Gitea versions prior to 28.3.0 contain an authorization flaw where personal access tokens restricted to public resources can be used to register organization-level runners that access private repository workflows, secrets, and variables. This vulnerability allows an attacker with a token having write:organization scope from a public organization owner to gain unintended access to private repository jobs. The issue is fixed in version 28.3.0.
A vulnerability in Gitea versions prior to 28.3.0 allowed exposure of contribution heatmap data derived from private repository activity to holders of public-only access tokens. While repository names and contents were not disclosed, unauthorized actors could view contribution counts and their 15-minute time buckets from private repositories. This issue has been fixed in version 28.3.0, where public-only tokens now restrict heatmap data to activity visible to anonymous users.
CVE-2026-113429 is a cross-site scripting (XSS) vulnerability in Gitea affecting legacy Actions artifact uploads (versions prior to 28.3.0). The vulnerability arises because the Content-Encoding header from artifact uploads was stored unchecked and used as the Content-Type when serving the artifact. If the Content-Type starts with video/ or audio/, it disables the sandboxing Content-Security-Policy, allowing an attacker to upload an artifact that a browser renders as HTML. This enables execution of attacker-controlled scripts in the context of the Gitea origin, potentially leading to session hijacking or full instance control if an administrator is targeted. The issue is fixed in version 28.3.0 by restricting stored headers and serve headers to valid media types only.
Gitea versions prior to 28.3.0 contain a vulnerability where Actions task tokens authenticate as a shared system user with no owner restrictions, allowing users who can run workflows to list all repositories on the instance, including private ones, via the REST API. Repository contents remain protected by permission checks. The issue is fixed in version 28.3.0 by ensuring system users own no repositories and restricting API routes accordingly.
CVE-2026-120263 is an improper authentication vulnerability in Gitea versions before 28.3.0. When the OAuth2 client setting ACCOUNT_LINKING is set to auto and registration is enabled, the system incorrectly linked external OAuth2 identities to existing Gitea accounts based on submitted user names or emails rather than the identity asserted by the OAuth2 provider. This flaw allowed an attacker with an account on the OAuth2 provider to link their external identity to another user's Gitea account and sign in as that user without a password, including administrator accounts. The vulnerability was fixed in version 28.3.0 by restricting automatic linking to cases where the existing account matches the provider's email or provider-derived user name, otherwise requiring the account password.
CVE-2026-118386 is an incorrect authorization vulnerability in Gitea affecting versions from 1.19.0 up to but not including 28.3.0. It allows a user with read access to a repository to trigger workflows with elevated privileges by submitting a pull request review referencing a commit they control, bypassing fork restrictions. This leads to unauthorized execution of workflows with access to repository secrets and a token granting write access. The issue is fixed in Gitea version 28.3.0.
A cross-site scripting (XSS) vulnerability exists in Typemill up to version 2.26.0 in the ControllerApiImage::saveImage function within SvgSanitizer.php. This flaw allows remote attackers to perform script injection via manipulated input. The issue is resolved by upgrading to Typemill version 2.27.0.
A vulnerability in Delgan loguru version 0.7.3 affects the AnsiParser.feed function in the ANSI Color-Markup Parser component. The issue involves inefficient regular expression complexity, which can be exploited remotely to cause a denial of service by resource exhaustion. Public exploit code is available, and the vendor has not responded to disclosure attempts. The CVSS score is 4.3, indicating a medium severity impact focused on availability.
This report discusses a tool enhancement for extracting embedded files from Microsoft OneNote (.one) files. The author added a JSON output feature to the onedump.py tool to facilitate batch extraction of embedded files, enabling easier analysis and file retrieval. There is no indication of a vulnerability or active exploitation associated with this activity.
The npm package @axon-web/icons version 99.0.0 is a malicious package that performs unauthorized data exfiltration upon loading. It constructs an identifier from the host's OS hostname and environment username, then leaks this information via DNS requests to an attacker-controlled domain. Additionally, it scans environment variables for names containing sensitive keywords related to credentials and outputs these key names, indicating reconnaissance activity. The package contains no legitimate icon content and appears to be a dependency confusion or typosquatting attack.
The hs-framework npm package declares a dependency on a tarball hosted at an external HTTPS URL rather than a versioned registry package. This tarball can be changed at any time by whoever controls the storage bucket, allowing arbitrary code execution during npm install via lifecycle scripts. The package itself is a minimal stub, with the dependency URL being the main attack surface. This setup enables supply chain compromise through dynamic, unpinned, and unverified code fetched during installation.
The privacypass-py package on PyPI masquerades as a Privacy Pass client library but contains no cryptographic functionality. Instead, it performs data exfiltration by collecting sensitive environment variables, system information, and process details, then sending this data to a hardcoded external IP address over unencrypted HTTP. This malicious behavior occurs both during package installation and on module import, bypassing typical build isolation protections.
The MapPress WordPress plugin versions before 2.97.15 have a vulnerability where the front-end map embed handler does not restrict access to user data. This flaw allows unauthenticated visitors to enumerate user IDs and retrieve email addresses and display names of any user on the site.
The Kali Forms WordPress plugin before version 2.4.30 has a vulnerability where its test-email handler does not perform proper capability checks. Instead, it relies solely on a nonce accessible to any logged-in user, allowing low-privileged users such as Subscribers to send emails from the site with arbitrary recipients and spoofed sender addresses.
A critical OS command injection vulnerability exists in the WAVLINK WL-WN530HG4B M_V260724 router, specifically in the port_forward function of /cgi-bin/wapi.cgi. This flaw allows remote attackers with low privileges to execute arbitrary OS commands by manipulating the Name argument. Exploit code has been published, increasing the risk of exploitation. No patch or remediation information is currently available.
The Media Library Assistant WordPress plugin versions before 3.44 have a vulnerability where the plugin does not verify if the requesting user has permission to view a post before revealing the taxonomy terms assigned to it. This flaw allows any authenticated user, including low-privileged roles such as subscribers, to discover the term assignments of private attachments and posts.
A critical command injection vulnerability exists in WAVLINK WL-WN530BE1-A version V251117 in the function con_check_read_finish of the file /protocol.csp. The vulnerability allows remote attackers to execute arbitrary commands by manipulating the argument group_name/IP-FROM. Public exploit code has been disclosed. No patch or remediation information is currently available.
A stored cross-site scripting (XSS) vulnerability exists in the Media Library Assistant WordPress plugin versions before 3.44. The flaw arises because one of the plugin's shortcode attributes is not properly escaped before being output in an attribute of generated links. This allows users with Contributor role or higher to inject malicious scripts that persist and execute in other users' browsers.
The Kali Forms WordPress plugin before version 2.4.30 contains a vulnerability where its test-email handler does not perform proper capability checks. This flaw allows any authenticated user with minimal privileges (Subscriber and above) to disclose the email addresses of any registered user, including administrators, by exploiting a nonce that is accessible to all logged-in users.
A critical vulnerability exists in WAVLINK WN579X3-D firmware version V250402 in the port_forward function of /cgi-bin/wapi.cgi. An attacker can remotely perform OS command injection by manipulating the Name argument. Public exploit code is available, increasing the risk of exploitation. No patch or remediation information is currently provided by the vendor.