🧠 NullSec Adversarial
Adversarial Machine Learning Attack Toolkit
Evasion, poisoning, and extraction attacks against ML models
🎯 Overview
NullSec Adversarial is a comprehensive toolkit for testing machine learning model robustness. It implements state-of-the-art adversarial attacks — evasion (FGSM, PGD, C&W, AutoAttack), model extraction, membership inference, and model inversion — across image classifiers, NLP models, and tabular ML pipelines.
⚡ Features
| Feature | Description |
|---|---|
| Evasion Attacks | FGSM, PGD, C&W, DeepFool, AutoAttack |
| Model Extraction | Query-based model stealing with knockoff networks |
| Membership Inference | Determine if a sample was in training data |
| Model Inversion | Reconstruct training data from model outputs |
| Transferability | Generate transferable adversarial examples |
| Defence Evaluation | Test adversarial training, certified defences |
| Framework Support | PyTorch, TensorFlow, scikit-learn, ONNX |
�� Attack Matrix
| Attack | Type | Domain | Threat Model |
|---|---|---|---|
| FGSM | Evasion | Image/NLP | White-box |
| PGD | Evasion | Image/NLP | White-box |
| C&W | Evasion | Image | White-box |
| AutoAttack | Evasion | Image | White-box |
| HopSkipJump | Evasion | Image | Black-box |
| Knockoff Nets | Extraction | Any | Black-box |
| Shadow Models | Membership | Any | Black-box |
| MI-FACE | Inversion | Image | White-box |
🚀 Quick Start
# Run PGD evasion attack on an image classifier
nullsec-adversarial evasion pgd --model resnet50.onnx --input samples/ --eps 0.03
# Black-box model extraction
nullsec-adversarial extract --target-url http://api.example.com/predict --queries 10000
# Membership inference attack
nullsec-adversarial membership --model target.pt --members train.csv --non-members test.csv
# Evaluate adversarial robustness
nullsec-adversarial benchmark --model model.pt --dataset cifar10 --attacks all
🔗 Related Projects
| Project | Description |
|---|---|
| nullsec-llmred | LLM red-teaming framework |
| nullsec-datapoisoning | Training data poisoning detection |
| nullsec-modelaudit | ML model security auditing |
| nullsec-promptinject | Prompt injection payloads |
| nullsec-linux | Security Linux distro (140+ tools) |
⚠️ Legal
For authorized ML security testing only. Do not use against models or systems without explicit permission.
📜 License
MIT License — @bad-antics
Part of the NullSec AI/ML Security Suite
Metadata
Release files for nullsec-adversarial 0.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| nullsec_adversarial-0.1.0.tar.gz | 5.9 kB | Without provenance | Details for nullsec_adversarial-0.1.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| nullsec_adversarial-0.1.0-py3-none-any.whl | Python 3 | none | any | Details for nullsec_adversarial-0.1.0-py3-none-any.whl |
Total release size: 13.1 kB
Release files / nullsec_adversarial-0.1.0.tar.gz
| Download URL | nullsec_adversarial-0.1.0.tar.gz |
|---|---|
| Size | 5.9 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
4642e72b1b8f2680a101c425b6b76f04b926a3bed14cdf8149808781c8985270
|
|
BLAKE2b-256 checksum How to use checksums |
07a8d0eef5b20306ad97b7fc2041a8b3089a648602545107e0ca716cfc7ef534
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.13.5
|
Release files / nullsec_adversarial-0.1.0-py3-none-any.whl
| Download URL | nullsec_adversarial-0.1.0-py3-none-any.whl |
|---|---|
| Size | 7.2 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
1b925cdac99422d261d2b32454428648c001463785c16bc5800577f554f0c330
|
|
BLAKE2b-256 checksum How to use checksums |
aaf81d55233a8b73e27c9f6b8c9f67b823b8f059e6d8f98c980b81b5b4d04a1e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.13.5
|
Security for nullsec-adversarial 0.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.