Langroid: Sandbox Escape to Remote Code Execution via Incomplete `eval()` Mitigation in TableChatAgent

Advisory Details Title: Sandbox Escape to Remote Code Execution via Incomplete eval Mitigation in TableChatAgent Description: Summary Langroid is vulnerable to a critical Sandbox Escape leading to Remote Code Execution RCE in its TableChatAgent and VectorStore capabilities. When these agents...

CVE-2026-14898

The OpenAI Codex desktop app for macOS rendered remote images from Markdown in model responses. An attacker who could place an indirect prompt injection in content processed by Codex, such as a connected-tool result or another untrusted source, could induce the model to construct a remote image U...

CVE-2026-55514

vLLM is a library for LLM inference and serving. From 0.12.0 to before 0.24.0, sending a pure prompt embeds payload in a /v1/completions request with a model using M-RoPE causes EngineCore to fail an assertion and fatally crash, shutting down the entire server application. Any remote user who is...

CVE-2026-14898

The OpenAI Codex desktop app for macOS rendered remote images from Markdown in model responses. An attacker who could place an indirect prompt injection in content processed by Codex, such as a connected-tool result or another untrusted source, could induce the model to construct a remote image U...

CVE-2026-14898

The CVE affects the OpenAI Codex desktop app for macOS, where rendering remote images from Markdown in model responses can exfiltrate data. An indirect prompt injection in content (e.g., from connected tools or untrusted sources) could cause the model to construct a remote image URL containing se...

EUVD-2026-41915

The OpenAI Codex desktop app for macOS rendered remote images from Markdown in model responses. An attacker who could place an indirect prompt injection in content processed by Codex, such as a connected-tool result or another untrusted source, could induce the model to construct a remote image U...

CVE-2026-14898

The OpenAI Codex desktop app for macOS rendered remote images from Markdown in model responses. An attacker who could place an indirect prompt injection in content processed by Codex, such as a connected-tool result or another untrusted source, could induce the model to construct a remote image U...

PT-2026-56085

Name of the Vulnerable Software and Affected Versions Langroid versions prior to 0.65.5 Description Neo4jChatAgent passes LLM-generated Cypher queries directly to the Neo4j driver without validation, a statement-type allowlist, or an opt-out gate. Because the query text can be influenced by promp...

PT-2026-55971

Name of the Vulnerable Software and Affected Versions OpenAI Codex for macOS affected versions not specified Description The desktop application renders remote images from Markdown within model responses. An attacker can use indirect prompt injection—a technique where malicious instructions are...

CVE-2026-13341

A vulnerability exists in the Kong Konnect Model Context Protocol MCP server prior to version 1.0.0, which could allow a remote attacker to perform an indirect prompt injection attack and execute unintended API requests...

CVE-2026-13341 Prompt Injection and Credential Exposure via Untrusted Analytics Data in Kong Konnect MCP

A vulnerability exists in the Kong Konnect Model Context Protocol MCP server prior to version 1.0.0, which could allow a remote attacker to perform an indirect prompt injection attack and execute unintended API requests...

EUVD-2026-41530

A vulnerability exists in the Kong Konnect Model Context Protocol MCP server prior to version 1.0.0, which could allow a remote attacker to perform an indirect prompt injection attack and execute unintended API requests...

CVE-2026-13341

Kong Konnect MCP server (before 1.0.0) is affected. A remote attacker could perform an indirect prompt injection and cause unintended API requests due to the MCP component. Impact aligns with high-severity potential exposure (CVSS 7.4); exploit details are not provided in the sources. Remediation...

CVE-2026-13341 Prompt Injection and Credential Exposure via Untrusted Analytics Data in Kong Konnect MCP

A vulnerability exists in the Kong Konnect Model Context Protocol MCP server prior to version 1.0.0, which could allow a remote attacker to perform an indirect prompt injection attack and execute unintended API requests...

CVE-2026-13341 Prompt Injection and Credential Exposure via Untrusted Analytics Data in Kong Konnect MCP

A vulnerability exists in the Kong Konnect Model Context Protocol MCP server prior to version 1.0.0, which could allow a remote attacker to perform an indirect prompt injection attack and execute unintended API requests...

CVE-2026-13341

A vulnerability exists in the Kong Konnect Model Context Protocol MCP server prior to version 1.0.0, which could allow a remote attacker to perform an indirect prompt injection attack and execute unintended API requests...

PT-2026-55525

Name of the Vulnerable Software and Affected Versions Kong Konnect MCP server versions prior to 1.0.0 Description An issue exists where the server fails to properly validate content returned to the Large Language Model LLM. This allows a remote attacker to perform an indirect prompt injection by...

Anthropic Claude Code 2.1.38 < 2.1.163 Sandbox Escape (CVE-2026-55607)

The version of Anthropic Claude Code installed on the remote host is 2.1.38 prior to 2.1.163. It is, therefore, affected by a sandbox escape vulnerability: - Claude Code's worktree handling allowed creation of worktrees named '.git' and navigation to worktrees outside the sandbox context, enablin...

SQL Injection

Overview langroid is a Harness LLMs with Multi-Agent Programming Affected versions of this package are vulnerable to SQL Injection via insufficient filtering in the validatequery process. An attacker can access arbitrary files on the database host and perform filesystem reconnaissance by injectin...

Critical Cursor Flaws Could Let Prompt Injection Escape Sandbox and Run Commands

Two flaws in Cursor, an AI code editor, could let a single, ordinary-looking prompt break out of the editor's safety sandbox and run any command on a developer's computer. There is no click to fall for and no approval box to ignore. Cato AI Labs found the pair and named them DuneSlide. They are...