CVE-2026-55605

Summary: CVE-2026-55605 affects the self-hosted HTTP transport of @arikusi/deepseek-mcp-server used with DeepSeek V4. starting in 1.4.2 and before 1.8.0, the POST /mcp endpoint is exposed without authentication due to createMcpExpressApp being called without an authProvider and missing route guar...

PT-2026-52169

Name of the Vulnerable Software and Affected Versions Drupal AI Agents versions 0.0.0 through 1.1.4 Drupal AI Agents versions 1.2.0 through 1.2.5 Drupal AI Agents versions 1.3.0 through 1.3.1 Description Missing authorization allows forceful browsing within the module that provides the entity typ...

AI Agents - Less critical - Access bypass - SA-CONTRIB-2026-056

This module provides the entity type and runtime for Drupal AI Agents, enabling agents to use tools. The module does not sufficiently check the required permissions when a tool loads content entities. This vulnerability is mitigated by the fact that an agent must be configured to use the affected...

EUVD-2026-30652

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.6, there is a vulnerability in chat completion API, which allows attackers to bypass tool restrictions, potentially enabling unauthorized actions or access. In the chatcompletion API, t...

SIR-Bench: Evaluating Investigation Depth in Security Incident Response Agents

We present SIR-Bench, a benchmark of 794 test cases for evaluating autonomous security incident response agents that distinguishes genuine forensic investigation from alert parroting. Derived from 129 anonymized incident patterns with expert-validated ground truth, SIR-Bench measures not only...

pentesting-writeups

🔐 Pentesting Writeups Personal penetration testing document...

pentesting-notes

🔐 Pentesting Notes Personal penetration testing documentati...

A Trajectory-Based Safety Audit of Clawdbot (OpenClaw)

Clawdbot is a self-hosted, tool-using personal AI agent with a broad action space spanning local execution and web-mediated workflows, which raises heightened safety and security concerns under ambiguity and adversarial steering. We present a trajectory-centric evaluation of Clawdbot across six...

Bypassing AI Control Protocols Via Agent-As-A-Proxy Attacks

As AI agents automate critical workloads, they remain vulnerable to indirect prompt injection IPI attacks. Current defenses rely on monitoring protocols that jointly evaluate an agent's Chain-of-Thought CoT and tool-use actions to ensure alignment with user intent. We demonstrate that these...

Gitbook

Personal Pentesting Knowledge Base 👋 Welcome! This reposit...

Agent Safety Alignment Via Reinforcement Learning

The emergence of autonomous Large Language Model LLM agents capable of tool usage has introduced new safety risks that go beyond traditional conversational misuse. These agents, empowered to execute external functions, are vulnerable to both user-initiated threats e.g., adversarial prompts and...

IP Leakage Attacks Targeting LLM-Based Multi-Agent Systems

The rapid advancement of Large Language Models LLMs has led to the emergence of Multi-Agent Systems MAS to perform complex tasks through collaboration. However, the intricate nature of MAS, including their architecture and agent interactions, raises significant concerns regarding intellectual...

Google Go 安全漏洞

Google Go is a static strongly typed, compiled, concatenated, and garbage-collected programming language from Google USA. A security vulnerability exists in Google Go that stems from the use of Apple's ld with special values in the cgo LDFLAGS instruction on Darwin systems, which could trigger...

SUSE CVE-2024-53863

Synapse is an open-source Matrix homeserver. In Synapse versions before 1.120.1, enabling the dynamicthumbnails option or processing a specially crafted request could trigger the decoding and thumbnail generation of uncommon image formats, potentially invoking external tools like Ghostscript for...

UBUNTU-CVE-2024-53863

Synapse is an open-source Matrix homeserver. In Synapse versions before 1.120.1, enabling the dynamicthumbnails option or processing a specially crafted request could trigger the decoding and thumbnail generation of uncommon image formats, potentially invoking external tools like Ghostscript for...

O365Spray - Username Enumeration And Password Spraying Tool Aimed At Microsoft O365

For educational, authorized and/or research purposes only. o365spray a username enumeration and password spraying tool aimed at Microsoft Office 365 O365. This tool reimplements a collection of enumeration and spray techniques researched and identified by those mentioned in Acknowledgments...

Waybackurls - Fetch All The URLs That The Wayback Machine Knows About For A Domain

Accept line-delimited domains on stdin, fetch known URLs from the Wayback Machine for .domain and output them on stdout. Usage example: ▶ cat domains.txt | waybackurls urls Install: ▶ go get github.com/tomnomnom/waybackurls Credit This tool was inspired by @mhmdiaa's waybackurls.py script. Thanks...

Exploit for Server-Side Request Forgery in Microsoft

CVE-2021-26855-PoC PoC exploit code for CVE-2021-26855. Orig...

APT-Hunter - Threat Hunting Tool For Windows Event Logs Which Made By Purple Team Mindset To Provide Detect APT Movements Hidden In The Sea Of Windows Event Logs To Decrease The Time To Uncover Suspicious Activity

APT-Hunter is Threat Hunting tool for windows event logs which made by purple team mindset to provide detect APT movements hidden in the sea of windows event logs to decrease the time to uncover suspicious activity . this tool will make a good use of the windows event logs collected and make sure...

Advanced Discovery of Privileged Accounts: ACLight

ACLight is a tool for discovering privileged accounts through advanced ACLs Access Lists analysis. It includes the discovery of Shadow Admins in the scanned network. The tool queries the Active Directory AD for its objects’ ACLs and then filters and analyzes the sensitive permissions of each one...