CVE Vulnerability Catalog

Vulnerability in PraisonAI before version 4.5.90, where MCPToolIndex.search_tools() compiles a caller-supplied string directly as a Python regular expression without validation, sanitization, or timeout. A crafted regex causes catastrophic backtracking in the re engine, blocking the Python thread for hundreds of seconds and causing a complete service outage.

Avahi, a system for service discovery on a local network via mDNS/DNS-SD, prior to version 0.9-rc4 is vulnerable to a DoS attack. Any unprivileged local user can send a single D-Bus method call with conflicting publish flags, causing avahi-daemon to crash.

A SQL injection vulnerability was found in Emlog version 2.6.2 and earlier in the updateTagName() function in include/model/tag_model.php. The function directly interpolates user input into SQL queries without parameterized queries, allowing SQL injection attacks.

In Emlog version 2.6.2 and earlier, a Local File Inclusion (LFI) vulnerability exists in admin/plugin.php. The $plugin parameter from the GET request is directly used in a require_once path without proper sanitization. If the CSRF token check can be bypassed, an attacker can include arbitrary PHP files from the server filesystem, leading to code execution.

Emlog prior to version 2.6.8 has a stored XSS vulnerability in the comment module via URI scheme validation bypass. Attackers can inject malicious scripts that execute when the comment page is viewed.

Emlog prior to version 2.6.8 contains a vulnerability in the backend upgrade interface that accepts remote SQL and ZIP URLs via GET parameters. The server downloads and executes the SQL file, then downloads the ZIP file and extracts it directly into the web root. The process does not validate a CSRF token, so an attacker can trick an authenticated administrator into visiting a malicious link, leading to arbitrary SQL execution and file write.

nimiq/core-rs-albatross prior to version 1.3.0 contains a vulnerability where an elected validator proposer can send an election macro block whose header.interlink does not match the canonical next interlink. Honest validators accept the proposal because verify_macro_block_proposal() does not check the interlink binding for election blocks. The same finalized block is later rejected by verify_block() with InvalidInterlink, but this happens after Tendermint decides, potentially causing consensus issues.

In LTI JupyterHub Authenticator prior to version 1.6.3, the LTI 1.1 validator stores OAuth nonces in a class-level dictionary that grows without bounds. Nonces are added before signature validation, so an attacker with knowledge of a valid consumer key can send repeated requests with unique nonces to gradually exhaust server memory, causing a denial of service.

Hirschmann HiLCOS products (OpenBAT, BAT450, WLC, BAT867) have a firewall filtering vulnerability that fails to correctly filter IPv4 multicast and broadcast traffic when management IP address filtering is disabled, allowing configured filter rules to be bypassed.

OpenPrinting CUPS up to version 2.4.16 has a heap-based buffer overflow in the CUPS scheduler when building filter option strings from job attributes. At the time of publication, no public patches are available.

In OpenPrinting CUPS versions 2.4.16 and prior, the RSS notifier allows path traversal in notify-recipient-uri, enabling a remote IPP client to write RSS XML bytes outside CacheDir/rss to any lp-writable location. By exploiting default group-writable permissions (root:lp, mode 0770), the notifier running as lp can overwrite root-managed state files, causing job cache parsing failures and loss of queued jobs after cupsd restart.

Discourse, an open-source discussion platform, has a vulnerability that exposes staged user custom fields and usernames on public invite pages without email verification. This affects versions from 2026.1.0-latest to before 2026.1.3, 2026.2.0-latest to before 2026.2.2, and 2026.3.0-latest to before 2026.3.0.

JupyterHub before version 5.4.4 has an open redirect vulnerability that allows attackers to construct links which, when clicked, take users to the JupyterHub login page, after which they are sent to an arbitrary attacker-controlled site, bypassing JupyterHub's check to prevent this.

Discourse versions from 2026.1.0 to before 2026.1.3, 2026.2.0 to before 2026.2.2, and 2026.3.0 to before 2026.3.0 have an authorization bypass vulnerability. Unauthenticated or unauthorized users can view staff-only tags and associated data.

In util-linux prior to version 2.41.4, a TOCTOU vulnerability has been identified in the SUID binary /usr/bin/mount. When setting up loop devices, mount validates the source file path with user privileges but later re-canonicalizes and opens it with root privileges without verifying that the path has not been replaced. A local unprivileged user can replace the source file with a symlink pointing to any root-owned file or device.

OpenPrinting CUPS versions 2.4.16 and prior contain an authorization bypass vulnerability due to case-insensitive username comparison. This allows an unprivileged user to gain unauthorized access to restricted operations by using a username that differs only in case from an authorized user. No public patches are available at the time of publication.

The Amazon Athena ODBC driver before version 2.1.0.0 contains an out-of-bounds write vulnerability in query processing components. Specially crafted data can crash the driver during query operations.

OpenClaw before version 2026.4.2 reuses the PKCE verifier as the OAuth state parameter in the Gemini OAuth flow, exposing it in the redirect URL. Attackers who capture the redirect URL can obtain both the authorization code and PKCE verifier, defeating PKCE protection.

Development and test API endpoints that mirror production functionality are present in the application. This means they may be accessible in the production environment, posing a risk of unauthorized access.

A specific administrative endpoint 'notifications' is accessible without proper authentication.