CVE-2026-102437

OS Command Injection in internal/gitcmd git diff filter.clean/smudge invocation in esengine DeepSeek-Reasonix Reasonix Studio allows a local attacker who controls repository content .gitattributes + .git/config to execute arbitrary commands via the desktop app's workspace-changes diff viewer...

CVE-2026-102437

The esengine DeepSeek-Reasonix (Reasonix Studio) desktop application contains an OS Command Injection vulnerability in the internal/gitcmd component. The root cause is the improper neutralization of special elements during the invocation of git diff filter.clean/smudge . A local attacker who can ...

CVE-2026-102437 Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') in DeepSeek-Reasonix

OS Command Injection in internal/gitcmd git diff filter.clean/smudge invocation in esengine DeepSeek-Reasonix Reasonix Studio allows a local attacker who controls repository content .gitattributes + .git/config to execute arbitrary commands via the desktop app's workspace-changes diff viewer...

EUVD-2026-88756

OS Command Injection in internal/gitcmd git diff filter.clean/smudge invocation in esengine DeepSeek-Reasonix Reasonix Studio allows a local attacker who controls repository content .gitattributes + .git/config to execute arbitrary commands via the desktop app's workspace-changes diff viewer...

Agentic-SOC-Simulation

🛡️ Agentic SOC Parallel Simulation ASPS: Autonomous Security Operations Parallel Simulation Center Redefining the Future of Security Operations: Full-Stack SOC Parallel Simulation, Multi-Agent Collaboration, and Data-Driven Autonomous Defense 🇨🇳 中文文档 Chinese Version This project is a cutting-edge...

CVE-2026-101132

A security flaw has been discovered in DeepSeek deepseek-harness up to 0.1.7-rc.2. The affected element is the function loadProfile of the file packages/boot/app-boot/src/profile.ts of the component Bundle Patch Handler. The manipulation of the argument dsh.bundle.patch results in path traversal...

CVE-2026-101131

A vulnerability was identified in deepseek-ai deepseek-harness up to 0.1.5-rc.3. Impacted is an unknown function of the file packages/e2b/e2b/src/index.ts of the component dsh. The manipulation of the argument E2BAPIKEY leads to reliance on untrusted inputs in a security decision. Local access is...

EUVD-2026-88438

A security flaw has been discovered in DeepSeek deepseek-harness up to 0.1.7-rc.2. The affected element is the function loadProfile of the file packages/boot/app-boot/src/profile.ts of the component Bundle Patch Handler. The manipulation of the argument dsh.bundle.patch results in path traversal...

CVE-2026-101132

A path traversal vulnerability exists in DeepSeek deepseek-harness (versions up to 0.1.7-rc.2 ). The flaw is located within the loadProfile function of the packages/boot/app-boot/src/profile.ts file in the Bundle Patch Handler component. Manipulation of the dsh.bundle.patch argument allows for pa...

EUVD-2026-88404

A vulnerability was identified in deepseek-ai deepseek-harness up to 0.1.5-rc.3. Impacted is an unknown function of the file packages/e2b/e2b/src/index.ts of the component dsh. The manipulation of the argument E2BAPIKEY leads to reliance on untrusted inputs in a security decision. Local access is...

CVE-2026-101131

A vulnerability exists in deepseek-ai deepseek-harness (dsh) up to version 0.1.5-rc.3 . The flaw is located in an unknown function within the file packages/e2b/e2b/src/index.ts , where the manipulation of the E2B_API_KEY argument leads to a reliance on untrusted inputs in a security decision . Th...

CVE-2026-101102

A vulnerability was found in deepseek-ai deepseek-harness up to 0.1.0-rc.7. Impacted is the function runcode of the component Code Mode Sandbox. The manipulation results in sandbox issue. The attack can be executed remotely. The vendor's own code, SAFETY.md, and design notes all explicitly state...

EUVD-2026-88402

A vulnerability was found in deepseek-ai deepseek-harness up to 0.1.0-rc.7. Impacted is the function runcode of the component Code Mode Sandbox. The manipulation results in sandbox issue. The attack can be executed remotely. The vendor's own code, SAFETY.md, and design notes all explicitly state...

CVE-2026-101102

A vulnerability exists in deepseek-ai deepseek-harness (versions up to 0.1.0-rc.7 ) within the Code Mode Sandbox component. Specifically, the run_code function is susceptible to manipulation, leading to a sandbox issue that can be triggered via a remote attack . Notably, the vendor's own document...

CVE-2026-101078

A vulnerability has been found in deepseek-ai deepseek-harness up to 0.1.7-rc.2. Affected is an unknown function of the file packages/sandbox/sandbox-local/src/profiles.ts of the component Landlock Backend. Such manipulation leads to improper isolation or compartmentalization. The attack must be...

CVE-2026-101078

A vulnerability exists in deepseek-ai deepseek-harness versions up to 0.1.7-rc.2 within the Landlock Backend component. The issue stems from an unknown function in the file packages/sandbox/sandbox-local/src/profiles.ts, which can lead to improper isolation or compartmentalization . Exploitation ...

EUVD-2026-88334

A vulnerability has been found in deepseek-ai deepseek-harness up to 0.1.7-rc.2. Affected is an unknown function of the file packages/sandbox/sandbox-local/src/profiles.ts of the component Landlock Backend. Such manipulation leads to improper isolation or compartmentalization. The attack must be...

llm-censorship-steering

LLM Censorship Steering This repository contains code implementation for Steering the CensorShip: Uncovering Representation Vectors for LLM "Thought" Control by Hannah Cyberey and David Evans. We introduce a method that finds "steering vectors" from LLM internals for detecting and controlling the...

AGATE: Provenance-Based Runtime Defense against Compositional Attacks on LLM Agents

LLM agents can produce harmful effects through sequences of ordinary operations. Judging such actions requires establishing both the authority that permits them and the origin of the data they carry. We present AGATE, an authorization and data-provenance gate at instrumented agent-harness...

DeepSeek Harness Flaw Let AI Agents Disable Their Own File Sandbox Without Approval

A flaw in DeepSeek Harness , DeepSeek's open-source tool for running AI coding agents on a developer's machine, let a sandboxed agent turn off its own sandbox with a single command. The tool runs an agent's commands inside an operating-system sandbox, so that an agent working on untrusted files...