ctf-ai-ml
CTF AI/ML 攻击技术。当挑战涉及 AI 模型攻击、对抗样本生成、模型提取、Prompt 注入/越狱、LoRA 权重操纵、LLM Token 走私、成员推理攻击、训练数据投毒、神经网络分析时使用。覆盖 FGSM/PGD/C&W 对抗攻击、模型反演、模型权重扰动还原、LLM 工具链劫持、上下文窗口操纵等 AI 安全全链路攻防技术
How do I install this agent skill?
npx skills add https://github.com/wgpsec/aboutsecurity --skill ctf-ai-mlIs this agent skill safe to install?
- Gen Agent Trust Hubwarn
This skill is an educational toolkit for AI and Machine Learning security challenges. It provides functional code and payloads for testing AI model robustness, including adversarial attacks and Large Language Model (LLM) prompt injections. It incorporates advanced techniques for bypassing security filters, such as Unicode-based obfuscation (zero-width characters and homoglyphs) and indirect prompt injection. Additionally, it uses potentially unsafe methods for loading machine learning models which could lead to code execution if applied to untrusted files outside of a controlled environment.
- Socketwarn
2 alerts: gptSecurity, gptAnomaly
- Snykpass
Risk: LOW · No issues
What does this agent skill do?
CTF AI/ML Attack Techniques
When to Use
- Challenge involves ML model files (.pt, .pth, .safetensors, .onnx, .h5)
- Target is an AI chatbot, LLM-based application, or ML classifier
- Need to craft adversarial examples to fool image/text classifiers
- Challenge provides model weights for analysis or manipulation
- AI/ML platform security testing (model extraction, membership inference)
Quick Start
pip install torch transformers numpy scipy Pillow safetensors scikit-learn
file model.*
python3 -c "import torch; m=torch.load('model.pt'); print(type(m), m.keys() if hasattr(m,'keys') else '')"
Decision Tree
- Model weight file (.pt/.safetensors) → model-attacks.md
- Weight perturbation negation, model inversion, LoRA merging, encoder collision
- Image classifier to fool → adversarial-ml.md
- FGSM, PGD, C&W attacks, adversarial patches, evasion, data poisoning
- LLM/chatbot target → llm-attacks.md
- Prompt injection, jailbreaking, token smuggling, tool use exploitation
- Pure math/crypto inside ML → Switch to
ctf-crypto - Compiled model binary → Switch to
ctf-reverse - Python jail wrapped in chatbot → Switch to
ctf-misc
Pivot Signals
- If challenge is pure cryptography/number theory with no ML →
/ctf-crypto - If reverse engineering compiled inference binary →
/ctf-reverse - If Python sandbox escape inside chatbot wrapper →
/ctf-misc - If web app with AI features (prompt injection via web) →
/ai-security/prompt-injection
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/wgpsec/aboutsecurity/ctf-ai-ml">View ctf-ai-ml on skillZs</a>