Raven Documentation¶
Raven is a penetration testing framework. Run the engagement, not just the report. It's free to download and free to run, and everything here works without AI. The AI layer is an optional add-on that draws from a credit balance.
Raven bundles the usual scanners (nmap, nuclei, and friends), engagement/findings management, and an AI advisor that reads your engagement data and suggests next steps, but never decides a finding for you. It runs as a web dashboard and a terminal REPL, both talking to the same server.
Start here¶
| Guide | What it covers |
|---|---|
| Quick Start | From download to your first finding |
| Install & Pair | One-line, Docker, and bare-binary installs, plus device pairing |
| CLI / REPL | The terminal operator client and its commands |
| Web Dashboard | Driving Raven from the browser |
| Testing Windows | Enforcing client-approved testing times |
| Scanning | Running and orchestrating the integrated tools |
| AI Analysis | The AI advisor, chat, and analysis |
| Offline Mode | What works without a connection, and what needs one |
| Troubleshooting | Pairing, AI errors, scanning, and dashboard fixes |
How it fits together¶
- Raven server: the engine + web dashboard. Run it via the Pro Docker image (toolchain bundled) or the bare
raven-serverbinary (you bring the tools, e.g. on Kali). - Raven REPL: the Claude-Code-style terminal client. Connects to any Raven server and drives the same actions as the web UI, keyboard-native.
- Chamber Door: the account + AI layer. You pair a device once (
raven-server auth login, ordocker exec -it raven-pro raven auth loginin Docker); AI requests proxy through Chamber Door, which meters them against your credit balance. Pairing is only needed for AI; everything else runs without it. Engagement data is anonymized before it leaves your machine.
Beta
These docs cover the closed beta. Raven itself is free to run; only AI draws credits. If something here doesn't match what you see, ping us.