ai-forensics

Windows forensics Engine

Valhuntir CLI — AI-augmented incident response platform

Valhuntir SIFT platform — MCP servers, gateway, Examiner Portal

VERDICT — a DFIR agent (Claude Code as the engine) that produces a signed, offline-verifiable verdict. SANS Find Evil! 2026.

🕵️ An AI-powered forensic orchestration engine for automated disk and memory analysis. Bridges the gap between raw data and actionable intelligence using LLMs (DeepSeek/OpenRouter).

A PoC-grade forensic data recovery and analysis tool designed for digital forensics investigators, incident responders, and cybersecurity professionals.

🧪 Hands-on AI-powered digital forensics lab. Analyze evidence, detect anomalies, and simulate real-world cyber investigations using machine learning, automation & forensic workflows. Built for DFIR learning, experimentation, and research.

Project AIR: forensic reconstruction and incident response for AI agents. Signed AgDR decision records, OWASP ASI01-ASI10 detection, signed forensic evidence exports.

Deepfake detection using temporal deep learning and physiological fingerprint analysis (rPPG)

Repository for AI Security research; will includes notes from various courses and resources.

Multi-agent AI image forensics pipeline that detects AI-generated and manipulated images using parallel vision agents (Gemini + Groq) with a context-aware pre-processing layer and Cerebras-based arbitration to reduce false positives.

This repository contains a Deep Fake Detection model that uses Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks to analyze videos and detect deepfake content. The model extracts frames from videos, processes them using deep learning techniques, and determines whether the video is real or fake based on trained data.

Post-incident forensics for AI agents — record, reconstruct, and analyze agent behavior after security events

AI video & deepfake forensic detection platform built with Django & OpenCV. Analyzes videos using 2D FFT spectrum heatmaps, PRNU sensor noise, and temporal continuity to classify real vs synthetic AI media.

CivilAI Sentinel: An AI-powered forensic structural engineer that uses Gemini 3 Pro's multimodal reasoning to detect, analyze, and diagnose building structural failures (cracks, spalling, settlement) instantly.

Adversarial AI engine that reverse-engineers video content to detect narrative manipulation, emotional coercion, and logical fallacies. Powered by Google Gemini 3.

A specification of what an agent execution trace must contain to support post-hoc causal reconstruction, with a checker, reference fixtures, and a methodology paper.

AI-assisted synthetic media forensics platform for evidence validation, metadata analysis, image signal detection, heuristic assessment, AI-assisted explanation, and PDF reporting.

HAAD: a forensic and assurance methodology for human attribution in AI-agent systems.

Security & compliance engineering ecosystem for cryptographic validation, forensic evidence workflows, and AI‑assisted audit automation.environments and Web3 integrations.