Semaine du 21 septembre au 27 septembre 2026

universal-tool-calling-protocol python-utcp

python-utcp versions before 1.1.4 contain a server-side request forgery vulnerability in HttpCommunicationProtocol.call_tool that validates the initial tool URL but follows HTTP redirects without re-validating the target. Attackers controlling a tool endpoint can return a 302 redirect to internal services, allowing the UTCP client to reach cloud metadata endpoints or internal HTTP services and return their response bodies to the caller.

universal-tool-calling-protocol python-utcp

utcp-http before 1.1.4 fails to validate the OAuth2 tokenUrl field from remote OpenAPI specifications, allowing attackers to redirect credential submission to arbitrary endpoints. When a victim registers an attacker-controlled OpenAPI spec and invokes a generated OAuth2-protected tool, the library POSTs the victim's client_id and client_secret to the attacker-supplied token endpoint without URL validation.

heymrun heym

Heym before 0.0.53 evaluates workflow condition expressions using Python's eval() with insufficient sandboxing in the workflow executor service. Authenticated users can edit workflow condition nodes or import malicious templates to execute arbitrary Python and OS commands as the backend process user.

heymrun heym

heym before 0.0.91 contains a sandbox escape vulnerability in the expression engine's DotList map/filter and fallback resolver that allows authenticated users to execute arbitrary Python code. Attackers can craft workflow expressions using dunder attribute access through item expressions or the fallback resolver to access os.system and execute commands as the backend process.

Project-MONAI MONAI

MONAI through 1.6.0 contains a remote code execution vulnerability in the bundle configuration engine that resolves _target_ values to arbitrary importable callables without an allow list and passes $ expressions to Python eval(). Attackers can publish a malicious bundle with crafted configuration containing arbitrary code that executes when a victim loads the bundle using monai.bundle.load() or monai.bundle.run().

OpenClaw

OpenClaw (npm package 'openclaw') versions >= 2026.3.28 and < 2026.8.1 allow an untrusted workspace .env file to set the CLOUDSDK_PYTHON_ARGS environment variable. When an operator starts OpenClaw in attacker-controlled workspace content and then runs the Gmail setup flow, that value is inherited when gcloud is launched, and the gcloud launcher passes it as arguments to the trusted Python interpreter. A crafted CLOUDSDK_PYTHON_ARGS value can therefore cause Python to execute attacker-supplied code with the OpenClaw host user's permissions, allowing credentials to be read, files to be modified, or other processes to be started. This issue is fixed in version 2026.8.1; as a workaround, run Gmail setup only from trusted workspaces and clear inherited CLOUDSDK_* variables beforehand.

horilla-hr

Horilla is an HR and CRM software. Prior to 2.0.0, HorillaListView.export_data in horilla_views/generic/cbv/views.py accepts an authenticated user's columns POST parameter, takes field_tuple[1], interpolates it into dynamic_fn_str as Python source, and passes the generated function definition to exec(). A crafted string that remains valid under ast.literal_eval can inject Python syntax into a default argument evaluated during function definition, allowing arbitrary operating-system commands to execute with the application process privileges, including root privileges in the shipped Docker image. This issue is fixed in version 2.0.0.

MongoDB Python Driver

An integer overflow in the BSON document encoding component of the MongoDB Python Driver's bundled native extension may occur when a single document is built from an unusually large amount of caller-supplied data. Size arithmetic is performed in a signed 32-bit type, and the guard meant to catch the overflow is written in a form whose behavior is not defined by the C language standard. A party with no privileges who can place a very large value into data that an application encodes may, depending on how the native extension was built, cause a write outside the bounds of an allocated buffer inside the application's own process.

python-social-auth social-core

Python Social Auth is a social authentication/registration mechanism. Prior to version 5.0.0, the `vk-app` backend accepted VK application callback data without verifying the callback signature when the `auth_key` parameter was omitted. Applications using this backend could treat unsigned attacker-controlled data as a verified VK identity. An attacker could choose callback fields such as `viewer_id`, `access_token`, `api_id`, and `api_result`, potentially allowing authentication as an arbitrary VK user ID. The issue affects only applications using the `vk-app` backend. The issue has been fixed in version 5.0.0 by requiring `auth_key` to be present and valid before callback data is trusted.

Red Hat Ansible Automation Platform 2.6 for RHEL 9 · Red Hat Ansible Automation Platform 2.6 · Red Hat Ansible Automation Platform 2.7 · Red Hat Ansible Automation Platform 2

A path traversal vulnerability was found in pulpcore. The content upload API accepts a 'file_url' parameter that allows users with file repository privileges to specify a local file URL for Pulp to download and store. A URL scheme validation check uses a string prefix comparison that only rejects URLs beginning with 'file://', but Python's URL parser recognizes the 'file:' scheme without double slashes, creating a mismatch between what is validated and what is dispatched to the file downloader. An authenticated user with low-privilege repository permissions can supply a specially crafted URL using relative path traversal sequences to read any file accessible to the Pulp server process. In deployments that include Pulp Container, successful exploitation allows an attacker to read the container registry token signing private key and forge bearer tokens, granting unauthorized access to all private container repositories in the affected registry.

Red Hat Ansible Automation Platform 2.5 for RHEL 8 · Red Hat Ansible Automation Platform 2.5 for RHEL 9 · Red Hat Ansible Automation Platform 2.6 for RHEL 9 · Red Hat Ansible Automation Platform 2.6

A flaw was found in Ansible Automation Platform's automation-controller. The custom Credential Type environment-variable injector validates variable names against a deny-list (an ANSIBLE_* prefix check plus a fixed ENV_BLOCKLIST) that omits process-hijacking loader variables such as BASH_ENV, ENV, LD_PRELOAD, LD_LIBRARY_PATH, PYTHONSTARTUP and GIT_SSH_COMMAND. Combined with the credential file injector, a privileged user can write an attacker-controlled script into the execution environment and point BASH_ENV at it, obtaining arbitrary code execution inside the execution-environment container for any job that attaches a credential of that type.

Red Hat Ansible Automation Platform 2.5 for RHEL 8 · Red Hat Ansible Automation Platform 2.5 for RHEL 9 · Red Hat Ansible Automation Platform 2.6 for RHEL 9 · Red Hat Ansible Automation Platform 2.6

A flaw was found in Red Hat Ansible Automation Platform's automation- controller. The setting that formats the log message emitted for API 4XX errors is an administrator-controlled Python format-string template that is rendered with a live user object as an argument. Because Python string formatting permits attribute and item traversal on its arguments, an administrator can craft a template that walks from the user object into the application settings and reads the Django secret key and the database password. The formatted message is written to a logger that can be forwarded to an external log aggregator, whose destination is also administrator-controlled, allowing the secrets to be sent off the host. An authenticated administrator can thereby obtain the master encryption key used to protect all stored credentials and the database service password, enabling offline decryption of every stored credential, forgery of user sessions, and direct access to the controller database.

OpenC3 cosmos

OpenC3 COSMOS provides the functionality needed to send commands to and receive data from one or more embedded systems. From 5.12.0 until 7.3.0, an authenticated actor can write the pypi_url setting through set_setting at POST /openc3-api/api, then cause OpenC3::PluginModel.install_phase2 in openc3/lib/openc3/models/plugin_model.rb to interpolate the value into a shell command while installing a plugin with Python dependency metadata. Shell metacharacters in the setting are interpreted by the command shell, allowing arbitrary operating-system commands to run as the openc3 service user with access to Redis and bucket credentials. Open-source deployments permit any authenticated user to reach the affected operations, while Enterprise deployments require an administrator. This issue is fixed in version 7.3.0.

webpy web.py 0.76 is vulnerable to server-side template injection (SSTI). The template engine can be tricked into executing attacker-controlled template code that built-in security checks are designed to reject. When an application precompiles templates from a directory the attacker can write to and later renders them through the precompiled template loader, the sandbox is bypassed and the attacker's code runs, resulting in arbitrary Python code execution and OS command execution on the server.

Pollen Robotics Reachy Mini

The Reachy Mini daemon exposes an HTTP API for managing the robot. Its app installation endpoint, POST /apps/install in src/reachy_mini/daemon/app/routers/apps.py, has no authentication. The handler's only dependency is Depends(get_app_manager), which just hands back the manager object from application state, so nothing in the chain ever checks a credential. The endpoint takes an AppInfo body naming a Hugging Face Space. The daemon downloads that Space and installs it as a Python package through install_package in src/reachy_mini/apps/sources/local_common_venv.py, using uv or pip. Installing a Python package runs the package's own build and setup code, so whoever chooses the Space chooses what code the robot runs. Anyone can publish a public Hugging Face Space, so this is not a meaningful restriction on the attacker. How far this reaches depends on the model. In _resolve_bind_host in src/reachy_mini/daemon/app/main.py the daemon binds 0.0.0.0 when it runs as the wireless version and 127.0.0.1 otherwise, with the vendor's own comment explaining that the robot has to be reachable on the LAN. On a wireless unit, then, any host on the same network can install and run code on the robot without credentials. One related change has already shipped but does not fix this. Version 1.8.2 replaced the wildcard CORS policy with an allow list of localhost and Tauri origins. That closes the browser drive-by route, where a web page the victim visits silently calls the endpoint in the background. It has no effect on this issue: CORS is enforced by browsers and governs whether script may read a response, while a direct HTTP request from another machine on the network involves no browser, no preflight and no CORS check at all.

psd-tools

psd-tools is a Python package for working with Adobe Photoshop PSD files. Prior to 1.17.4, PSDImage.composite() and PSDImage.numpy() allocated output buffers from attacker-controlled PSD header geometry, including width, height, channels, depth, and per-layer rectangles, before validating those values against the available file data. A tiny crafted PSD could therefore cause multi-gigabyte memory allocation, and PSDImage.composite() could return a black image with only a warning instead of raising an exception. Services that composite untrusted PSD files could be terminated by out-of-memory handling. This issue is fixed in version 1.17.4.

Crypt::SelfCertificate versions from 1.01 through 1.05 for Perl contains malware which executes Python code from an obfuscated URL. The generate_certificate runs a Python script saved as a certificate file. The pyhton script attempts to retrieve code from a hardcoded http URL that is obfuscated with base64 encoding and run the response body directly. The impact is that arbitrary code can be invoked as the user, without a dropped script being saved on the affected host. The releases have no test scripts nor build hooks. The intention may have been to trigger the payload after installation. For version 1.01, the dropper script is in lib/Crypt/SelfCertificate/sample/validate.p12. For version 1.05, the dropper script is in lib/Crypt/SelfCertificate/sample/cert7.pem. The SHA-256 digests of the files are fbff21f45ff748365062a5e36fb2d72558cad82a507a6f357f320b4fcdf07760 Crypt-SelfCertificate-1.01.tar.gz 27b2d2d3174ad771474fff2521f5084ec231e9218ea8c832515aef1cbd5897bc lib/Crypt/SelfCertificate/sample/validate.p12 9fdfa7d69b034b77d4510cda567e8da1e486ca81c7daaadc5732a45c41d71991 Crypt-SelfCertificate-1.05.tar.gz 27b2d2d3174ad771474fff2521f5084ec231e9218ea8c832515aef1cbd5897bc lib/Crypt/SelfCertificate/sample/cert7.pem

sooperset mcp-atlassian · Mcp-atlassian mcp atlassian

MCP Atlassian is a Model Context Protocol (MCP) server for Atlassian products (Confluence and Jira). Prior to 0.22.0, validate_safe_path defaults its base directory to os.getcwd(), and affected Confluence attachment call sites omit base_dir, allowing attacker-selected writes within the working directory. This Python module overwrite can provide code execution when the application later imports the modified module, bypassing the remediation tracked as CVE-2026-27825. This issue is fixed in version 0.22.0.

kap-sh zapros

Zapros, a Python HTTP client, prior to version 0.14.0 is vulnerable to denial of service via memory exhaustion. The issue affects all callers who streamed compressed responses relying on the chunk size — explicit (`iter_bytes(chunk_size=...)`) or the default — to bound memory. The decoder ignored that bound, so a chunk could be far larger than requested and a single compressed response could overflow memory. Version 0.14.0 contains a patch. Some workarounds are available. Read the still-compressed body with `Response.iter_raw()` / `Response.async_iter_raw()`, which bypass the built-in decoders, and decompress it yourself with an explicit output-size bound (e.g. `zlib`'s `max_length`), aborting once a configured limit is exceeded. Where feasible, send `Accept-Encoding: identity` to disable response compression so bodies are not decompressed client-side. Avoid decoding response bodies from untrusted servers.

Tautulli

Tautulli is a Python based monitoring and tracking tool for Plex Media Server. Prior to 2.17.2, the import_config handler and the database_file branch of import_database in plexpy/webserve.py join the attacker-controlled config_file.filename or database_file.filename directly to CACHE_DIR without basename reduction or a containment check. An administrator or caller with the instance API key can submit a multipart filename containing parent-directory segments, causing the upload to be created or overwritten outside CACHE_DIR before file-content validation runs. The write is limited to paths permitted to the Tautulli process, but it can enable configuration tampering, service disruption, or code execution. This issue is fixed in version 2.17.2.

conda

Conda is a system-level binary package and environment manager that runs on major operating systems and platforms. Prior to 26.5.2, parse_entry_point_def in conda/common/path/python.py accepted an unvalidated entry-point command from a noarch:python package's info/link.json metadata. CreatePythonEntryPointAction in conda/core/path_actions.py interpolated that command into target_short_path, and PrefixPathAction.target_full_path joined it to the installation prefix without verifying that the result remained under the intended bin or Scripts directory. create_python_entry_point in conda/gateways/disk/create.py then wrote an executable wrapper to the resulting path. A malicious package could use path separators, traversal segments, or an absolute command path to write outside the prefix or overwrite another in-prefix entry point during default install and environment transactions. Out-of-prefix writes require the target parent directory to exist, while an overwritten entry point can execute attacker-controlled Python when later invoked with the installing user's privileges. This issue is fixed in version 26.5.2.

SepineTam mcp-for-stata

MCP-for-Stata is a MCP server for integrating Stata into agent loops with a safety-first design. Prior to version 1.19.0, the ado_package_install MCP tool in stata-mcp concatenates user-controlled input directly into a Stata command string without any validation or sanitization. An attacker who can invoke the MCP tool or the equivalent Python API can embed newline characters in the package argument to inject arbitrary Stata commands. Because Stata supports a shell escape command, this leads to full OS-level arbitrary command execution (RCE) under the account running the Stata-MCP server. The tool is registered in the default all profile, so no non-default configuration is required. This issue has been patched in version 1.19.0.