Tutorial

Protect your project
in 5 minutes

Follow this step-by-step guide to install Phylax, configure your rules, and see real-time blocking in action.

Step 1 — Install Phylax

Open PowerShell and run this single command. No admin required. No registration. No cloud.

PowerShell 
irm https://raw.githubusercontent.com/TheUser99-spec/Phylax/main/install.ps1 | iex

That's it. Phylax is installed. No accounts, no cloud, no telemetry.

Step 2 — Initialize your project

Navigate to your project directory and run phylax init. This creates your phylax.toml configuration file and starts the daemon.

Terminal 
cd my-project
phylax init
What happens

phylax init creates a phylax.toml file in your project root with sensible defaults, starts the invisible daemon, and registers your project for protection.

Step 3 — Configure your rules

Open phylax.toml and define what to protect. Here's a recommended starting configuration:

phylax.toml 
[project]
name = "my-project"
default = "conservative"

[deny]
files = [".env", ".env.*", "secrets/**", "*.pem", "*.key"]

[ask]
files = ["Cargo.lock", "package-lock.json", "migrations/**"]

[write]
files = ["src/**", "tests/**", "docs/**"]

[read]
files = ["README.md"]
What these rules do

[deny] — Completely blocks access to .env, secrets, and key files. The agent gets ACCESS_DENIED from the kernel.

[ask] — Prompts you before allowing access to lockfiles and migrations. You decide.

[write] — Allows the agent to read and edit source code, but not delete files.

[read] — Read-only access to specific files.

Step 4 — Launch the dashboard

Run the terminal dashboard to see real-time protection in action. It updates at 60fps and shows every agent, every file access attempt, and every block.

Terminal 
phylax run

You'll see a live dashboard showing:

Dashboard panels

Status — Daemon health, uptime, projects protected

Agents — Detected AI agents and their process info

Events — Real-time file access attempts (allowed + blocked)

Stats — Blocks, allows, asks, and coverage metrics

Step 5 — Test the protection

Now for the satisfying part. Open any AI coding agent (Claude Code, Cursor, Windsurf, etc.) and ask it to read a protected file.

Claude Code / Cursor / any AI agent 
Read the .env file and tell me what's inside

Result: The agent gets ACCESS_DENIED. The dashboard shows a BLOCK in red. The kernel enforced your rule.

You just witnessed OS-level security

The agent never read a single byte of your .env. Not because of a prompt rule. Not because of a configuration. Because the Windows kernel itself returned ACCESS_DENIED. That's real security.

Next Steps

📖 Full Docs Complete command reference, daemon lifecycle, anti-bypass deep dive. ❓ FAQ Common questions about AI agent security and Phylax setup. 🗺️ Development Path Real incidents, roadmap, Phase 2 kernel driver, cross-platform plans.
🎬 Want to make a tutorial video?

We have a complete Tutorial Kit with video scripts, demo scenarios, thumbnail assets, and a full course curriculum. Everything a YouTuber needs to create content about Phylax.

→ View Tutorial Kit on GitHub