third-thoughts

Third Thoughts A Lightless Labs research project studying AI agent behavior at scale through multi-disciplinary corpus analysis basically, throwing stuff at the wall and seeing what sticks. Named after Tiffany Aching's concept from Discworld: first thoughts agents thinking, second thoughts agents...

SysTrace

SysTrace 2.0 ptrace-based Linux syscall monitor and sandbox for analyzing ELF binaries. SysTrace runs a target inside an isolated Linux namespace environment, traces its system calls and child processes, applies rule-based detection and weighted risk scoring, then uses a Random Forest classifier ...

PowerShellProfiler

PowerShellProfiler This project code is from https://github.com/pan-unit42/publictools/tree/master/powershellprofiler Introductation PowerShellProfiler.py is a script for statically analyzing PowerShell scripts by de-obfuscating and normalizing the content which is then profiled for behavioral...

nginx-rift-detect

nginx-rift-detect Behavioral detection script for CVE-2026-42945 NGINX Rift — a heap buffer overflow in ngxhttprewritemodule present in nginx since 2008. ⚠️ Disclaimer This tool is intended solely for authorized security testing and research. Only run this against systems you own or have explicit...

vi-bot

BotDetect v2 — Production Bot Detection Library Client-side bot and automation detection library with weighted scoring, behavioral analysis, browser fingerprinting, and configurable thresholds. Detects headless browsers, Selenium, Puppeteer, Playwright, CDP-based tools, and stealth automation...

CVE-2026-108789

A vulnerability was identified in deepseek-ai deepseek-harness up to 0.1.1-rc.2. This affects the function agent of the component dsh-workflow-worker-thread. The manipulation leads to enforcement of behavioral workflow. The attack is possible to be carried out remotely. The exploit is publicly...

CVE-2026-108789

A vulnerability exists in deepseek-ai deepseek-harness up to version 0.1.1-rc.2 . The flaw specifically affects the agent function within the dsh-workflow-worker-thread component. The root cause involves improper enforcement of behavioral workflow and business logic errors (CWE-841, CWE-840), whi...

CVE-2026-108789 deepseek-ai deepseek-harness dsh-workflow-worker-thread agent behavioral workflow

A vulnerability was identified in deepseek-ai deepseek-harness up to 0.1.1-rc.2. This affects the function agent of the component dsh-workflow-worker-thread. The manipulation leads to enforcement of behavioral workflow. The attack is possible to be carried out remotely. The exploit is publicly...

EUVD-2026-97158

A vulnerability was identified in deepseek-ai deepseek-harness up to 0.1.1-rc.2. This affects the function agent of the component dsh-workflow-worker-thread. The manipulation leads to enforcement of behavioral workflow. The attack is possible to be carried out remotely. The exploit is publicly...

CVE-2026-108789 deepseek-ai deepseek-harness dsh-workflow-worker-thread agent behavioral workflow

A vulnerability was identified in deepseek-ai deepseek-harness up to 0.1.1-rc.2. This affects the function agent of the component dsh-workflow-worker-thread. The manipulation leads to enforcement of behavioral workflow. The attack is possible to be carried out remotely. The exploit is publicly...

EmailXpose

EmailXpose Specification AI-Powered Email Clarity HTML Mirror: https://roxanneardary.com/emailxpose-specification/ Specification Overview EmailXpose is an open source, modular AI email security and intelligence system designed to analyze email content, sender identity, links, attachments, images,...

dynmx

dynmx Prototipo dynmx pronunciato dinamiche è un approccio di rilevamento basato su firme per caratteristiche comportamentali di malware basate su sequenze di chiamate API di Windows. In modo semplificato, puoi pensare a dynmx come una sorta di YARA per tracce di chiamate API i cosiddetti functio...

VISION-ProcMon

About VISION-ProcMon is a tool dedicated for malware analysis and specifically when performing behavioral analysis. VISION-ProcMon is using a combination of the rust and web technologies in order to provide a fast and better visualization of the behavior of a malware after capturing its activity...

windows-malware-behavioral-analysis

Analisi Comportamentale del Malware con REMnux & FLARE-VM Un'indagine pratica di malware per Blue Team che dimostra come eseguibili Windows sospetti possano essere analizzati in sicurezza all'interno di un laboratorio isolato utilizzando tecniche di analisi comportamentale. Piuttosto che invertir...

SkillPoison: Progressive Skill Poisoning Via Successful Experiences

Self-improving LLM agents increasingly distill successful experiences into persistent, reusable skills. Existing skill attack methods corrupt this learning pipeline by injecting malicious triggers, behaviors, or false facts into individual experiences or extracted skills. However, such attacks ar...

Evaluating Behavioral Context for Interpretable IAM Policy Risk Scoring in Cloud Environments

IAM policy analysis typically emphasizes the authorization capabilities encoded in a policy, but security analyst review priority may also depend on the behavioral and environmental context surrounding a policy event. This paper evaluates whether contextual information provides measurable...

The Fine Art of Frustrating the Adversary

For Cybersecurity Awareness Month, eight Cisco Talos researchers share practical ways defenders can frustrate adversaries at different stages of an operation. Deception techniques such as honeypot accounts, false infrastructure, and tarpits can slow adversaries down while giving defenders earlier...

ActBench

ActBench ActBench is a self-evolving benchmark of behavioral safety in cowork agents. It defines behavioral safety as whether an agent's execution remains within the permissions and state changes required by a benign task, and evaluates realized behavioral risk from execution trajectories rather...

Mikrotik RouterOS Improper Enforcement of Behavioral Workflow Vulnerability

Mikrotik RouterOS contains an improper enforcement of behavioral workflow vulnerability that could allow an unauthenticated client to open a session channel and send an exec request. This vulnerability can be chained to achieve unauthenticated exploitation of CVE-2026-86060...

CVE-2026-82406: Improper Enforcement of Behavioral Workflow

Klever-Go is the Go implementation of the Klever blockchain protocol. Prior to 1.7.20, the native marketplace function core/kapp/market/market.go Buy does not check IsClaimed before accepting a bid. A seller can use the Claim seller-accept branch to settle a resting-bid auction while leaving the...