vulnerability in the custom URL parser of Cisco Webex App could allow an unauthenticated, remote attacker to persuade a user to download arbitrary files, which could allow the attacker to execute arbitrary commands on the host of the targeted user. This vulnerability is due to insufficient input validation when Cisco Webex App processes a meeting invite link. An attacker could exploit this vulnerability by persuading a user to click a crafted meeting invite link and download arbitrary files. A successful exploit could allow the attacker to execute arbitrary commands with the privileges of the targeted user.

OpenClaw before 2026.3.12 automatically discovers and loads plugins from .OpenClaw/extensions/ without explicit trust verification, allowing arbitrary code execution. Attackers can execute malicious code by including crafted workspace plugins in cloned repositories that execute when users run OpenClaw from the directory.

pnpm is a package manager. Prior to 10.34.2 and 11.5.3, pnpm can install configDependencies declared in pnpm-workspace.yaml before command dispatch. Before the patch, a repository could declare pacquet or @pnpm/pacquet as a config dependency and pnpm treated that repository-controlled dependency as an install-engine opt-in. During install, pnpm resolved a platform-specific @pacquet/<platform>-<arch>/pacquet binary from node_modules/.pnpm-config/<packageName> and spawned it as the developer or CI user. This vulnerability is fixed in 10.34.2 and 11.5.3.

pnpm is a package manager. Prior to 10.34.2 and 11.5.3, the generic peer-suffix normalizer also stripped parenthesized text from git, URL, tarball, file, and other opaque locators. Approval for one source string could therefore authorize a different attacker-controlled source whose locator normalized to the same value. This vulnerability is fixed in 10.34.2 and 11.5.3.

A vulnerability was found in ua-parser-js 0.7.29/0.8.0/1.0.0. It has been rated as critical. This issue affects the crypto mining component which introduces a backdoor. Upgrading to version 0.7.30, 0.8.1 and 1.0.1 is able to address this issue. It is recommended to upgrade the affected component.

The WebDorado Contact Form plugin before 1.13.5 for WordPress allows CSRF via the wp-admin/admin-ajax.php action parameter, with resultant local file inclusion via directory traversal, because there can be a discrepancy between the $_POST['action'] value and the $_GET['action'] value, and the latter is unsanitized.

Untrusted data inclusion in pg_dump in PostgreSQL allows a malicious superuser of the origin server to inject arbitrary code for restore-time execution as the client operating system account running psql to restore the dump, via psql meta-commands. pg_dumpall is also affected. pg_restore is affected when used to generate a plain-format dump. This is similar to MySQL CVE-2024-21096. Versions before PostgreSQL 17.6, 16.10, 15.14, 14.19, and 13.22 are affected.

An iframe that was not permitted to run scripts could do so if the user clicked on a <code>javascript:</code> link. This vulnerability affects Firefox < 102, Firefox ESR < 91.11, Thunderbird < 102, and Thunderbird < 91.11.

Inappropriate implementation in Offline in Google Chrome on Android prior to 90.0.4430.212 allowed a remote attacker who had compromised the renderer process to bypass site isolation via a crafted HTML page.

Cursor is a code editor built for programming with AI. In versions 1.7 and below, automatic loading of project-specific CLI configuration from the current working directory (<project>/.cursor/cli.json) could override certain global configurations in Cursor CLI. This allowed users running the CLI inside a malicious repository to be vulnerable to Remote Code Execution through a combination of permissive configuration (allowing shell commands) and prompt injection delivered via project-specific Rules (<project>/.cursor/rules/rule.mdc) or other mechanisms. The fix for this issue is currently available as a patch 2025.09.17-25b418f. As of October 3, 2025 there is no release version.