Bazarr is a companion application to Sonarr and Radarr. Prior to 1.6.3-beta.11, the /api/subtitles/subtitles-info handler uses SubtitleNameContents.get() in subtitles_contents.py and reads the caller-supplied subtitlePath after checking only its subtitle extension; because the resolved path is not confined to files indexed by Bazarr, traversal can read files outside the media library. The /api/subtitles/subtitles handler in subtitles.py passes caller-supplied path and reference to sync and translation without per-path authorization, allowing an out-of-library reference file to be read and output to overwrite a file outside the media library. The authenticate decorator gates both endpoints when API authentication is enabled, while auth = None allows access without credentials. Both paths can affect only files that the Bazarr process can read or write. This issue is fixed in version 1.6.3-beta.11.
Rucio is a scientific data management framework. From 1.14.0 until 35.9.1, 38.6.1, 40.4.2, and 41.1.1, validate_auth_token in lib/rucio/core/authentication.py does not distinguish short-lived SSH challenge nonces from authentication tokens stored in the same token table. An unauthenticated attacker can request /auth/ssh_challenge_token for an arbitrary account by supplying X-Rucio-Account, then present the returned X-Rucio-SSH-Challenge-Token as X-Rucio-Auth-Token to authenticate to other API endpoints as that account, including an administrative account. This issue is fixed in versions 35.9.1, 38.6.1, 40.4.2, and 41.1.1.
The Model Context Protocol Python SDK is the official Python SDK for Model Context Protocol servers and clients. From 1.9.1 until 1.30.0 and 2.2.0, OAuth client support in mcp.client.auth can send credentials to an authorization server chosen by an untrusted MCP server. ClientCredentialsOAuthProvider and PrivateKeyJWTOAuthProvider cannot identify the authorization server that owns their credentials. Affected HTTP clients using those providers, OAuthClientProvider, or the deprecated 1.x RFC7523OAuthClientProvider can disclose a client_secret, authorization code, PKCE code_verifier, or signed client assertion. MCP servers, stdio clients, and clients that attach their own tokens or headers are not affected. Interactive OAuthClientProvider use requires a person to start sign-in, while the unattended providers do not. For ClientCredentialsOAuthProvider and PrivateKeyJWTOAuthProvider, callers must configure issuer= after upgrading because omission still follows the MCP server's advertised authorization server, and the deprecated RFC7523OAuthClientProvider must be replaced because it has no issuer option. A stored client registration without an issuer remains unbound after upgrading, so stored OAuth client information must be cleared once to register again or a manually pre-registered record must be updated with its issuer. This issue is fixed in versions 1.30.0 and 2.2.0.
C-Kermit 8.0.200 before 11.1.515 contains a buffer overflow vulnerability in getreply() in ckcftp.c that allows malicious FTP servers to overwrite client memory by sending overlong security replies. Attackers controlling or intercepting an FTP server can send long PBSZ replies after AUTH TLS, or long ADAT replies in GSSAPI, Kerberos 4, and SRP builds, overflowing the static reply_buf with attacker-controlled bytes.
A cross-site request forgery (CSRF) vulnerability exists in the SAML backend of social-auth-core versions 4.7.0 up to but not including 7.0.0. This flaw allows attackers to deliver a valid identity provider (IdP) response and RelayState to a victim's browser, causing the victim to be signed into the attacker's account. The vulnerability arises because the SAML backend restores sessions named in the unsigned RelayState parameter, enabling session fixation attacks.
Yeti bridges the gap between CTI and DFIR practitioners by providing a Forensics Intelligence platform and pipeline. Prior to 2.5.1, YETI registers the audit router without Depends(auth.get_current_active_user), and core/web/apiv2/audit.py does not perform a route-level authentication check for /api/v2/audit/timeline/{id}. An unauthenticated network attacker can use an empty, partial, or specific wildcard path to query TimelineLog records containing the actor, target_id, action, and details fields, including analyst usernames, object identifiers and values, and before-and-after object data. This exposes the operational history of threat-intelligence objects while other application routers require authentication. This issue is fixed in version 2.5.1.
A vulnerability in the filelock Python package versions 4.1.0 through 4.2.0 allows a local user with write permission on the directory containing the lock database to rename directories and cause the lock user's process to recursively delete unintended directories. This occurs because the ReadWriteLock and AsyncReadWriteLock implementations remove their private .filelock-* directories without verifying that the path still points to the originally created directory. The issue affects macOS since version 4.1.1 and POSIX environments without a usable /proc/self/fd, such as chroot or containers without /proc mounted. The vulnerability leads to high integrity impact by enabling deletion of directories owned by the lock user without error reporting. The CVSS 3.1 score is 6.1 (Medium).
A vulnerability in filelock versions 4.1.0 through 4.2.0 allows a local unprivileged user who can write to a shared lock directory without the sticky bit to exploit a race condition in the soft-lock stale break restore process. This leads to the creation of a hard link to a file owned by the lock user, exposing file metadata and enabling the attacker to manipulate the file's name. On version 4.2.0, this can result in truncation of the lock user's file. The issue arises due to os.link following symbolic links on CPython 3.12+ on Linux and macOS, combined with directory permissions and Linux kernel settings.
MLflow versions from 3.12.0rc0 up to but not including 3.17.0 contain a critical code injection vulnerability in the third-party scorer deserialization mechanism. This flaw allows an attacker to place a malicious Python module under a writable allowed package and invoke it via MLflow artifact endpoints, leading to remote code execution. Deployments without authentication are fully exposed, while basic-auth deployments require credentials with experiment creation and artifact upload permissions. The vulnerability is fixed in version 3.17.0.
A vulnerability in python-social-auth's social-core prior to version 6.2.0 allows an attacker to bypass nonce validation in OpenIdConnectAuth. The nonce is stored globally rather than being bound to the browser session and OAuth state, enabling an attacker who obtains a victim's unused authorization code to impersonate the victim if DEFAULT_USE_PKCE is disabled. This issue is resolved in version 6.2.0.
Netcore NR256CN devices up to version 1.5.140116 have a critical pre-authentication stack-based buffer overflow vulnerability in the Boa web server's user_ok() function. This flaw arises from unbounded strcpy() calls copying Basic-auth usernames into fixed 64-byte stack buffers. An unauthenticated attacker can exploit this by sending a username longer than 63 bytes to any protected /router/*.cgi route, potentially crashing the Boa process or executing arbitrary code.
A Cross-Site Request Forgery (CSRF) vulnerability exists in python-social-auth's social-core LINE backend prior to version 5.0.2. The vulnerability arises because the LINE backend exchanges an authorization code without validating the returned OAuth state parameter. This flaw allows an attacker to cause a victim to be logged into the attacker's LINE account or link the attacker's LINE identity to the victim's account when account linking is enabled. The issue is fixed in version 5.0.2.
HoloViz Panel versions from 1.2.2 up to but not including 1.9.0 contain an open redirect vulnerability. This flaw exists in panel/auth.py where the next query parameter is not validated, allowing attackers to redirect users to external malicious sites after login. The vulnerability affects the BasicLoginHandler and PasswordLoginHandler, which store the next parameter in a cookie and redirect users accordingly. The CVSS 4.0 score is 5.1, indicating a medium severity issue.
CVE-2026-108818 is a medium severity vulnerability in agno-agi agno versions 2.7.0 through 2.7.3 and 3.0.0 through 3.0.6. It involves missing authentication due to improper handling of the security_key argument in the get_authentication_dependency function within auth.py. The vulnerability can be exploited remotely without privileges or user interaction. Public exploit code exists, but no vendor response or patch is currently available.
Tugtainer is a self-hosted app for automating updates of Docker containers. Prior to version 1.31.3, when the OIDC login flow completes, backend/modules/auth/providers/auth_oidc_provider.py decodes the id_token returned by the identity provider's token endpoint using jose.jwt.get_unverified_claims() instead of jwt.decode(). This skips signature verification, audience (aud) validation, issuer (iss) validation, and expiry (exp) checking entirely. The extracted claims (email/sub/preferred_username) are then used directly as the user_id for the resulting Tugtainer session. This issue has been patched in version 1.31.3.
Tugtainer is a self-hosted app for automating updates of Docker containers. Prior to version 1.30.4, Tugtainer Agent allows unauthenticated access to Docker management APIs when AGENT_SECRET is not configured. The Agent uses request signatures to protect its API routes. However, in agent/auth.py, the signature verification function returns successfully if Config.AGENT_SECRET is empty. This causes protected Agent APIs to become accessible without authentication. This issue has been patched in version 1.30.4.