This page explains how BastionGuard thinks about security. It documents the threat model, design principles, and assumptions behind BastionGuard’s behavior.
If you are looking for predictable behavior, explicit policies, and visible decisions, this page describes exactly what BastionGuard is designed to protect — and what it deliberately does not attempt to do.
Design Philosophy
Security must be understandable, observable, and controllable.
BastionGuard is not a “trust us” security product. It is a security control plane for Linux desktops, where every protection mechanism is explicit, deterministic, and inspectable.
Many security products rely on opaque engines, cloud scoring, or silent automation. BastionGuard intentionally avoids these approaches.
What BastionGuard Assumes
- The user wants visibility into security decisions
- Silent background actions are undesirable
- Security policies should be inspectable and reversible
- Local systems should not depend on external trust services
Threat Model
BastionGuard addresses a specific and clearly defined set of threat classes. Each one is handled using explicit rules and visible enforcement.
Malware & File-Based Threats
- Real-time on-access scanning
- Manual and scheduled scans
- USB device scanning
- Samba and network share scanning
Ransomware & Destructive Behavior
- YARA-based detection
- SaneSecurity rules
- Explicit allowlists and false-positive controls
Phishing & Malicious Domains
- Deterministic phishing blacklists
- DNS-based enforcement
- Optional firewall IP blocking
- User-controlled whitelists
Identity Exposure
- On-demand breach checks
- Clear source attribution
- No continuous background surveillance
Secure Payments
- Sandboxed secure browser
- Explicit domain allowlists
- No wildcard trust
Privacy: Webcam & Microphone
- Explicit access control
- No assumed trust
Deterministic Behavior
BastionGuard favors deterministic rules over probabilistic models. Every security decision can be reasoned about and reproduced.
- No machine-learning guesswork
- No hidden scoring systems
- No silent remediation
Open Source Transparency
BastionGuard is fully open source. This is a deliberate security decision, not a marketing feature.
- Auditable implementation
- Verifiable runtime behavior
- No proprietary or opaque components
- No hidden cloud dependencies
Support Scope
This document is part of BastionGuard support documentation. It defines expected behavior and design intent.
If behavior differs from what is described here, it should be treated as a bug or misconfiguration.