Instructions to use Cyb3rDudu/cyb3r with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Cyb3rDudu/cyb3r with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="Cyb3rDudu/cyb3r")# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("Cyb3rDudu/cyb3r", dtype="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use Cyb3rDudu/cyb3r with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "Cyb3rDudu/cyb3r" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Cyb3rDudu/cyb3r", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/Cyb3rDudu/cyb3r
- SGLang
How to use Cyb3rDudu/cyb3r with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "Cyb3rDudu/cyb3r" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Cyb3rDudu/cyb3r", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "Cyb3rDudu/cyb3r" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "Cyb3rDudu/cyb3r", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use Cyb3rDudu/cyb3r with Docker Model Runner:
docker model run hf.co/Cyb3rDudu/cyb3r
cyb3r
cyb3r is an offensive-security reasoning model project for a Pi-based agent harness.
The model is one part of a larger operator stack:
- Pi agent provides the runtime layer; in the system analogy, think of it as the operating-system layer.
- cyb3r is the agent harness that coordinates planning, execution, reassessment, and tool use.
- The model provides penetration-testing reasoning, branch ranking, pivot decisions, and operator discipline; think of it as the CPU inside the harness.
- pi-pen-kit is the tool layer for reverse shells, Sliver, Metasploit, and operator-level C2 workflows.
Current Status
Training is in progress. We are currently in the first epoch of an epoch 01-03 reasoning-focused pass.
The current corpus is a cybersecurity-focused training corpus of approximately 1.6B cybersecurity-related training rows. The first training focus is reasoning behavior:
- structured penetration-testing reasoning
- global reassessment during multi-step tasks
- attack-path branch ranking
- reduced rabbit-hole fixation
- preserved tool-call competence
After the first reasoning-focused pass, we plan to broaden and strengthen security knowledge across tools, techniques, procedures, vulnerabilities, and defensive context.
Hardware / Partner Note
Training hardware is currently the limiting factor. We are actively looking for partners who can help with compute, evaluation capacity, applied security feedback, or controlled real-world workflows.
Intended Use
cyb3r is intended for authorized security research, lab environments, defensive validation, and controlled penetration-testing workflows.
It is not intended for unauthorized access, abuse, or deployment against systems without permission.
Availability
This repository is a teaser/model-card placeholder while the first training run is in progress. Weights, adapters, evaluation notes, and usage examples will be added as artifacts become available.
Related Tooling
pi-pen-kit provides Pi extensions for offensive-security workflows:
- controlled reverse-shell listeners and session handling
- Sliver C2 operator workflows
- Metasploit RPC workflows
- interactive shell read/write/interrupt/probe/exec primitives
Model tree for Cyb3rDudu/cyb3r
Base model
Qwen/Qwen3.6-27B