Infinite loop

Overview justhtml is an A pure Python HTML5 parser that just works. Affected versions of this package are vulnerable to Infinite loop via custom sanitization policies or programmatic DOM manipulation. An attacker can inject and execute arbitrary scripts, cause resource loading, or trigger externa...

justhtml has sanitization bypass in custom policies and programmatic DOM

Summary justhtml 1.17.0 fixes multiple security issues in sanitization, serialization, and programmatic DOM handling. Most of these issues affected advanced or custom configurations rather than the default safe path. Affected versions - justhtml , MathML , SVG / , and MathML text integration poin...

Uncontrolled Recursion

Overview xmldom is an A pure JavaScript W3C standard-based XML DOM Level 2 Core DOMParser and XMLSerializer module. Affected versions of this package are vulnerable to Uncontrolled Recursion in the recursive processing of deeply nested XML documents by several DOM-related operations, including...

XML Injection

Overview xmldom is an A pure JavaScript W3C standard-based XML DOM Level 2 Core DOMParser and XMLSerializer module. Affected versions of this package are vulnerable to XML Injection via the createProcessingInstruction function. An attacker can inject arbitrary XML nodes into the serialized output...

CVE-2026-3837 Frappe Framework 16.10.0 - Stored DOM XSS in Multiple Field Formatters

An authenticated attacker can persist crafted values in multiple field types and trigger client-side script execution when another user opens the affected document in Desk. The vulnerable formatter implementations interpolate stored values into raw HTML attributes and element content without...

CVE-2026-3673

CVE-2026-3673 affects Frappe Framework 16.10.10. An authenticated attacker can store a crafted value in _user_tags and trigger JavaScript execution when a victim opens the list/report view where tags are rendered. The vulnerable renderer interpolates tag content into HTML attributes and element c...

Discord-Linked Group Accessed Anthropic’s Claude Mythos AI in Vendor Breach

Anthropic is investigating a vendor breach after a Discord-linked group accessed its Claude Mythos AI model, with no evidence of impact on core systems...

GHSA-RXPQ-XGQX-FR7P InstructLab Includes Functionality from Untrusted Control Sphere

A flaw was found in InstructLab. The linuxtrain.py script hardcodes trustremotecode=True when loading models from HuggingFace. This allows a remote attacker to achieve arbitrary Python code execution by convincing a user to run ilab train/download/generate with a specially crafted malicious model...

InstructLab Includes Functionality from Untrusted Control Sphere

A flaw was found in InstructLab. The linuxtrain.py script hardcodes trustremotecode=True when loading models from HuggingFace. This allows a remote attacker to achieve arbitrary Python code execution by convincing a user to run ilab train/download/generate with a specially crafted malicious model...

LLM-and-MCP

Detection and Exploitation of Vulnerabilities in Android Appli...

CVE-2026-6859

A flaw was found in InstructLab. The linuxtrain.py script hardcodes trustremotecode=True when loading models from HuggingFace. This allows a remote attacker to achieve arbitrary Python code execution by convincing a user to run ilab train/download/generate with a specially crafted malicious model...

CVE-2026-6859 Instructlab: instructlab: arbitrary code execution due to hardcoded `trust_remote_code=true`

A flaw was found in InstructLab. The linuxtrain.py script hardcodes trustremotecode=True when loading models from HuggingFace. This allows a remote attacker to achieve arbitrary Python code execution by convincing a user to run ilab train/download/generate with a specially crafted malicious model...

CVE-2026-6859

A flaw was found in InstructLab. The linuxtrain.py script hardcodes trustremotecode=True when loading models from HuggingFace. This allows a remote attacker to achieve arbitrary Python code execution by convincing a user to run ilab train/download/generate with a specially crafted malicious model...

CVE-2026-6859

A flaw was found in InstructLab. The linuxtrain.py script hardcodes trustremotecode=True when loading models from HuggingFace. This allows a remote attacker to achieve arbitrary Python code execution by convincing a user to run ilab train/download/generate with a specially crafted malicious model...

CVE-2026-6859 Instructlab: instructlab: arbitrary code execution due to hardcoded `trust_remote_code=true`

A flaw was found in InstructLab. The linuxtrain.py script hardcodes trustremotecode=True when loading models from HuggingFace. This allows a remote attacker to achieve arbitrary Python code execution by convincing a user to run ilab train/download/generate with a specially crafted malicious model...

Attacking the MCP Trust Boundary

Every secure API draws a line between code and data. HTTP separates headers from bodies. SQL has prepared statements. Even email distinguishes the envelope from the message. The Model Context Protocol MCP, the fast-growing standard for connecting AI agents to external services, inherits that gap...

EUVD-2026-24629

Missing bounds validation for operator could allow out of range operator-code lookup during model loading Affected version is prior to commit 1.30.0...

CVE-2026-6840

Missing bounds validation for operator could allow out of range operator-code lookup during model loading Affected version is prior to commit 1.30.0...

CVE-2026-6840

Missing bounds validation for operator could allow out of range operator-code lookup during model loading Affected version is prior to commit 1.30.0...

CVE-2026-6840

CVE-2026-6840 describes missing bounds validation for an operator during model loading, enabling a out-of-range operator-code lookup. Affected versions are those prior to commit 1.30.0. The CVSS 3.1 base score is 5.5 (Medium) with Local attack vector, Low attack complexity, No privileges required...