# letai

> Kubernetes for coding agents.

**Status:** Beta | **Language:** Go | **License:** MIT
**Repo:** https://github.com/liwala/letai

## The Problem

Running AI agents with unrestricted access on your host machine is a security incident waiting to happen. There's no hierarchy, no isolation, no coordination.

## Architecture

```
Human (you)
  └── CTO orchestrator (host, all-seeing)
       └── VP orchestrator (per project)
            ├── Producer (implements in VM)
            ├── Reviewer (reviews in VM)
            └── Tester (tests in VM)
```

## What It Does

- Hierarchical CTO → VP → Worker agent structure
- VM isolation (Lima, Docker, Podman, OrbStack)
- Network hardening with iptables and domain allowlists
- Git-based coordination protocol
- Live agent attachment via tmux
- Task templates for common workflows
- Session recording via Entire.io integration

## Quick Start

```bash
git clone https://github.com/liwala/letai && cd letai
make build

letai init                  # setup wizard
letai up my-app             # start hierarchy
letai task dispatch my-app "Add auth"  # dispatch
letai watch                 # monitor everything
```

**Supports:** Claude Code, Codex, Amp, any CLI-based agent