CVE-2026-113723 is a vulnerability in virtualenv's bash activation prompt handling where attacker-controlled prompt text can be interpreted as executable shell syntax under certain older bash compatibility modes or when using dash. This occurs because the prompt string assigned to PS1 is expanded twice, allowing command execution when the prompt is drawn. The issue affects POSIX activation scripts and arises when an attacker can supply the prompt argument to virtualenv and a user activates the environment in an affected shell context. The vulnerability was introduced in virtualenv 20.0.0b1 and 21.14.7 and fixed in 2026.10.8_1. Exploitation requires user interaction to activate the environment and draw the prompt.
A vulnerability in filelock's ReadWriteLock and AsyncReadWriteLock components allows a local attacker with write permissions in the directory containing the lock database to manipulate SQLite lock files via symlinks or planted journal files. This can cause the lock user to write arbitrary data to their own files, create or unlink files, or cause lock acquisition failures. The issue affects versions 3.20.4 through 4.2.0, with a fix available in version 2026.10.8_1. The vulnerability requires a writable directory without the sticky bit and local user access. It has a CVSS score of 7.1 (High) and impacts integrity and availability but not confidentiality.
The ReadWriteLock and AsyncReadWriteLock classes in the pdm product resolve lock file paths by following symlinks and pass these resolved paths to SQLite without preventing symlink traversal. This allows a local attacker with write access to the lock directory to plant symlinks that cause the victim process to create or lock files at unintended locations, potentially breaking mutual exclusion and corrupting protected resources. The issue affects pdm versions from 2.26.7 up to but not including 2.29.3 and has a CVSS score of 6.1 (medium severity).
GitPython has three wrapper methods—Git.ls_remote, Repo.merge_base, and IndexFile.move—that omit important unsafe protocol and option checks designed to prevent command injection vulnerabilities. The ls_remote method forwards URLs unchecked, allowing remote code execution (RCE) if the victim's git transport policy permits unsafe protocols like 'ext'. The merge_base and move methods forward caller arguments without guards, exposing latent injection surfaces though no direct RCE is currently confirmed. These issues remain unfixed as of the main branch commit dated 2026-09-21, after the latest release 3.0.0_15 which contains the fix. The vulnerability is rated high severity with a CVSS score of 7.5.
Authlib versions up to 1.7.2 accept an empty HMAC key for JWS/JWT HS256/HS384/HS512 signature verification. This allows attackers to forge tokens with arbitrary claims that pass signature verification, leading to token forgery and authentication bypass in OAuth 2.0 resource servers and OIDC relying parties using Authlib. The vulnerability arises because empty keys produce a deterministic digest anyone can reproduce, and no check rejects zero-length keys. A fix is available in Authlib version 0.26.1.
Authlib versions from 0.13 up to 1.7.2 contain a vulnerability in the is_secure_transport() function, which incorrectly allows certain HTTP URLs with a userinfo component to bypass TLS enforcement. This occurs because the function uses a simple string prefix check that can be fooled by URLs like 'http://localhost:80@evil.example/cb', where the actual host is not a loopback address but an external domain. This flaw affects OAuth 2.0 and OIDC endpoint and redirect URI validation, potentially allowing cleartext HTTP URLs to be accepted as secure in contexts such as OIDC RP-Initiated Logout and authorization server metadata validation. The severity is rated low and a fix is available in Authlib version 0.24.0.
A vulnerability in filelock versions 3.22.0 through 4.2.0 allows a local user with write access to a lock marker file (but without permission to remove it) to forge the process ID in the marker. This causes another process running as the lock owner or root to mistakenly consider the lock stale, remove the live marker, and acquire the lock concurrently. This results in two processes entering the critical section simultaneously without error, potentially corrupting data integrity. The issue arises when the marker file is created with group or other write permissions, such as with an explicit permissive mode or a lax umask (002 or 000) in a sticky or non-removable directory. The default umask 022 is not affected. The vulnerability has a CVSS 3.1 score of 4.7 (Medium).
filelock versions 3.22.0 through 4.2.0 contain a vulnerability in their stale-lock break mechanism for SoftFileLock and related locks. The flaw arises when a stale lock file is identified by inode number and modification time, but inode reuse on certain filesystems (ext4, xfs, overlayfs) allows a new live lock file to reuse the same inode number. This can cause two processes to simultaneously hold the same lock without detection, compromising data integrity. The issue requires local access by another user who can create files in the lock directory (e.g., a sticky /tmp). The vulnerability has medium severity with a CVSS score of 4.7 and no confidentiality or availability impact, but high integrity impact.
GitPython versions up to 3.1.62 contain a vulnerability in the Head.checkout() method where an attacker-controlled branch name can be used to perform an arbitrary file read. This occurs because the method does not properly sanitize its positional argument (the branch name), allowing malicious branch names that inject Git command options such as --pathspec-from-file to read arbitrary files. The vulnerability was introduced due to an incomplete fix of previous advisories and affects all published releases up to 3.1.62. The issue is rated high severity with a CVSS score of 7.5.
Authlib's `OctKey.import_key()` function contains an incomplete fix for an algorithm confusion vulnerability that allows an attacker to forge HMAC-signed JWT tokens using an asymmetric public key with a prefixed byte. The vulnerability arises because the guard that prevents asymmetric keys from being used as HMAC secrets only checks if the key bytes start exactly with certain PEM headers, but the loader accepts keys with leading whitespace or comments before the header. This discrepancy allows an attacker who possesses the issuer's public key (with a leading byte) to create valid HS256 tokens that the system accepts as authentic. The issue affects Authlib versions from 3.2.4 up to but not including 3.8.1_3 and was fixed in 3.8.1_3. The vulnerability has a CVSS 3.1 score of 7.4 (high severity).
A command injection vulnerability exists in the bash activate script of virtualenv used by conda-lock. The script improperly escapes environment names containing backslashes before dollar signs, allowing attacker-controlled environment names to execute arbitrary commands in the user's shell each time the prompt is drawn. This affects virtualenv versions including and after 21.14.6. The issue is fixed in version 21.14.7 by properly escaping special characters for both bash prompt parsing passes.
A vulnerability in filelock versions >=4.1.0 and <=4.2.0 allows a local unprivileged user to evict a fresh lock file held by another user by exploiting inode reuse and coarse clock timestamp manipulation on ext4 or overlayfs filesystems. This causes two processes of the lock owner to enter a critical section simultaneously, violating lock integrity without affecting confidentiality or availability. The attack requires specific filesystem behavior, timing precision, and the ability to create files in the lock directory, such as a sticky /tmp. The CVSS 3.1 score is 4.7 (Medium), reflecting local access with high attack complexity and high integrity impact.