Critical · CVSS 10EPSS 1.47%Public exploitFix available
Decepticon: Role-boundary forgery via ChatML special-token literals in web crawl output composed into LLM context
Decepticon is an autonomous hacking agent for red teams. Versions prior to 1.1.17 wrap web crawl results — the output of agent reconnaissance against target services — into LLM messages without neutralizing ChatML special-token literals. Under the BYOK (Bring Your Own Key) deployment model, users configure their own LLM credentials to any OpenAI-compatible endpoint. Most open-source and self-deployed model providers (vLLM, SGLang, Ollama, LM Studio, text-generation-webui, etc.) do not filter special-token literals from user content in their default configurations. Those literals are parsed into structural role-boundary token IDs, meaning an a
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 decora
social-auth-core: VK App backend accepts unsigned callback data when auth_key is missing
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 pr
social-auth-core has Login CSRF via Missing State Parameter in LoginRadius Backend
Python Social Auth is a social authentication/registration mechanism. Prior to version 5.0.0, the LoginRadius backend did not validate OAuth state during the authentication flow. Applications using this backend were vulnerable to login CSRF. An attacker could cause a victim's browser session to complete authentication using an attacker-controlled LoginRadius token, making the victim authenticated as the attacker's LoginRadius identity. The issue affects only applications using the LoginRadius backend. The issue has been fixe in version 5.0.0 by enabling callback state validation for the LoginRadius backend.
social-auth-core Vulnerable to Account Takeover via Identity Binding Flaw in Vend Backend
Python Social Auth is a social authentication/registration mechanism. Prior to version 5.0.0, the Vend OAuth2 backend used only the numeric Vend user_id as the social-auth UID. When multiple Vend shops authenticate through the same application, users from different shops with the same internal Vend user ID could collide in the social-auth association table. A user from one shop could then be authenticated as the local account previously associated with the same numeric user ID from another shop. The issue affects applications using the Vend OAuth2 backend with more than one Vend shop. Version 5.0.0 patches the issue.
social-auth-core has an Improper Authentication issue
Python Social Auth is a social authentication/registration mechanism. Prior to version 5.0.0, the SAML backend accepted SAML responses on the Assertion Consumer Service endpoint without verifying that they matched a previously issued `AuthnRequest`. Applications using SAML account association could allow an attacker with a valid account on a trusted IdP to link the attacker's SAML identity to a logged-in victim's local account. The attacker could then authenticate through SAML and gain access to the victim's account. The issue affects applications using the SAML backend together with authenticated account association. The issue has been fixed
hpack: Unbounded variable integer decoding can cause run-away computation on malformed input
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.
plone.app.contenttypes has a Denial of Service in File Upload due to excessive filename length
### Impact
When creating an Image or File with a very large filename, Plone can become unresponsive. Even if the content loads, it can be hard to edit or delete, because the UI is unwieldy.
### Patches
The problem has been patched in `plone.app.contenttypes`.
* On Plone 6.2, upgrade to `plone.app.contenttypes ` 5.0.1.
* On Plone 6.1, upgrade to `plone.app.contenttypes ` 4.0.10.
* On Plone 6.0, upgrade to `plone.app.contenttypes ` 3.0.12
Another part of this problem has been patched in `plone.app.dexterity`. See [this advisory](https://github.com/plone/plone.app.dexterity/security/advisories/GHSA-5426-92w4-wvhv).
### Workarounds
There are
MemoryOS 2.0.34 was published with a credential-stealing binary
An attacker with write access to the GitHub repository pushed malicious
commits and tagged v2.0.34, and the project's own GitHub Actions release
workflow built and uploaded 2.0.34 to PyPI.
Importing the package runs memos/_stage0.py, which launches
a bundled sckit binary that collects credentials
(.pypirc, .npmrc, .git-credentials, SSH keys, token-like environment variables)
and sends them to *.skyleen[.]fr.
Remove 2.0.34 and rotate any credentials reachable from affected machines.
- wheel SHA256: 39ee644406829a4b630b31759c20478bc22d576d6a59b253ed86f72c360aa5ef
- sdist SHA256: 92b46d18fc553c494eda714f204459edb74c205bf53b18a9092bcf02c7a6c5be
Medium · CVSS 5.3EPSS 0.53%Public exploitFix available
Mesop: Unauthenticated ANSI Escape Sequence Injection in CSP Reporting Endpoint
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
Streamlink: HTTPSession follows HTTP redirects into file:// URLs, reading local files
Streamlink is a CLI utility which pipes video streams from various services into a video player. Prior to 8.6.0, HTTPSession mounts a FileAdapter for the file scheme and inherits redirect handling from requests.Session without rejecting cross-protocol redirects. A remote server controlling an HTTP or HTTPS URL reached by Streamlink can return a redirect to a local file URL, causing HTTPSession to read the local file and return its contents to the response consumer. This bypasses the direct file URL checks added for HLS and DASH content because the manifest contains an ordinary network URL and the scheme transition occurs later during fetch ha
plone.app.dexterity and plone.app.contenttypes have a Denial of Service due to excessive title or description length
plone.app.dexterity is a content-type system for the Plone content management system, and plone.app.contenttypes provides Plone’s Dexterity-based content types. Plone.app.dexterity versions through 3.2.2, 4.0.0 through 4.1.2, and 5.0.0, and plone.app.contenttypes versions through 3.0.11, 4.0.0 through 4.0.9, and 5.0.0 are vulnerable to denial of service because an authenticated user can create content with excessively long titles, descriptions, or uploaded-file names, causing Plone to become unresponsive and potentially making the resulting content difficult to edit or delete. The vulnerability is patched in plone.app.dexterity versions 3.2.3
wlc may disclose API tokens to project-configured URLs
wlc is a Weblate command-line client using Weblate's REST API. Prior to 2.0.1, automatically discovered configuration from .weblate, .weblate.ini, or weblate.ini can select the API URL while an unscoped API token is supplied through WLC_KEY or --key without a matching WLC_URL or --url. When wlc runs in an untrusted repository, pull request checkout, or directory with untrusted ancestor configuration, it can send the token to an attacker-controlled project-configured URL. URL-scoped keys in [keys] are not affected. This issue is fixed in version 2.0.1.
MPXJ is an open source library to read and write project plans from a variety of file formats and databases. From 5.5.5 until 16.4.1, MerlinReader creates a DocumentBuilder with default settings while parsing XML from the ZTIMEINTERVALS column of a Merlin project SQLite database, leaving doctype declarations and external entities enabled. A crafted database can cause the parser to read an arbitrary local file, although MPXJ's subsequent handling of the parsed XML makes disclosure of the file contents unlikely. This issue is fixed in version 16.4.1.
MPXJ: Potential Path Traversal Vulnerability in Primavera P3 PRX and SureTrak STX readers
MPXJ is an open source library to read and write project plans from a variety of file formats and databases. From 7.3.0 until 16.5.0, reading a suitably crafted Primavera P3 PRX or SureTrak STX file can cause MPXJ to write files to arbitrary locations in the filesystem. This issue is fixed in version 16.5.0.
Nautobot: Authorization bypass in approval workflow REST API allows self-approval and unauthorized activation of scheduled jobs
Nautobot is a Network Source of Truth and Network Automation Platform. From 3.0.0 until 3.1.8, the generic ApprovalWorkflowStageResponse create endpoint does not enforce approver-group membership, change permission on the object under review, or the one-response-per-user restriction applied by the intended approve and deny actions. A user with only extras.add_approvalworkflowstageresponse can submit approved responses directly, while writable user and state fields permit responses to be attributed to arbitrary users. These forged responses can satisfy min_approvers, approve the workflow, and activate its gated ScheduledJob without a legitimat
Nautobot: Stored cross-site scripting (XSS) in object create/edit form help text
Nautobot is a Network Source of Truth and Network Automation Platform. Prior to 2.4.37 and 3.1.8, a user with extras.add_relationship or extras.change_relationship permission can store HTML or JavaScript in a Relationship description, and a user with dcim.add_modulefamily or dcim.change_modulefamily permission can store it in a Module Family name. Nautobot assigns these values to form field help_text rendered by render_field.html through Django's |safe filter without adequate neutralization. The stored content executes in the authenticated browser session of any user, including an administrator or superuser, who opens an affected create or ed
psd-tools: Uncontrolled memory allocation in psd-tools composite/numpy via crafted PSD geometry
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.
Home Assistant is open source home automation software focused on local control and privacy. Prior to 2026.2.3, the IPP integration automatically processed unauthenticated _ipp._tcp.local mDNS announcements in homeassistant/components/ipp/config_flow.py, where async_step_zeroconf passed attacker-controlled host, port, and base_path values to validate_input for printer metadata retrieval. Because the shared HTTP client followed attacker-controlled cross-origin redirects without blocking loopback targets, a local-network attacker could redirect the request to 127.0.0.1 or another internal service without user interaction or prior IPP configurat
Home Assistant is open source home automation software focused on local control and privacy. Prior to 2026.7.0, the Statistics Graph card in src/components/chart/statistics-chart.ts passed entity names through getStatisticLabel and computeStateName and interpolated param.seriesName into ECharts tooltip HTML without escaping. An authenticated user or an integration that supplies a malicious default entity name could cause script-related HTML to execute when a viewer hovered over a data point. Mean, State, Sum, and Change fields in the default Line chart configuration were affected, while Bar charts were not. This issue is fixed in version 2026