gmh5225/awesome-ai-security
A curated list of AI Security materials and resources for Pentesters, Bug Hunters, and Security Researchers.
# `awesome-ai-security`[](https://awesome.re) [](https://github.com/gmh5225/awesome-ai-security/blob/main/LICENSE) A curated list of AI Security materials and resources for Pentesters, Bug Hunters, and Security Researchers. ``` If you find that some links are not working, you can simply replace the username with gmh5225. Or you can send an issue for me. ``` > Show respect to all the projects below, perfect works of art :saluting_face: ## How to contribute? - https://github.com/HyunCafe/contribute-practice - https://docs.github.com/en/get-started/quickstart/contributing-to-projects ## Skills for AI Agents This repository provides skills that can be used with AI agents and coding assistants such as [Cursor](https://www.cursor.com/), [OpenClaw](https://docs.openclaw.ai/), [Claude Code](https://docs.anthropic.com/en/docs/claude-code), [Codex CLI](https://github.com/openai/codex), and other compatible tools. Install skills to get specialized knowledge about game security topics. - https://github.com/vercel-labs/skills [The open agent skills tool - npx skills] **[View on learn-skills.dev](https://learn-skills.dev/skills/gmh5225/awesome-ai-security)** **Installation:** ```bash npx skills add https://github.com/gmh5225/awesome-ai-security --skill <skill-name> ``` **Available Skills:** | Skill | Description | |-------|-------------| | `adversarial-machine-learning` | Adversarial machine learning: adversarial examples, data poisoning, model backdoors, and evasion attacks | | `ai-powered-pentesting` | AI-powered penetration testing tools, red teaming frameworks, and autonomous security agents | | `llm-attacks-security` | LLM security attacks: prompt injection, jailbreaking, and data extraction | | `awesome-ai-security-overview` | Overview of this repository and contribution guidelines | | `ai-security-tooling` | AI security tooling: detectors, analyzers, guardrails, and benchmarks | **Example:** ```bash # Install LLM attacks skill npx skills add https://github.com/gmh5225/awesome-ai-security --skill llm-attacks-security # Install multiple skills npx skills add https://github.com/gmh5225/awesome-ai-security --skill adversarial-machine-learning npx skills add https://github.com/gmh5225/awesome-ai-security --skill ai-powered-pentesting ``` ## AI Security Starter Pack - **CTFs / Practice** - https://github.com/verialabs/ctf-agent [ctf-agent - autonomous CTFd solver: coordinator LLM + parallel model swarms in Docker; BSidesSF 2026 1st] - https://aivillage.org/ [AI Village @ DEF CON - LLM Jailbreak Challenges] - https://doublespeak.chat/#/handbook [Doublespeak - AI Security Challenges] - https://github.com/EasyJailbreak/EasyJailbreak [Framework for adversarial jailbreak prompts] - https://github.com/microsoft/AI-Red-Teaming-Playground-Labs [Microsoft AI Red Teaming Playground Labs] - https://github.com/schwartz1375/genai-security-training [GenAI Red Teaming Training] - **Blogs / Resources** - https://genai.owasp.org/ [OWASP GenAI Security Project] - https://llm-stats.com [LLM Leaderboard] - https://www.aidaily.win [AI Daily News] - https://baoyu.io/blog/how-to-write-good-prompt [How to Write Good Prompts] - https://rootissh.in/ [LLM Pentesting Series Blog] - https://github.com/Abdowaer098/Wa3r-OffSec-Kit [Wa3r OffSec Kit - offensive-security knowledge base with practical workflows, payload patterns, case studies, and forensics notes] - **Newsletters / Collections** - https://mlsecops.com/podcast [MLSecOps Podcast] - https://podcasts.apple.com/ph/podcast/the-genai-security-podcast/id1782916580 [GenAI Security Podcast] - https://avidml.org/ [AI Vulnerability Database (AVID)] - https://calder-brief.pages.dev/ai-hub.html [Calder Brief AI news hub — daily-updated AI news, including AI security/safety coverage] - **Certifications / Courses** - https://cs229.stanford.edu/ [Stanford CS229: Machine Learning] - https://course.fast.ai/ [fast.ai Practical Deep Learning] - https://www.coursera.org/specializations/deep-learning [Deep Learning Specialization by Andrew Ng] - https://huggingface.co/reasoning-course [Build DeepSeek-R1 like Reasoning Model] ## AI/LLM Guide - **Foundations** - https://d2l.ai/ [Dive into Deep Learning - Interactive book with PyTorch/JAX/TensorFlow] - http://neuralnetworksanddeeplearning.com/ [Neural Networks and Deep Learning by Michael Nielsen] - https://www.deeplearningbook.org/ [Deep Learning by Goodfellow, Bengio, Courville] - https://github.com/karminski/one-small-step [AI/LLM Tutorial] - https://github.com/datawhalechina/happy-llm [LLM Principles and Practice Tutorial] - https://github.com/rasbt/LLMs-from-scratch [Build LLM from Scratch] - https://github.com/naklecha/llama3-from-scratch [LLaMA3 from Scratch] - https://github.com/ZJU-LLMs/Foundations-of-LLMs [Foundations of LLMs] - **Awesome Lists** - https://github.com/WangRongsheng/awesome-LLM-resourses [Comprehensive LLM Resources] - https://github.com/mahseema/awesome-ai-tools [Awesome AI Tools] - https://github.com/Shubhamsaboo/awesome-llm-apps [Awesome LLM Apps] - https://github.com/mahonzhan/awesome-agent-harness [Curated list of agent harnesses, agent frameworks, workflow frameworks, and emerging agent protocols] - https://github.com/punkpeye/awesome-mcp-servers [Awesome MCP Servers] - https://github.com/wong2/awesome-mcp-servers [Awesome MCP Servers] - https://github.com/deepseek-ai/awesome-deepseek-integration [Awesome DeepSeek Integration] - https://github.com/lmmlzn/Awesome-LLMs-Datasets [Awesome LLMs Datasets] - **From-scratch LLMs / Reasoning** - https://github.com/rasbt/LLMs-from-scratch/tree/main/ch05/11_qwen3 [Qwen3 From Scratch - Chinese walkthrough] - https://github.com/rasbt/LLMs-from-scratch/blob/main/ch05/11_qwen3/standalone-qwen3-moe-plus-kvcache.ipynb [Implement Qwen3 MoE with KV cache from scratch] - https://github.com/rasbt/LLMs-from-scratch/tree/main/ch05/12_gemma3 [Build Gemma 3 270M from scratch] - https://github.com/rasbt/reasoning-from-scratch [Reasoning models from scratch] - https://github.com/mingyin0312/RLFromScratch [Reinforcement learning from scratch (Chinese tutorial)] - https://github.com/karpathy/nanochat [End-to-end nanochat training loop in ~8K lines] - https://github.com/kyegomez/OpenMythos [OpenMythos - first-principles theoretical reconstruction of Claude Mythos architecture based on public research literature] - https://github.com/vixhal-baraiya/microgpt-c [MicroGPT-C — train and infer a tiny GPT in pure dependency-free C (single file); fp32/AVX2 style CPU path; MIT] ## AI Security & Attacks ### Prompt Injection - https://www.lakera.ai/blog/guide-to-prompt-injection [Prompt Injection Guide] - https://genai.owasp.org/llmrisk/llm01-prompt-injection/ [OWASP LLM01:2025 Prompt Injection] - https://redbotsecurity.com/prompt-injection-attacks-ai-security-2025/ [Prompt Injection Attacks 2025] - https://github.com/protectai/rebuff [Self-hardening Prompt Injection Detector] - https://github.com/NVIDIA/garak [NVIDIA LLM Vulnerability Scanner] - https://github.com/deadbits/vigil-llm [Detects Prompt Injections and Risky Inputs] - https://github.com/alphasecio/prompt-guard [Prompt Defense for LLM] - https://github.com/tml-epfl/llm-adaptive-attacks [Adaptive Attacks on LLMs] - https://github.com/RomiconEZ/llamator [LLM Vulnerability Testing Framework] - https://github.com/gh0stOo/claude-md-vorlagen-de/blob/main/guides/prompt-hardening.md [German-language prompt-injection hardening guide with 10 concrete before/after code patterns (system/user separation, delimiters, output validation, RAG source distrust)] - https://github.com/Vick606/subcanopy-guard [Context-Aware Indirect Prompt Injection Scanner] ### Adversarial Attacks - https://gradientscience.org/intro_adversarial/ [Introduction to Adversarial Examples] - https://cset.georgetown.edu/publication/key-concepts-in-ai-safety-robustness-and-adversarial-examples/ [AI Safety and Adversarial Examples] - https://github.com/Trusted-AI/adversarial-robustness-toolbox [IBM Adversarial Robustness Toolbox] - https://github.com/QData/TextAttack [Adversarial Attacks on NLP Models] - https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-2e2025.pdf [NIST Adversarial ML Taxonomy] - https://llm-vulnerability.github.io/ [ACL 2024 Tutorial: LLM Vulnerabilities] - https://github.com/tensorflow/cleverhans [CleverHans - ML Vulnerability Benchmark] - https://github.com/bethgelab/foolbox [Foolbox - Adversarial Examples Toolbox] - https://github.com/cchio/deep-pwning [Deep-pwning] ### Poisoning & Backdoors - https://arxiv.org/abs/2009.02276 [Witches' Brew: Industrial Scale Data Poisoning] - https://arxiv.org/abs/2402.09179 [Instruction Backdoor Attacks on Customized LLMs] - https://arxiv.org/abs/2510.07192 [Poisoning Attacks Need Only a Few Points] - https://arxiv.org/abs/1910.03137 [MNTD: Detecting AI Trojans] - https://owasp.org/www-project-top-10-for-large-language-model-applications/ [OWASP Top 10 for LLM Applications 2025] - https://github.com/git-disl/awesome_LLM-harmful-fine-tuning-papers [LLM Harmful Fine-tuning Papers] ### Privacy & Extraction - https://www.usenix.org/conference/usenixsecurity21/presentation/carlini-extracting [Extracting Training Data from LLMs] - https://arxiv.org/abs/2309.10544 [Model Leeching: Extraction Attack on LLMs] - https://arxiv.org/abs/2301.10226 [Watermark for Large Language Models] - https://arxiv.org/abs/2103.07853 [Membership Inference Attacks Survey] - https://arxiv.org/abs/2503.19338 [MIAs on Large-Scale Models Survey] - https://trustllmbenchmark.github.io/TrustLLM-Website/ [TrustLLM Benchmark] - https://github.com/stratosphereips/awesome-ml-privacy-attacks [Awesome ML Privacy Attacks] - https://github.com/chawins/llm-sp [LLM Security Papers] - https://github.com/journey-ad/gemini-watermark-remover [Client-side Gemini AI image watermark remover - Reverse Alpha Blending] ### Model Security - https://arxiv.org/html/2507.02737v1 [Steganography Capabilities in Frontier LLMs] - https://jplhughes.github.io/bon-jailbreaking/ [AI Jailbreaking] - https://github.com/Goochbeater/Spiritual-Spell-Red-Teaming [A repo for jailbreaking various LLMs, mainly Claude] - https://huggingface.co/blog/mlabonne/abliteration [Model Abliteration] - https://github.com/p-e-w/heretic [Heretic - fully automatic censorship removal for LLMs, abliteration + Optuna TPE optimizer, dense/MoE/multimodal] - https://github.com/FailSpy/abliterator [abliterator - Python library to ablate refusal/features in LLMs, TransformerLens, cache activations, refusal direction] - https://github.com/spkgyk/abliteration [Abliteration - uncensor LLMs via refusal-vector removal, PyTorch hooks, no TransformerLens] - https://github.com/jim-plus/llm-abliteration [llm-abliteration - make abliterated models with Transformers, batch inference, dense/MoE, norm-preserving biprojected, low VRAM] - https://github.com/Tsadoq/ErisForge [ErisForge - dead simple LLM abliteration, transform internal layers, AblationDecoderLayer/AdditionDecoderLayer, ExpressionRefusalScorer] - https://github.com/protectai/llm-guard [LLM Guard - Security Tool] - https://github.com/protectai/modelscan [ModelScan - Scan Models for Unsafe Code] - https://github.com/fr0gger/nova-framework [Nova Framework - Jailbreak Detection] - https://github.com/fr0gger/nova_mcp [Nova MCP Server] - https://github.com/0xAIDR/AIDR-Bastion [GenAI Protection System] - https://github.com/CAU-ISS-Lab/AIGT-Detection-Evade-Detection [AI-Generated Text Detection & Evasion] - https://github.com/AUGMXNT/deccp [deccp - Evaluating and unaligning Chinese LLM censorship, abliteration PoC for Qwen2] - https://github.com/gpiat/AIAE-AbliterationBench [AbliterationBench - benchmark model resilience to residual stream/abliteration attacks] ## AI Pentesting & Red Teaming ### AI-Powered Pentesting - https://github.com/GreyDGL/PentestGPT [GPT-4 Powered Pentesting Agent] - https://github.com/zakirkun/guardian-cli [AI-Powered Pentesting CLI with Gemini] - https://github.com/usestrix/strix [AI Security Pentesting] - https://github.com/aliasrobotics/cai [CAI - Cybersecurity AI Framework] - https://github.com/promptfoo/promptfoo [AI Agent Pentesting Framework] - https://github.com/antoninoLorenzo/AI-OPS [AI Assistant for Penetration Testing] - https://github.com/yz9yt/BugTrace-AI [AI Automated Web Pentesting] - https://github.com/six2dez/reconftw_ai [ReconFTW with AI Analysis] - https://github.com/projectdiscovery/katana [Katana (ProjectDiscovery) - fast web crawler/spider for automation: standard & headless, JS endpoint parsing, scope/regex filters, JSONL; stacks with httpx/nuclei and AI pentest agent workflows] - https://github.com/Ed1s0nZ/CyberStrikeAI [AI-Native Security Testing Platform with 100+ Tools] - https://github.com/vxcontrol/pentagi [PentAGI - Fully autonomous AI agents for penetration testing] - https://github.com/XenoCoreGiger31/GEMMA-by-GOOGLE [HALO (GEMMA-by-GOOGLE) - Fully local, autonomous AI pentesting agent; a local Gemma model drives a 29-tool MCP server through recon, attack, and reporting] - https://github.com/Agnuxo1/EnigmAgent [EnigmAgent - autonomous AI security research/pentest agent with CTF benchmarking, decentralized swarm coordination, exploit validation, and Lean 4-backed security claims] - https://github.com/KeygraphHQ/shannon [Shannon - Autonomous AI pentester, finds and executes real exploits in web apps] - https://github.com/wudidike/pentest_skill [Black-box web penetration-testing automation framework for AI agents with phase-based workflow and report generation] - https://github.com/yv1ing/Z3r0 [Z3r0 - controlled multi-agent workbench for authorized security assessment, code auditing & research: CSO coordinator + intelligence/penetration/reverse/crypto specialists, Docker-bound sandboxes (shell/noVNC/file mgr), persistent delegation, LiteLLM/OpenAI-compatible; FastAPI + React; MIT] - https://github.com/ASCIT31/Dark-Moon [Dark-Moon - autonomous AI penetration testing platform where Markdown methodology playbooks orchestrate 80+ offensive tools via MCP across web, cloud, Active Directory, Kubernetes and API targets, with an evidence trail per finding] - https://github.com/Strategic-Automation/violin [Violin - supervised, agentic Hermes Agent pentest profile: 31 skill-based playbooks (OWASP Top 10, API Top 10, LLM Top 10) with interactive scoping, scope validation, and approval gates. Installs via `hermes profile install https://github.com/Strategic-Automation/violin`. Hermes-native, no extra API keys; MIT] - https://github.com/secorizon/SecorizonAI [SecorizonAI - terminal-native AI shell for pentesters: single Go binary, local LLM via Ollama, JSON tool-use ReAct loop with shell execution + built-in web search, on-demand methodology guides, MCP (Burp) support; Apache-2.0 + Commons Clause; use only on authorized engagements] ### AI Red Teaming Tools - https://github.com/Azure/counterfit [Microsoft ML Penetration Testing Tool] - https://github.com/Azure/PyRIT [Microsoft Red-Teaming Framework for GenAI] - https://github.com/meta-llama/PurpleLlama [Meta Open-Source LLM Safety Tools] - https://github.com/NVIDIA/NeMo-Guardrails [NVIDIA Programmable Guardrails] - https://github.com/NoDataFound/hackGPT [LLM Toolkit for Offensive Security] - https://github.com/ipa-lab/hackingBuddyGPT [Autonomous Red-Teaming Agent] - https://github.com/Yanlewen/TradeTrap [TradeTrap - test LLM-based trading agents: prompt injection, MCP hijacking, state tampering, memory poisoning; AI-Trader/Valuecell] - https://github.com/humanbound/humanbound [Humanbound - open-source CLI that runs LLM-generated adversarial attacks against an agent's HTTP endpoint, scored against OWASP LLM Top 10, OWASP Agentic Top 10, NIST, and EU AI Act mappings] - https://github.com/toby-bridges/api-relay-audit [API Relay Audit - local audit CLI for AI API relays and LLM proxies; probes for prompt injection signals, package-command changes, error-response leakage, and Anthropic SSE anomalies; Markdown reports] - https://github.com/taoq-ai/ziran [Ziran - security testing framework for AI agents: graph-based tool chain discovery, execution-level side-effect detection, multi-phase campaigns; LangChain, CrewAI, MCP, A2A, remote HTTP] ### AI Security MCP Tools - https://github.com/tomelias10/mcp-drift-check [MCP Drift Check — passive local Python CLI that statically flags mutable/unpinned npm/npx package references in MCP client configs; does not execute servers, download packages, or upload configs; MIT] - https://github.com/AndrewXuTurtle/mcpaudit [mcpaudit — MCP server + CLI that audits the MCP servers you already have installed across Claude Desktop, Claude Code, Cursor, Windsurf and VS Code. Tool poisoning (incl. zero-width/bidi chars in tool descriptions), credential blast radius, privilege scope, transport, and supply-chain provenance: homoglyph publisher scopes, registry-removed packages, install hooks, and OSV/GHSA advisories matched to the resolved version. Zero dependencies; nothing installed or executed; MIT] - https://github.com/ArmorerLabs/Armorer-Guard [Armorer Guard - local Rust scanner and MCP proxy for prompt injection, credential leakage, exfiltration, and risky tool-call detection before execution] - https://github.com/0x4m4/hexstrike-ai [HexStrike AI - 150+ Cybersecurity Tools MCP] - https://github.com/cyproxio/mcp-for-security [Pentesting MCP] - https://github.com/johnhalloran321/mcpSafetyScanner [MCP Safety Scanner] - https://github.com/Karthikathangarasu/pentest-mcp [Pentest MCP] - https://github.com/Sicks3c/hackerone-mcp-server [HackerOne MCP — unofficial Hacker API over stdio: reports, programs, scope, earnings, hacktivity; submit/comment/close; MIT] - https://github.com/zhizhuodemao/android_proxy_mcp [Android Proxy MCP - MCP-based Android traffic capture, let AI analyze HTTP/HTTPS via natural language] - https://github.com/MHaggis/Security-Detections-MCP [Security Detections MCP - unified Sigma/Splunk ESCU/Elastic/KQL, 71+ tools, 11 prompts, autonomous detection platform] - https://github.com/ai-blueteam/http-detection-agent [http-detection-agent - capability-aware HTTP attack detection: Rust CLI and local MCP server over a normalized rule catalog of 76 detections across 62 behavior families; MIT] - https://github.com/Correctover/correctover-scan [correctover-scan - zero-config npm CLI that scans MCP client configs (Claude Desktop, Cursor, VS Code) for credential exposure, SSRF, missing auth/transport encryption and other misconfigurations; findings mapped to OWASP AISVS; JSON/SARIF output for CI; MIT] - https://github.com/rolandpg/zettelforge [ZettelForge — CTI agentic memory MCP server with entity extraction (CVEs, threat actors, IOCs, MITRE ATT&CK), knowledge graph with alias resolution, STIX 2.1, intent-classified retrieval, OCSF audit logging; offline; MIT] - https://github.com/doomsday-network/bounty-guard [BountyGuard — Open-source prompt-injection firewall, published honeypot blocklist (data/honeypots.json), and zero-dependency MCP security server (mcp_server.py / https://doomsday.network/mcp) that screens GitHub repositories for trap signatures and classifies issues into SAFE_WITH_ESCROW, SAFE_NO_ESCROW, or QUARANTINE; MIT] ### AI-Powered C2 - https://github.com/Red-Hex-Consulting/Ankou [AI C2 Framework] ### AI Password Cracking - https://github.com/d-sec-net/VPK [AI Automated Password Cracking] ## AI Security Tools & Frameworks ### AI SOC & SecOps - https://github.com/Vigil-SOC/vigil [Vigil - open-source AI-native SOC: 12 specialized agents, multi-agent workflows, MCP integrations (SIEM/EDR/TI/sandbox/ticketing), FastAPI + React] - https://github.com/ccjmcc/jevsec [JevSec - Jev-compatible decision-model security triage for short cross-request web behavior; combines deterministic rules with local Qwen3-4B, privacy-aware normalization, and shadow-mode review; MIT] ### AI Reverse Engineering - https://github.com/ZeroDaysBroker/GhidraGPT [GPT Integration for Ghidra] - https://github.com/jtang613/GhidrAssist [LLM Extension for Ghidra] - https://github.com/0xeb/windbg-copilot [WinDbg Copilot - Agentic Debugging extension] - https://github.com/agentrebench/AgentRE-Bench [AgentRE-Bench - Agentic benchmark for long-horizon binary RE: C2/encoding/anti-analysis, deterministic scoring, 13 ELF tasks] - https://github.com/banteg/bn [bn - Agent-friendly Binary Ninja CLI: decompile, xrefs, types, mutations via Unix socket to GUI plugin] - https://github.com/amruth-sn/kong [Kong - agentic reverse engineer, LLM-orchestrated binary RE via in-process Ghidra, call-graph analysis, agentic deobfuscation] - https://github.com/CSIT-SG/AETHER [An AI-powered reverse-engineering copilot for assisting tedious malware analysis in IDA Pro] - https://github.com/thatskriptkid/re-harness [RE-harness - Windows PE static-analysis OpenCode plugin for Qwen 27B/35B: structured read-only IDA/IDASQL tools, NeverD/LLVM large-function lift fallback; malware analysis harness; macOS/Linux] - https://github.com/mrphrazer/ghidra-headless-mcp [ghidra-headless-mcp — headless Ghidra over MCP] - https://github.com/vwww-droid/Mira [Mira - mobile runtime detection workbench (Android/iOS): AI-native live runtime analysis with Relay + MCP, remote shell/Frida workflows, and reusable detection knowledge] - https://github.com/buzzer-re/ToCode [ToCode - export binaries/IDA databases into source-like project trees (recovered C, asm, summaries, rich JSON metadata + AGENTS.md) so coding agents can traverse large binaries as an oracle; supports IDA Pro and radare2] - https://github.com/cellebrite-labs/ghidra-rpc [ghidra-rpc - agentic Ghidra skill: runs Ghidra as a persistent PyGhidra daemon exposing decompile/xrefs/type-recovery/patch/binary-diff via a JSON CLI over Unix socket, drivable by any shell-capable AI assistant (Cursor, Claude Code, etc.); by Cellebrite Labs] - https://github.com/morluto/rea [REA - local CLI and MCP server for agent-assisted reverse engineering of native binaries, JavaScript/Electron apps, and .NET assemblies; evidence provenance and explicit limitations; native deep analysis uses separately installed Hopper, Ghidra, or IDA] ### AI Vulnerability Detection - https://github.com/Mayaaa311/LLMBugScanner [LLM BugScanner - GPTLens-style pipeline: pluggable HF code LLMs as auditor + critic, rank findings by correctness/severity; Solidity-oriented datasets] - https://github.com/LLMAudit/LLMSmartAuditTool [LLM-SmartAudit - multi-agent Solidity auditing (BA/TA modes), task-oriented roles, 40+ detector prompts, batch notebooks + web visualizer, OpenAI API; arXiv:2410.09381] - https://github.com/scabench-org/hound [AI Auditor with Adaptive Knowledge Graphs] - https://github.com/416rehman/DeepZero [DeepZero - automated vulnerability research pipeline engine (YAML-defined stages) for large-scale Windows kernel driver analysis, decompilation, and IOCTL exploitability assessment with AI agents] - https://github.com/kpolley/redai [RedAI - terminal workbench for AI-driven vulnerability discovery with live validation evidence (browser/iOS validator environments, PoCs, logs, screenshots, reproducible reports)] - https://github.com/vercel-labs/deepsec [deepsec - coding-agent-powered security harness for large codebases: candidate-site scanning, AI investigation/revalidation, and exportable findings workflows] - https://github.com/Agent-Field/sec-af [SEC-AF - AI-native security auditor on AgentField: proves exploitability with verdicts, data-flow traces, and actionable evidence; adversarial hunt/prove agent DAG; Apache-2.0] - https://github.com/evilsocket/audit [audit - 8-stage vulnerability-discovery agent (Cloudflare Glasswing-style): Recon/Hunt/Validate/Gapfill/Dedupe/Trace/Feedback/Report; narrow agents + adversarial disprove + reachability gate; Claude Code Agent SDK, subscription OAuth; MIT] - https://github.com/visa/visa-vulnerability-agentic-harness [VVAH (Visa) - agentic SAST pipeline for autonomous vulnerability discovery: 9-stage threat modeling → multi-lens research → adversarial verification → SARIF; multi-model (Claude CLI/SDK, OpenAI); Glasswing-inspired; `vvaharness` CLI; authorized use only] - https://github.com/scadastrangelove/rust-in-peace [rust-in-peace - agentic Rust vulnerability-discovery harness: autonomous recon/find/triage/patch loop for unsafe/FFI memory bugs, panic-DoS, and deserialization-trust issues; verifies findings with Miri/ASan/panic/hang detectors and cargo-fuzz] - https://github.com/zakirkun/deep-eye [Deep Eye - AI-driven vulnerability scanner and pentesting framework with multi-provider LLM support, payload generation, recon modules, and report export] - https://semgrep.dev/ [AI-Assisted SAST] - https://github.com/squirrelscan/squirrelscan [Website audit tool for agent/LLM workflows (security/performance/SEO)] - https://github.com/rohansx/vgx [Git pre-commit security scanner with LLM integration (detect AI code + vulnerabilities)] - https://github.com/HikaruEgashira/vulnhuntrs [AI Web Security Audit Tool] - https://github.com/xvnpw/ai-security-analyzer [AI Security Doc Generator] - https://github.com/aress31/burpgpt [BurpGPT - AI Vulnerability Scanning] - https://github.com/haroonawanofficial/AISA-Scanner [AI Security Scanner] - https://github.com/youpengl/OpenVul [OpenVul - Post-training framework for LLM-based vulnerability detection (SFT/DPO/ORPO/GRPO)] - https://github.com/Pinperepette/snakebite [snakebite - PyPI supply-chain scanner: heuristics + optional LLM to cut false positives; local or RSS feed mode] - https://github.com/rushter/hexora [hexora - Rust static analyzer for malicious Python (supply-chain, pasted scripts, IoCs); AST rules with confidence levels] - https://github.com/sashiko-dev/sashiko [Sashiko - agentic Linux kernel patch review (Rust): lore.kernel.org + local git, multi-stage LLM protocol (implementation, concurrency, security, drivers), web UI/CLI; self-contained; patch/kernel context sent to LLM—authorize sharing & monitor API cost; Apache-2.0] ### AI CVE Analysis - https://github.com/arschlochnop/VulnWatchdog [CVE Monitoring with GPT Analysis] - https://github.com/suhasgowtham-x/aegis-security-co-pilot [AI CVE Scanner] - https://github.com/ucsb-mlsec/VulnLLM-R [Specialized Reasoning LLM for Vulnerability] - https://github.com/RogoLabs/VulnRadar [VulnRadar - Vulnerability radar via GitHub Actions: CVE watchlist, KEV/EPSS/PatchThis, issues, Discord/Slack/Teams] ### AI OSINT - https://ai.cylect.io/ [AI OSINT] - https://github.com/apurvsinghgautam/robin/ [AI-Powered Dark Web OSINT Tool] - https://github.com/calesthio/Crucix [Crucix - self-hosted OSINT terminal: 27 open feeds (satellite, flights, conflict, markets), Jarvis dashboard, optional LLM alerts & Telegram/Discord bots] ### AI Security Libraries - https://secml.readthedocs.io/ [SecML - Secure and Explainable ML Library] - https://github.com/google/oss-fuzz-gen [AI Code Audit Fuzzing Tool] - https://github.com/Invicti-Security/brainstorm [AI Fuzzer for Web Applications] - https://github.com/NativeStar/js-hooker [js-hooker - lightweight JavaScript hooking library for browser environments (runtime interception/instrumentation helper)] ### TLS, fingerprint & bot signals (web / automation) - https://github.com/rawandahmad698/noble-tls [noble-tls - Python HTTP client with TLS/JA3 impersonation (requests-like API, auto-updated fingerprints)] - https://github.com/lexiforest/curl_cffi [curl_cffi - Python bindings to libcurl-impersonate: browser-aligned TLS/JA3 and HTTP/2 fingerprints without a full browser; strong default for scripted fetches] - https://github.com/0x676e67/rnet [rnet - Rust HTTP client with TLS JA3/JA4 and HTTP/2 fingerprint control] - https://github.com/fingerprintjs/BotD [BotD (FingerprintJS) - open-source client-side bot detection SDK you embed in your own pages (self-hosted / first-party integration)] - https://github.com/tiagozip/cap [Cap - privacy-first self-hosted CAPTCHA alternative (proof-of-work + instrumentation challenges); ~20kb, no telemetry, standalone Docker, invisible mode; reCAPTCHA/hCaptcha/Turnstile alternative; Apache-2.0] - https://github.com/botswin/BotBrowser [BotBrowser - cross-platform Chromium for automation/QA against anti-bot stacks (Cloudflare, Akamai, Kasada, Shape, DataDome, PerimeterX, hCaptcha, FunCaptcha, Imperva, reCAPTCHA, ThreatMetrix, Adscore, etc.); use only on systems you own] - https://github.com/MiddleSchoolStudent/BotBrowser [BotBrowser (alt distribution) - headless-oriented Chromium builds for anti-bot automation; compare with botswin fork; authorized targets only] - https://github.com/zhom/donutbrowser [Donut Browser - open-source anti-detect Chromium (Wayfern) with isolated profiles, proxy/VPN, Local API & MCP for Claude/automation; AGPL-3.0; authorized targets only] - https://github.com/daijro/camoufox [Camoufox - stealth-oriented Firefox for scraping/automation; pairs well with Browser-Use-style stacks and Cloudflare-challenge helper tools (e.g. solver proxies); use only where you have authorization] - https://github.com/AlloryDante/undetected-browser [undetected-browser - modified Puppeteer/Chromium stack for lower-detection automation testing] - https://github.com/ultrafunkamsterdam/nodriver [nodriver - undetected-style Chrome control without classic WebDriver surface; Python driver for hardened automation research] - https://github.com/Xewdy444/CF-Clearance-Scraper [CF-Clearance-Scraper - scriptable retrieval of Cloudflare `cf_clearance` / session artifacts for HTTP clients; for authorized testing and research only] - https://github.com/FlareSolverr/FlareSolverr [FlareSolverr - self-hosted HTTP API/proxy that solves Cloudflare challenges and returns cookies/HTML for downstream clients; deploy only on networks and sites you are allowed to test] - https://github.com/xKiian/awswaf [awswaf - AWS WAF browser-challenge / token handling for scripted clients; for authorized security research and targets you own or have explicit permission to test] - https://github.com/fingerprintjs/fingerprintjs [FingerprintJS - browser fingerprinting library (open-source visitor identification)] - https://github.com/abrahamjuliot/creepjs [CreepJS - browser fingerprinting + anti-spoofing / lie detection; modular parallel collection for privacy and bot research] - https://github.com/juu17/browser-fingerprint-shuffler [browser-fingerprint-shuffler - browser extension to shuffle fingerprint-related signals (privacy / QA research)] - https://pixelscan.net/fingerprint-check [Pixelscan - online browser fingerprint consistency / leak checker] - https://github.com/Myronfr/RISC-Fingerprinting2025 [RISC-Fingerprinting2025 - browser fingerprinting research materials] - https://github.com/Myronfr/AkamaiBmpGen2025 [AkamaiBmpGen2025 - Akamai BMP/sensor research tooling and notes (e.g. Akamai-ACF related artifacts)] - https://nullpt.rs/compiling-browser-to-bypass-antibot-measures [nullpt.rs - building Chromium variants for anti-bot / automation research (write-up)] - https://github.com/zmzimpl/chrome-power-app [Chrome Power App - companion app for customized Chromium / fingerprint-oriented workflows] - https://github.com/zmzimpl/chrome-power-chromium [chrome-power-chromium - Chromium sources for Chrome Power–style builds] ### AI Agent Security - https://github.com/maximhq/bifrost [Bifrost Edge - endpoint governance for AI apps and MCP servers with guardrails, approvals, audit logs, and device management] - https://github.com/NVIDIA/NemoClaw [NVIDIA plugin for secure installation of OpenClaw - sandboxed agents with Landlock/seccomp/netns, policy-enforced egress and inference] - https://github.com/peg/rampart [Firewall for AI agents - policy engine for OpenClaw, Claude Code, Cursor, Codex] - https://github.com/openguardrails/openguardrails [OpenGuardrails - Runtime security for AI agents: prompt injection, credential leakage, exfiltration, behavioral threats] - https://github.com/cisco-ai-defense/skill-scanner [Security scanner for agent skills - prompt injection, exfiltration, malicious code] - https://github.com/Tencent/AI-Infra-Guard [AI-Infra-Guard (A.I.G) - Tencent Zhuque Lab's full-stack AI red-teaming platform: Agent/Skill/MCP scan, AI infra CVE scan (146 components, 2000+ rules), and LLM jailbreak evaluation; ships `aig-skill-scan` as the official OpenClaw ClawHub scanner (SARIF 2.1.0 output, SkillTrustBench T01-T09 taxonomy); web UI + CLI + Docker; Apache-2.0] --- [Read more](https://github.com/gmh5225/awesome-ai-security)
