cyberpulse 0.1.3

Risk-based vulnerability fix list generator

CyberPulse

CyberPulse is a CLI tool to ingest vulnerability scanner exports (JSON, CSV, XML), apply risk‑based prioritisation, enrich findings with CVE data, and generate actionable remediation steps.

Features

• Parse JSON, CSV, and XML scanner outputs
• Apply CVSS‑based prioritisation, public‑facing host checks, known‑exploited CVE flags
• Enrich findings from a local NVD database
• One‑line remediation summaries via on‑device Llama‑2 or GPT‑4 (Pro)
• Output grouped to‑do lists (Critical, High, Medium) in Markdown, plain text, or Slack

Installation

pip install cyberpulse

Quick Start

# Free mode – on‑device Llama‑2 (auto‑fallback to cloud if model missing)
cyberpulse --input scan.json --output fixes.md

# Pro mode – GPT‑4 via CyberPulse Cloud
cyberpulse login            # enter your CyberPulse API key
cyberpulse --input scan.json --cloud --slack

Local, no‑cloud summarisation (offline)

# 1. Download or copy a quantised GGUF model, e.g. Llama‑2‑7B‑Chat.Q4_K_M.gguf
# 2. Point CyberPulse at it
export LLAMA_MODEL_PATH=/Volumes/WEXLER/models/cyberpulse/llama-2-7b-chat.Q4_K_M.gguf

# 3. Generate a report using the on‑device LLM
cyberpulse --input scan.json --summarize local --output fixes.md

If the model is missing, CyberPulse prints:

📎  Local model not installed. See docs → https://cyberpulse.dev/local-llm

The default --summarize auto first tries the local model and transparently falls back to GPT‑4 Cloud.

Development

# From project root
source venv/bin/activate
pip install -r requirements-dev.txt
pytest --cov=src/cyberpulse
flake8 src/cyberpulse
black --check src/cyberpulse

Project Structure

cyberpulse/
├── src/
│   └── cyberpulse/
├── tests/
├── docs/
├── requirements.txt
├── requirements-dev.txt
├── README.md
└── .gitignore

License

MIT