agentbox

Agentbox - Coding Agent Sandbox There's no perfect way to sandbox agents yet, but containers are a practical start. Agentbox is a Docker-based coding agent sandbox, originally inspired by Batrachian Toad, and now generalized to more tools. It provides an isolated environment that works well with...

cve-bench

CVE-Bench A benchmark for evaluating LLM agents on fixing real-world security vulnerabilities. Agents run inside sandboxed Docker containers and are scored against the maintainer's security test suite. Requirements Python 3.12+ Docker OPENAIAPIKEY, ANTHROPICAPIKEY, and/or POOLSIDEAPIKEY in your...

readme2demo

readme2demo — verified tutorials & demo videos from your README ▶ readme2demo generating its own tutorial: an AI agent runs this repo's README in a sandbox, a fresh container replays every step, then the demo is rendered. Full self-run output in examples/readme2demo · run against another project ...

poodle-attack-sandbox

poodle-attack-sandbox Overview This code is the test code for POODLE CVE-2014-3566. POODLE stands for "Padding Oracle On Downgraded Legacy Encryption" and is a vulnerability in SSLv3.0. Warning If you use this code to carry out an attack, you may be arrested by the judicial authorities of your...

pentagi

PentAGI P enetration testing A rtificial G eneral I ntelligence Join the Community! Connect with security researchers, AI enthusiasts, and fellow ethical hackers. Get support, share insights, and stay updated with the latest PentAGI developments. ⠀ Table of Contents Overview Features Architecture...

kurukshetra

Kurukshetra Kurukshetra est un framework web développé dans le but d'être le premier framework open source à fournir une base solide pour héberger des défis de codage sécurisé raisonnablement complexes, tout en offrant la possibilité d'exécuter chaque défi de manière efficace et dynamique en...

livewire-honeypot

Livewire Honeypot A high-interaction honeypot that masquerades as a vulnerable Laravel/Livewire application. It captures exploit attempts targeting known Livewire CVEs, stores uploaded malicious files webshells and remote code execution RCE payloads with SHA-256 deduplication, and optionally...

CVE-2026-72713

XAgent contains a path traversal vulnerability in the workspace file endpoint that allows self-registered or default-credential users to read arbitrary files on the host by supplying parent-directory segments in the filename form field with no path containment check. Attackers can register an...

EUVD-2026-56851

XAgent contains a path traversal vulnerability in the workspace file endpoint that allows self-registered or default-credential users to read arbitrary files on the host by supplying parent-directory segments in the filename form field with no path containment check. Attackers can register an...

CVE-2026-72713 XAgent Path Traversal Arbitrary File Read via /workspace/file

XAgent contains a path traversal vulnerability in the workspace file endpoint that allows self-registered or default-credential users to read arbitrary files on the host by supplying parent-directory segments in the filename form field with no path containment check. Attackers can register an...

CVE-2026-72713 XAgent Path Traversal Arbitrary File Read via /workspace/file

XAgent contains a path traversal vulnerability in the workspace file endpoint that allows self-registered or default-credential users to read arbitrary files on the host by supplying parent-directory segments in the filename form field with no path containment check. Attackers can register an...

PT-2026-70827

XAgent contains a path traversal vulnerability in the workspace file endpoint that allows self-registered or default-credential users to read arbitrary files on the host by supplying parent-directory segments in the file name form field with no path containment check. Attackers can register an...

CVE-2026-72713: Improper Limitation of a Pathname to a Restricted Directory

XAgent contains a path traversal vulnerability in the workspace file endpoint that allows self-registered or default-credential users to read arbitrary files on the host by supplying parent-directory segments in the filename form field with no path containment check. Attackers can register an...

CVE-2026-14891 Nomad vulnerable to sandbox escape in Docker task driver

HashiCorp Nomad and Nomad Enterprise are vulnerable to a sandbox escape in the Docker task driver that may allow a job submitter to bind-mount a host path into a container even when volume bind mounts are disabled, potentially leading to reading and writing files on the host. This vulnerability,...

CVE-2026-12539

Docker Sandboxes sbx blocks ICMP egress with an authorizer applied only at network-creation time, and does not re-apply it to networks rebuilt from disk when the Docker daemon restarts, so a restart-surviving sandbox forwards ICMP to arbitrary hosts. A workload inside a sandbox, which the threat...

CVE-2026-12539 Docker Sandboxes ICMP egress restriction bypass after daemon restart

Docker Sandboxes sbx blocks ICMP egress with an authorizer applied only at network-creation time, and does not re-apply it to networks rebuilt from disk when the Docker daemon restarts, so a restart-surviving sandbox forwards ICMP to arbitrary hosts. A workload inside a sandbox, which the threat...

CVE-2026-12539 Docker Sandboxes ICMP egress restriction bypass after daemon restart

Docker Sandboxes sbx blocks ICMP egress with an authorizer applied only at network-creation time, and does not re-apply it to networks rebuilt from disk when the Docker daemon restarts, so a restart-surviving sandbox forwards ICMP to arbitrary hosts. A workload inside a sandbox, which the threat...

MalSkillBench: A Runtime-Verified Benchmark of Malicious Agent Skills

AI coding agents such as Claude Code and Gemini CLI increasingly extend themselves with third-party skills: markdown packages bundling natural-language instructions, executable scripts, and tool permissions. Because a skill is at once code and agent-facing instruction, it introduces a supply chai...

vulnhunt-agent

Vulnerability Hunting Agent An LLM agent that reads code,...

OpenClaw Encryption Problem Vulnerability

OpenClaw is openclaw open source an intelligent artificial assistant. OpenClaw suffers from a cryptographic issue vulnerability that stems from the use of SHA-1 hashed Docker and browser sandbox configurations of the sandbox identifier cache key, which can be exploited by an attacker to enable...