Getting Started

Cyberstrike is an AI-augmented offensive security harness. You describe a target — a web app, an API, a network — and Cyberstrike runs the assessment. It picks the tools, dispatches the right agents, tests for vulnerabilities, and reports what it finds.

Think of it as an AI pentester that works alongside you. You stay in control through a single conversational interface while the harness handles the heavy lifting: crawling, traffic analysis, injection testing, authorization checks, and finding chains across individual results.

Under the hood, it brings together 100+ security tools, 15+ specialized agents, 65+ deterministic test scripts based on the OWASP WSTG, plus dedicated engines for web crawling, traffic analysis, containerized tools, and external integrations.

System Requirements

Before installing Cyberstrike, verify your system meets these requirements:

Installing Cyberstrike

Choose your preferred installation method based on your operating system and package manager.

npm (Recommended)

1. Open your terminal
2. Run the installation command:

npm install -g @cyberstrike-io/cyberstrike@latest

3. Verify the installation:

cyberstrike --version

Bun Coming Soon

Bun support is not yet available.

Homebrew (macOS) Coming Soon

Homebrew support is not yet available.

curl (Linux/macOS)

Run the installation script:

curl -fsSL https://cyberstrike.io/install | bash

iwr (Windows/Powershell)

Run the installation script:

iwr -useb https://cyberstrike.io/install.ps1 | iex

Docker

Connecting an AI Provider

Cyberstrike requires an AI provider to function. The following providers are supported:

Setting Up Authentication

1. Launch Cyberstrike and run the connect command:

/connect

2. Select your provider from the list and enter your API key (or authorize via OAuth for GitHub Copilot)

3. Switch models anytime with:

/models

Using Environment Variables

Alternatively, set API keys as environment variables:

# Anthropic
export ANTHROPIC_API_KEY="sk-ant-..."

# OpenAI
export OPENAI_API_KEY="sk-..."

# Google
export GOOGLE_GENERATIVE_AI_API_KEY="AI..."

# OpenRouter
export OPENROUTER_API_KEY="sk-or-..."

Add these to your shell profile (~/.bashrc, ~/.zshrc) for persistence.

Running Your First Assessment

Launch

cyberstrike

The TUI appears with an input prompt, response area, and status bar.

Describe the Target

Type your task directly. The harness decides what to run:

> Test https://app.example.com for OWASP Top 10 vulnerabilities

What happens under the hood:

1. The cyberstrike primary agent reads your goal
2. /hackbrowser launches to crawl the target and capture HTTP traffic
3. Vortex processes each request — applies rules, flags anomalies
4. The wstg-agent dispatches specialist agents based on request signals
5. 13 WSTG specialists test for injection, auth bypass, IDOR, session flaws, crypto issues, and more
6. The escalation-hunter chains findings to detect compound vulnerabilities
7. The triage agent independently verifies each finding by replaying HTTP evidence
8. You get triaged findings with evidence, severity, and remediation

What’s in the Harness

You interact with one prompt. The harness orchestrates the rest.

Configuration

Create a cyberstrike.json file in your project directory to set defaults:

{
  "$schema": "https://cyberstrike.io/config.json",
  "model": "anthropic/claude-sonnet-4-20250514"
}

When you run cyberstrike in this directory, it automatically loads these settings. You can also switch models anytime with /models in the TUI.

Configuration Hierarchy

Cyberstrike loads configuration from multiple sources in this order:

1. Environment variables (highest priority)
2. Project config (cyberstrike.json)
3. User config (~/.config/cyberstrike/cyberstrike.json)
4. Default values (lowest priority)

For detailed configuration options, see the Configuration Reference.

Key Slash Commands

Keyboard Shortcuts

See Keyboard Shortcuts for the full reference.

Next Steps

Continue learning with these guides:

1. Configuration Reference — Customize Cyberstrike settings
2. AI Providers — Configure your preferred AI provider
3. AI Models — 9,000+ models across 230+ providers
4. Security Agents — Learn about specialized agents
5. Agentic Pipelines — End-to-end assessment workflows for web, cloud, and more
6. Features — HackBrowser, Vortex, Bolt, MCP