AI-native security testing for Dockerized apps

Catch security issues before you get hacked.

CyberDuty combines AI-native security scanning and a remedy dashboard into one developer-friendly CLI. Give your developers fast, repeatable security scanning built for the AI age.

Active Session · Agent Active
0High
0Medium
0Low
0Info

"If you don't use a tool like CyberDuty, you will get hacked sooner or later. We have it embedded in our CI & dev process, so we can ship with confidence."

Sameer CTO, SpeedBank
✓ Verified CTO
Why traditional pentesting isn't enough

MPT was built for a world that no longer exists.

When software shipped quarterly, an annual penetration test made sense. Today, engineering teams ship dozens of times a week. Traditional MPT doesn't just fall short — it creates a dangerous illusion of security while your real attack surface grows undetected between sprints.

The deployment–security gap

Your annual pentesting is pointless.

You ship every sprint. Your pentest happens once a year. Every feature or config change in between is a business liability, or worse, a company killer.

Deploys this year
◀ Last pentest Today ▶
This gap is your exposure window
  • Every sprint widens the security gap. New routes, dependencies, APIs, container images can introduce vulnerabilities the last pentest has never seen.
  • Enterprise deals, compliance audits, and cyber insurers want ongoing evidence. A year-old PDF document is not a security program. Timestamped security scans give the evidence you need.
  • Fixing is cheapest when context is fresh. CyberDuty lets the engineer who shipped the change see the security issue and fix it before a malicious actor finds it.
before every release after auth changes before sec audits insurance renewals
For CTOs and engineering leaders

Know your risk before your customers do.

CyberDuty: when the code changes, security gets checked. No waiting for an annual report. No vague scanner dump that nobody owns or understands.

Fix early Catch vulnerabilities while the dev still has context and the fix is cheap.
Save money Enterprise-grade security checks without the security headcount.
Audit ready Timestamped reports that show what was tested, and what was found.
What you get as a leader
  • $
    Stop absorbing breach costs.Problems found in local development cost a fraction of problems found by customers, regulators, or attackers.
  • Clear ownership, not chaos.The engineer who shipped the change sees the finding immediately — no ticket lost in a security backlog.
  • Security becomes a habit.Run CyberDuty before every release. It becomes invisible — until it catches something that would have been very expensive in production.
  • Evidence for everyone who asks.Show board members, customers, insurers, and auditors that security is embedded in your delivery process.
Simple by design

Docker ready. No new infrastructure.

CyberDuty works with the Docker setup your team already uses. Just install the CLI and get going. There's nothing to provision, no agents to deploy, no cloud to configure.

Step 01

Your app, running normally

Start your application in Docker or Docker Compose — exactly as your team already does.

Step 02

One command to scan

Point CyberDuty at the running container and tell it what to test. Everything runs automatically.

Step 03

Prioritized findings

Open the local dashboard to review risks by severity, each with clear evidence and remediation.

Step 04

Ship with evidence

Fix what matters and ship with confidence. Make security real-time, not an afterthought.

AI-native security

Let Claude scan while you build.

CyberDuty fits your Claude Code workflow in two ways — you invoke it, or Claude does.

In the Claude terminal

Run cyber pentest directly inside a Claude Code session. Claude reads the structured output and proposes fixes before you've switched tabs.

As an MCP tool

Register CyberDuty as an MCP server. Claude invokes cyber_pentest automatically during code review — structured results come back, remediations go straight into your editor. No copy-paste. No context switch.

CLI Mode
Claude Code Terminal
$cyber pentest payment-api --port 3000
✓ ZAP baseline scan complete
✓ Nuclei template scan complete
CRITICALJWT secret hardcoded in config.js:12
HIGH Missing CORS policy on /api/checkout
HIGH SQL injection vector in /api/users
Claude ›I see 1 critical and 2 high findings.
         Let me fix config.js:12 first…
MCP Server
Claude is invoking a tool ···
cyber_pentest({
  container: "payment-api",
  port: 3000,
  engines: ["zap","nuclei"]
})
← Tool result · 3 findings
CRITICALconfig.js:12 — hardcoded JWT secret
HIGH /api/checkout — CORS misconfiguration
Claude ›Applying fix to config.js — review
         the diff in the editor panel →
CLI Reference

One command from a security signal.

CyberDuty Scanner runs from your terminal against any Dockerized application. The full reference covers every command, flag, engine option, and configuration setting.

Docker container refs ZAP baseline / full Nuclei templates Local dashboard JSON & HTML export
Quick start
# 1. Install
unzip cyber-duty-scanner.zip && cd cyber-duty-scanner
npm install && npm run build && npm link --workspace @cyber/cli

# 2. Pentest a running container
cyber pentest my-app --port 3000 --engine all

# 3. Open the findings dashboard
cyber dashboard
→ http://localhost:8787
View the full CLI Reference — all commands, flags, and examples
Get started today

Stop guessing. Start scanning.

CyberDuty installs in minutes and runs against any Dockerized application. Get a real security signal before your next release.