MIT License · Open Source · Go

Open source server security
you can audit

The agent that runs as root on your server is fully open source. Every detection pattern, every firewall rule, every malware signature — auditable on GitHub under the MIT license.

Why open source matters for security

You are installing software that runs as root on your server. It reads your authentication logs, modifies your firewall, scans every file on disk, and communicates with an external API. That is an extraordinary level of trust.

With proprietary security agents like BitNinja or Imunify360, you cannot see what the binary does. You cannot verify it does not exfiltrate data, phone home to third parties, or open backdoors. You trust the vendor blindly. With Defensia, every line of detection logic is public. Security researchers, your team, or anyone can audit the code before deploying it. That is how security software should work.

MIT
License — use, modify, distribute
100%
Agent detection logic is public
0
Closed-source dependencies

What's open source, what's not

We believe in transparency. Here is the exact boundary between open source and commercial components.

OPEN SOURCEMIT License

The Agent

  • SSH brute force detection engine (15 patterns)
  • Web application firewall (15+ OWASP attack types)
  • Malware scanner (64K+ hashes, 684 dynamic signatures)
  • CVE vulnerability matching
  • Firewall rule management (iptables/ipset)
  • Log watchers (auth, nginx, Apache, Docker)
  • Security posture scoring
  • Bot detection and geoblocking logic
  • All Go source code on GitHub
COMMERCIALSaaS

The Dashboard

  • Web dashboard (Vue 3 + Laravel)
  • API server (event storage, configuration sync)
  • Database (MySQL, event history, ban records)
  • WebSocket server (real-time updates)
  • Alert routing (Slack, email, Discord, webhooks)
  • Threat intelligence feeds and CVE database
  • Team management and billing

Free tier: 1 server. Pro: €9/server/month.

This is the same model used by GitLab (Community Edition vs Enterprise), Grafana (OSS vs Cloud), and Elastic (Basic vs Cloud). The critical security logic that runs on your infrastructure is open. The management layer is a hosted service.

The agent: technical details

A single statically compiled binary. No runtime dependencies. No Python, no Java, no Node.js on the target server.

Agent specifications
LanguageGo 1.26
Binary size~40 MB
Runtime memory<30 MB
Memory during scans<50 MB
Architectureslinux/amd64, linux/arm64
CompilationCGO_ENABLED=0 (static)
LicenseMIT
Dependencies on targetNone (static binary)

Source code: github.com/defensia/agent. Supports linux/amd64 (standard VPS) and linux/arm64 (AWS Graviton, Raspberry Pi, Oracle Ampere).

What the agent does on your server

Full transparency. Here is exactly what the agent reads, writes, and connects to. No hidden behavior.

Reads

  • auth.log / journald (SSH authentication events)
  • nginx and Apache access logs (HTTP requests)
  • mail.log (Postfix/Dovecot authentication)
  • Docker container logs (if Docker present)
  • System package lists (for CVE matching)

Writes

  • iptables / ipset rules (ban malicious IPs)
  • /var/lib/defensia/ (quarantine, state files)
  • systemd service file (auto-restart on boot)

Connects to

  • defensia.cloud API over HTTPS (heartbeat, events, config sync)
  • No third-party connections
  • No telemetry to external services

Sends to dashboard

  • Security events (attack type, source IP, timestamp)
  • Server metrics (CPU, RAM, disk usage)
  • Agent version and OS information
  • Never sends raw log contents or file data

What the agent does NOT do

  • ✗ Send raw log contents to the dashboard
  • ✗ Access file contents (except during malware scanning of known-suspicious paths)
  • ✗ Mine cryptocurrency or run background computation
  • ✗ Phone home to third parties — only connects to defensia.cloud API
  • ✗ Install additional software or dependencies

How to build from source

Clone the repository, build with Go, run on any Linux server. No proprietary toolchain required.

Build from source

# Clone the repository

git clone https://github.com/defensia/agent.git

cd agent

# Build for your architecture (amd64)

CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -o defensia-agent ./cmd/defensia-agent

# Or build for ARM64 (Graviton, Raspberry Pi)

CGO_ENABLED=0 GOOS=linux GOARCH=arm64 go build -o defensia-agent ./cmd/defensia-agent

# Cross-compile with Docker (from macOS/Windows)

docker run --rm -v "$(pwd):/app" -w /app golang:1.26-alpine \

  sh -c "CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -o build/defensia-agent ./cmd/defensia-agent"

The build produces a single binary with no external dependencies. Copy it to your server, register with your Defensia account token, and it starts protecting immediately. Or use the one-line installer: curl -fsSL https://defensia.cloud/install.sh | sudo bash

How to contribute

The agent is an active open source project. Contributions are welcome.

Report bugs

Open an issue on GitHub with steps to reproduce. Include agent version and OS details.

Submit pull requests

Fork the repo, create a branch, submit a PR. New detection patterns, performance improvements, and bug fixes are all welcome.

Fork and customize

The MIT license means you can fork, modify, and distribute the agent — even for commercial use. No restrictions.

Open source vs proprietary: comparison

How does Defensia compare to other server security tools on openness, licensing, and architecture?

ToolLicenseAgent open?DashboardSelf-hosted?Language
DefensiaMITYes (full)SaaS (free tier)NoGo
CrowdSecMITYesSaaSPartialGo
fail2banGPL v2YesNoneYesPython
WazuhGPL v2YesSelf-hostedYes (heavy)C / Python
WordfenceGPL v2WP pluginWP adminYesPHP
BitNinjaProprietaryNoSaaSNoUnknown
Imunify360ProprietaryNocPanel pluginNoPython / C

Defensia is the only tool that combines an MIT-licensed agent with a full web dashboard, covering SSH, WAF, malware scanning, CVE detection, and bot management in a single binary. fail2ban covers SSH only. CrowdSec requires paid blocklists for full protection. Wazuh requires a multi-node Elasticsearch cluster. BitNinja and Imunify360 are entirely proprietary.

Frequently asked questions

Is the Defensia agent fully open source?

Yes. The agent is MIT licensed and all detection logic is in the public GitHub repository at github.com/defensia/agent. This includes SSH detection patterns, WAF rules, malware scanner, CVE matching, firewall management, and every other feature that runs on your server.

Can I use the agent without the dashboard?

The agent requires the dashboard API for configuration and event storage. It connects to defensia.cloud to register, receive configuration updates, and report security events. A self-hosted dashboard is not available yet.

Can I modify the agent?

Yes. The MIT license allows modification, redistribution, and commercial use with no restrictions. You can add custom detection patterns, modify scoring thresholds, or fork the entire project for your own use.

Is the dashboard open source?

No. The dashboard is a commercial SaaS built with Laravel and Vue. Free tier includes 1 server with SSH protection and the full dashboard. Pro costs €9/server/month and adds WAF, malware scanning, CVE intelligence, geoblocking, bot management, and alerts.

How do I report a security vulnerability?

Contact security@defensia.cloud or open a private GitHub security advisory on the agent repository. We follow responsible disclosure practices and will credit researchers who report valid vulnerabilities.

What data does the agent send to the dashboard?

Security events (attack type, source IP, timestamp, matched pattern), server metrics (CPU, RAM, disk usage), and agent version information. The agent never sends raw log contents, file data, or any information unrelated to security monitoring.

Sources and references

Try the open source agent

One command. Under 30 seconds. Free for one server. Audit the code first if you want.

$ curl -fsSL https://defensia.cloud/install.sh | sudo bash

No credit card required.