raven-vapt 0.1.0

Offline-first AI-powered VAPT framework for Linux

RAVEN

Recon Automation & Vulnerability Enumeration Network (RAVEN) is an offline-first, AI-assisted VAPT framework for Linux.

Prerequisites

• Python 3.12+
• uv package manager
• nmap
• ollama with a local model (example: qwen2.5:14b)
• Optional for Phase 2: Metasploit (msfconsole, msfrpcd)

Installation

git clone <repo-url>
cd raven
uv sync
cp .env.example .env

Set your .env for local LLM operation:

RAVEN_LLM_PROVIDER=ollama
RAVEN_MODEL=qwen2.5:14b
OLLAMA_BASE_URL=http://localhost:11434
RAVEN_AUTHORIZED=true

Quick Start

Show commands:

uv run raven --help

Run reconnaissance:

uv run raven scan --target 192.168.1.0/24 --profile normal

Auto-detect local subnet(s):

uv run raven scan

Resume interrupted scan:

uv run raven resume --session <session_id>

Update local CVE database:

uv run raven update-cve-db

Run exploitation phase on saved recon:

uv run raven exploit --session <session_id>

Example Output (abbreviated)

RAVEN starting... profile=quick
RAVEN Target Resolution
manual | - | 127.0.0.1 | - | -
Recon complete: live_hosts=1, total_open_ports=49
Saved MD report: ~/.raven/runs/recon-<id>/report.md
Saved HTML report: ~/.raven/runs/recon-<id>/report.html
Saved JSON report: ~/.raven/runs/recon-<id>/report.json

Safety Notes

• Only scan assets you are explicitly authorized to test.
• Use RAVEN_AUTHORIZED_SCOPE to restrict allowed targets.
• Exploitation requires HITL approval for each action.