Review behavior, not just diffs. Jev prioritizes human attention; OpenAI explains the changes. Local CLI + agent skill + GitHub extension.
Deletion-first Agent Skill for removing test bloat, verification theater, and speculative fallbacks while preserving behavior.
Reproducible MATLAB behavioral model for TI-ADC mismatch generation, spectral validation, and calibration benchmarking.
EntraTrace is a defensive security research tool for tracking and identifying the behavior of offensive tooling targeting Microsoft Entra ID.
Jev-powered semantic code search for coding agents β find behavior across repositories via CLI or MCP, with exact source excerpts and line numbers.
Whetstones Lab: measuring coevolution between adaptive malware and behavioral defense. De-weaponized research release β genomes, evolution rounds, telemetry, training harness (stubbed primitives), and lab range tooling. Payloads, real primitives, and C2 infrastructure intentionally withheld. Research and defensive use only.
AISimulate predicts LLM serving behavior and searches for strong deployment configurations offline, without bringing up a GPU serving cluster
A collection of mobile ad countermeasures for bylassing splash ads, such as shake-to-open redirects, ad-triggered navigation, and related intrusive advertising behaviors.
Adaptive Human-Agent Teaming (HAT) & Smart Company (SmC) Infrastructure This repository implements a Human-Model-Free adaptive agent architecture. Instead of relying on rigid, computationally expensive human behavioral models, the system leverages real-time trajectory analysis and the Cross-Entropy
Behavioral testing for AI agents.
Perceptive behavioral foundation models for terrain-adaptive humanoid control: motion tracking, goal reaching, reward inference and teleoperation.
A proof-of-concept PDF file designed for authorized security testing of PDF upload functionality and PDF viewer behavior, including XSS and credential-harvesting scenarios.
Educational reverse engineering of the Elektron Octatrack MKII firmware (OS 1.40C, ColdFire + DSP56xxx) β tooling, notes and behavior patches
Human-Like Roblox Dragon PvP Bot AI 2026: Realistic Flight, Dodging & Combat
Python SDK for DefuzeX agent behavior testing.
match functions in binaries by what they do, not what their bytes look like. behavioral function fingerprinting via microexecution.
Adaptive behavioral VPN routing for MikroTik RouterOS. Pilot ARM64 project in C11.
Behavior-guardrail hooks for Claude Code: test-tampering + outbound-action guards
Claude Opus 5.5 system prompt leak split into readable components (tools, behavior, agentic rules, memory). Reorganized from a public extract for study.
ζι Β· Personal Investment Twin β an AI investment decision coach: rebuilds your investment episodes from your own trading history, attributes P&L with deterministic computation, and reminds you before you repeat past mistakes. No stock picks, no prediction, no auto-trading.
An interactive real-time ocean simulation combining FFT/JONSWAP waves, depth-aware coastal behavior, foam, caustics, refraction and dynamic weather around a tropical island.
API documentation patterns for prompt caching, reasoning tokens, model fields, and model-specific behavior, with Argolink boundary notes.
Static analyzer for Flutter/Dart AOT snapshots β recovers function names, class hierarchies, call graphs, and behavioral signals from libapp.so without embedding or executing the Dart VM. Supports ARM64 and x86_64, Dart 2.10β3.12.
Code for Test-Time Compute for Humanoid Behavior Foundation Models
A collection of AI prompts for generating comprehensive manual test scenarios from software projects, covering features, edge cases, security, validation, and expected behavior.
chrome said that this is not a bug (intended behavior)
Userland EDR for Windows. ETW telemetry, behavioral correlation, MITRE ATT&CK.
Behavioral-discipline skill turning an AI agent into a methodical, non-robotic bug-bounty & pentest hunter - 42 always-on rules: scope, coverage, no false positives, never quits early.
Runtime behavior diffs for pull requests
An elimination death game to study AI and human agent behavior
Project 02 of StudyBuild: Analyze customer behavior, generate business insights, and make data-driven decisions using Python.
Annotation-free multi-animal direction estimation using self-supervised learning
An open-source tool for running controlled behavioral experiments on AI systems at scale.
Vela: User Behavior as Supervision Sparks LLM Intelligence β official paper, task templates, and prompts from AllSpark at Xiaohongshu.
Agent skills for the product process: design, build, review, behavior specs, verification recordings, brand assets.
Machine-readable JSON dataset of AWS service version end-of-support and extended-support dates across EKS, RDS, Aurora, Lambda, ElastiCache, OpenSearch, MSK, Amazon MQ, and DocumentDB. Each entry includes lifecycle status, dates, post-support behavior, a confidence level, and a source URL. Sourced from official AWS documentation.
Modular instrumentation framework for studying real-time memory behavior in PUBG: BATTLEGROUNDS