Get Started with Vreko

Install the CLI

npm install -g @vreko/cli

Initialize your project

cd your-project
vr init

vr init scans your codebase and starts the Vreko service:

$ vr init
◇ Scanning workspace…
  ├─ indexing files         ████████████████  1,284 files
  ├─ detecting frameworks   next.js, drizzle, vitest
  └─ wiring agent context   CLAUDE.md
◇ Vreko service started
✓ Workspace ready  -  agents.workspace.json created

Vreko is now watching your codebase. Run your next AI prompt - Vreko is already informing the context.

Install the VS Code extension

Install Vreko from the VS Code Marketplace (extension ID marcelle-labs.vreko) for ambient status-bar signals and file decorations:

Add VS Code Extension →

The extension is optional - the CLI alone gives you the full intelligence layer and automatic agent delivery.

What just happened

Vreko detected your agent setup automatically. A pointer to .vreko/docs/INTELLIGENCE.md has been added to your CLAUDE.md, .cursorrules, or project instructions file - that’s what your agent reads on every prompt now.

It’s probably short right now. Vreko is still learning your codebase. Over the next few weeks it will surface fragile files, AI drift zones, and co-change patterns specific to your code.

If the block is empty or missing after a few sessions, run vr status to verify the service is running.

What to Expect

When	What’s Happening
Right away	Vreko detects your agent setup and starts writing intelligence into your rules file
First week	Vreko observes which files change together and which churn under AI edits; it starts building a picture of your codebase
Day 30	Patterns emerge. Vreko starts knowing your codebase, not just watching it
Day 90	Your agent sees fragile files before it touches them - delivered automatically on every prompt

Connect your AI agent

Vreko delivers intelligence to your agent through your project’s rules file automatically. To let your agent call Vreko’s tools directly, connect the MCP server - see MCP Setup for Claude Code and Cursor.

Where to Go Next

How It Works - What Vreko actually detects and why
MCP Setup - Connect Vreko’s tools to your AI agent
CLI Reference - Full command reference
Pioneer Program - Beta benefits and community perks