Mercredi 30 septembre 2026

jupyterlab

JupyterLab is an extensible environment for interactive and reproducible computing, based on the Jupyter Notebook Architecture. From JupyterLab 4.0.0 until 4.5.11 and 4.6.4, the PyPI Extension Manager uninstall request reaches ExtensionHandler.post, which validates extension names for installation but passes uninstall names to PyPIExtensionManager.uninstall and python -m pip uninstall without rejecting option-like values. The security impact requires that the PyPI Extension Manager is enabled, the account can call the extension API, and kernels and terminals are disabled or delegated to remote hosts; otherwise the user can already read files and make outbound requests directly. An authenticated user with extension API access can supply a pip requirements option to make the server read a local file or fetch an internal URL, and reflected parse errors can return the first unparsable line or response content. A pip log option can also create or corrupt a chosen path with pip-generated log text, but the requester cannot select an arbitrary disclosed line or arbitrary file content, and the injection does not add code execution or availability impact beyond ordinary package removal. This issue is fixed in JupyterLab 4.5.11 and 4.6.4.

langflow-ai langflow

Langflow 1.0.16 before 1.12.0 and 0.0.94 before 1.12.0 contain an unsafe eval() vulnerability in schema.py that allows authenticated attackers to achieve code execution by placing a Python object with a malicious __repr__ method into component input options lists. The eval() sink is triggered when a component is converted into a LangChain tool via ComponentToolkit.get_tools(), including during custom component saves through the API, by interpolating options into a Literal type string that is passed directly to eval() without safe evaluation controls.

sqlalchemy mako

Mako is a template library written in Python. Prior to 1.4.2, on Windows, TemplateLookup.get_template() in mako/lookup.py resolves template URIs with posixpath, while Template.__init__() in mako/template.py validates them with os.path, which uses ntpath. A URI beginning with a drive designator causes ntpath to absorb the traversal segments before the leading dot-dot check, while posixpath resolution can escape the configured template directory. An application that passes attacker-controlled template names or include paths can disclose process-readable files on the same volume, and a targeted file containing Mako template syntax may also be parsed and executed as a template. Raw URL paths are generally normalized before reaching this form, but query strings, form or JSON bodies, route parameters, and dynamic include expressions can preserve it. This issue is fixed in version 1.4.2.

pypa virtualenv

virtualenv is a tool for creating isolated virtual python environments. Prior to 21.7.11, PyEnvCfg.write() writes prompt values verbatim to the line-oriented pyvenv.cfg format while PyEnvCfg._read_values() parses the file with str.splitlines() and accepts the last value for duplicate keys. An attacker who influences --prompt, VIRTUALENV_PROMPT, or configuration input can insert a recognized line boundary and additional keys, including home, causing consumers to use an attacker-selected base interpreter or corrupted environment metadata. The security impact requires prompt input from outside the operator's trust boundary; directly supplied prompt content primarily corrupts the operator's own environment. This issue is fixed in version 21.7.11.

devpi

devpi is a Python package index staging server and packaging, testing, and release tool. Prior to 6.20.2 and 7.0.0b3, a server configured with the primary or deprecated master role allows an unauthenticated, modified GET request to the +changelog route because verify_primary does not reject a missing identity and therefore fails to require ReplicaIdentity. The response can disclose complete database content, including Argon2 password hashes and identifiers and salts for devpi-tokens; exposed hashes may be subject to dictionary attacks, and public tokens may assist attempts to derive the server secret. Large responses can also consume significant CPU, input/output capacity, and bandwidth. Servers using the standalone role are not exposed through replication, and an instance served exclusively through nginx with devpi-lockdown redirects the request to login with no known exploit. This issue is fixed in devpi-server versions 6.20.2 and 7.0.0b3.

psd-tools

psd-tools is a Python package for working with Adobe Photoshop PSD files. Prior to version 1.17.1, `SmartObject.save()` writes an embedded smart object to a path taken verbatim from the PSD file. Because that name is attacker-controlled and unsanitised, a tool that extracts embedded objects from an untrusted `.psd` can be made to write attacker-chosen bytes to an attacker-chosen path (absolute or `../`-traversing), outside its intended output directory. A secondary issue in `SmartObject.open()` for external-kind smart objects allows the attacker-controlled `fullPath` descriptor to be used as an arbitrary file read path, enabling exfiltration of the read content to the controlled write destination. Both issues are fixed in v1.17.1.

gitpython-developers GitPython

GitPython before 3.1.62 does not validate the `path` field read from an untrusted .gitmodules file when updating submodules. While a prior fix (GHSA-hmq2-w58f-27jc) added Submodule._validated_name() to constrain the `name` field, and GitPython's own containment guard Submodule._to_relative_path() is applied in add() and move(), Submodule.update() derives the absolute checkout location from the raw `path` value without that guard. A .gitmodules entry containing directory traversal components (e.g., path = ../../../tmp/escaped) can therefore cause directories to be created via os.makedirs() outside the repository working tree, populated from the submodule URL on the clone path, and removed via shutil.rmtree() when force_remove is used. Exploitation requires an application flow that updates submodules at a non-HEAD commit (such as a historical-commit API); the common clone-then-update flow re-derives the path from a canonical tree lookup and is not affected. The issue is fixed in GitPython 3.1.62.

lmfit asteval

ASTEVAL is an evaluator of Python expressions and statements. Prior to 1.0.9, FROM_PY in asteval/astutils.py exposes BaseException, SystemExit, KeyboardInterrupt, and GeneratorExit to expressions evaluated by asteval.Interpreter.eval(), while run() and eval() in asteval/asteval.py catch Exception rather than these non-Exception BaseException subclasses. When an attacker-controlled expression raises one of these classes, on_raise() passes the class to raise_exception(), and the resulting exception bypasses the interpreter's safety handlers and propagates into the calling application. A consuming service that evaluates untrusted expressions can therefore be terminated or have signal and cleanup handling disrupted, causing denial of service. The separately documented read-only open() capability is not part of this vulnerability. This issue is fixed in version 1.0.9.

facelessuser soupsieve

Soup Sieve is a CSS selector library designed to be used with Beautiful Soup 4. Prior to 2.9, the selector parser in src/soupsieve/css_parser.py defines IDENTIFIER with adjacent quantified groups over overlapping character classes, and VALUE embeds IDENTIFIER for attribute selectors. When an attacker-controlled selector contains a long identifier or unquoted attribute-value run followed by input that makes the overall match fail, the regular expression engine explores quadratically many splits between the overlapping groups. User-controlled selectors can reach this path through soupsieve.compile(), soupsieve.select(), or BeautifulSoup.select(), while applications using only hard-coded selectors are unaffected. The resulting CPU consumption can hold the Python GIL, exhaust application workers, and stall a service; successful plain identifier matches are linear, and the issue does not cause memory corruption or code execution. The issue is fixed in version 2.9.

ronf asyncssh

AsyncSSH is a Python package which provides an asynchronous client and server implementation of the SSHv2 protocol on top of the Python asyncio framework. Prior to 2.24.0, _process_channel_open and _process_channel_open_confirmation in asyncssh/connection.py accept a peer-supplied send_pktsize value of zero. When channel data reaches SSHChannel._flush_send_buf in asyncssh/channel.py, the zero value causes each loop iteration to slice and remove zero bytes without reducing the send window, leaving the synchronous loop permanently true with no await point. A malicious SSH server can trigger the client path through SSH_MSG_CHANNEL_OPEN_CONFIRMATION before the first channel write, while an authenticated client can trigger the server path through SSH_MSG_CHANNEL_OPEN and freeze every current and future connection handled by the process. This vulnerability is fixed in 2.24.0.

zopefoundation AccessControl

Zope AccessControl provides a general security framework for use in Zope. Prior to 7.4, applications that allow untrusted users to create and execute AccessControl-controlled Python code do not safely guard str.format and str.format_map when those methods are reached through a str subclass. In both ImplPython.py and cAccessControl.c, Python formatting can recursively access attributes and subscriptions using unrestricted getattr and getitem behavior instead of the policy-restricted getattr and getitem operations. A controlled format string can therefore disclose objects reachable from values available to the formatting operation. This issue is fixed in version 7.4.

universal-tool-calling-protocol python-utcp

utcp-gql before 1.1.1 and utcp-websocket before 1.1.1 contain server-side request forgery vulnerabilities due to incomplete application of CVE-2026-44661 fixes. The GraphQL plugin uses a vulnerable prefix check allowing bypass URLs like http://127.0.0.1.attacker.example, while the WebSocket plugin performs no URL validation despite documented security requirements. Attackers can force connections to internal services and cloud metadata endpoints by supplying malicious tool URLs in call templates, and receive configured API keys and OAuth tokens sent to attacker-controlled hosts.

python-hyper hpack

hpack is an HTTP/2 Header Encoding for Python. Prior to version 4.2.0, unbounded variable integer decoding can cause run-away computation on malformed input leading to O(n^2) runtime, effectively blocking further processing with large enough unsanitized input. A fix is available in python-hyper/hpack v4.2.0 to restricted variable integer decoding to uint32 to prevent run-away computation. As a workaround, sanitize input to hpack decoder for long sequences of `0xFF` values to prevent malicious use.

mesop-dev mesop

Mesop is a Python-based UI framework that allows users to build web applications. Prior to 1.3.4, the unauthenticated /__csp__ endpoint passes attacker-controlled document-uri, blocked-uri, and violated-directive values to the csp_report handler in mesop/server/static_file_serving.py, which prints them to standard output without neutralizing terminal control sequences. When an operator views the resulting logs in an ANSI-capable terminal, injected ANSI or VT100 sequences can clear or reposition the display, hide text, or present forged messages, reducing the integrity of monitoring and incident-response output. This issue is fixed in version 1.3.4.

kap-sh zapros

Zapros, a Python HTTP client, prior to version 0.14.0 is vulnerable to denial of service when an application requests content from an untrusted server, or follows a redirect to one, because a malicious response containing an excessive number of chained `Content-Encoding` values causes Zapros to construct a deeply nested decompression chain that consumes excessive resources. Version 0.14.0 patches the vulnerability by limiting responses to five content-encoding layers and raising `DecodingError` when that limit is exceeded. As a workaround, applications can add response middleware that inspects the `Content-Encoding` header and rejects responses containing more than a safe number of encoding layers.

sanic-org sanic

Sanic is an opensource python web server/framework. In version 25.12.0, Sanic's core HTTP/1.1 chunked-body handling does not fully consume the trailer-part after the terminating zero chunk before reusing the keep-alive connection buffer. A remote unauthenticated client can place attacker-controlled bytes in that trailer region, causing Sanic to parse and route them as a hidden second request after the outer request. This breaks HTTP request-boundary integrity and can provide a request-smuggling primitive when Sanic is deployed behind intermediaries. This issue is fixed in version 25.12.1.

python-social-auth social-core

Python Social Auth is a social authentication/registration mechanism. Prior to version 5.0.0, the partial-pipeline resume mechanism accepted `partial_token` as a bearer credential without binding it to the browser session that created it. Applications using resumable partial pipeline steps could allow an attacker to start an authentication flow, obtain a valid partial token and verification data, and cause a victim's browser to resume that attacker-controlled flow. This could authenticate the victim's browser as the attacker's account. The issue affects applications using partial pipeline steps such as `mail_validation` or custom steps decorated with `@partial`. The issue has been fixed in version 5.0.0 by binding partial pipeline resumes to the originating browser session.

heymrun heym

Heym versions 0.0.90 and earlier contain two server-side request forgery (SSRF) egress gaps, both remediated in app/services/ssrf_guard.py in 0.0.91. First, the LLM image-edit input loader (_load_image_bytes) fetched caller-controlled HTTP/HTTPS URLs with a bare httpx.get, applying only a scheme check and bypassing the egress-pinning HTTP client; because the workflow DSL supports "imageInput": "$userInput.body.imageUrl", a webhook or API caller can choose the fetch target when a workflow author uses that expression, allowing requests to loopback, RFC1918, and cloud metadata endpoints. Second, _is_public_address unwrapped only IPv4-mapped IPv6 addresses, so IPv6 transition forms — the NAT64 well-known prefix 64:ff9b::/96, deprecated IPv4-compatible ::x.x.x.x addresses, and 6to4 (2002::/16, classified as globally routable by Python 3.11.0 through 3.11.9) — could carry loopback, RFC1918, link-local, or cloud-metadata IPv4 destinations past both the initial URL validation and the dial-time IP pin. Version 0.0.91 routes the image loader through guard_http_url and the guarded client, evaluates NAT64 and IPv4-compatible addresses by their embedded IPv4 address, and refuses 64:ff9b:1::/48, 6to4, and Teredo (2001::/32) outright.