Accepting early access requests

Dev environments for
AI software development.

With AI writing most of the code, you need isolated, reproducible workstations - not your laptop. werkr gives you one in seconds.

The problem

AI is writing the code.
Your environment wasn't built for that.

50 to 90% of code is now AI-generated. But the environments developers work in haven't changed. That's a problem.

Your laptop isn't a sandbox

AI coding agents need filesystem access, network access, and credentials to do their job. Running them directly on your machine means sharing everything - your files, your keys, your identity.

VDI and hand-rolled VMs don't scale

Enterprise teams are duct-taping solutions - VDI desktops, manually provisioned VMs, locked-down containers. None of them were designed for AI-assisted development, and none of them are turnkey.

Every AI tool works differently

Claude Code, Codex, Cursor - every AI tool has its own setup, its own auth flow, its own system requirements. Managing this across developers and tools is unsustainable.

The solution

Isolated AI workstations.
Defined in YAML. Ready in seconds.

werkr creates fully isolated workstations with your AI tools, skills, and configurations pre-installed. Define it once, run it anywhere - your laptop or the cloud. One command.

workstation.yaml
# workstation.yaml
apiVersion: werkr.dev/v1alpha1
kind: Workstation
metadata:
name: my-dev
labels:
team: backend
spec:
os:
id: ubuntu-24.04
provider: lima
resources:
cpu: 4
memory: 4
---
apiVersion: werkr.dev/v1alpha1
kind: WorkstationConfig
metadata:
name: ai-dev-tools
spec:
selector:
matchLabels:
team: backend
portForward:
autoForward: true
contents:
devtools:
- claude
- gh
- docker
- www-browser-remote
packages:
- python3
- python3-pip

Capabilities

Built for how software
gets written now

Isolated by default

Every workstation is a full VM with its own filesystem, network, and credentials. AI agents can't touch your laptop. That's the point.

AI tools, pre-configured

Claude Code, Codex, Cursor - declare them in your config and they're installed, authenticated, and ready. Skills are placed where each tool expects them.

Local or cloud, same config

Run on your laptop today. Switch the provider to GCP, AWS, or Azure when you need more power. Same YAML, same tools, same experience.

Skills, managed centrally

Define approved skills and tool configurations once. They're automatically deployed to every workstation that matches. No per-developer setup, no drift.

Ephemeral by design

Spin up a workstation, do the work, tear it down. The config is in git. Need it again? One command. This is how agent workloads should run.

Free for developers

werkr is free for individual use. Local or cloud, unlimited workstations. No license, no signup, no catch.

Seamless Connectivity

Feels local.
Runs anywhere.

Moving an environment to the cloud shouldn't mean losing the developer experience you depend on. werkr automatically detects running services and transparently forwards connections to your desktop - closing the gap between local and remote development.

Automatic Connection Handling

Services are discovered in real-time and forwarded to your machine. No port mapping, no tunnel scripts, no manual steps.

Transparent & Secure

All traffic routes through encrypted tunnels. Your workflow stays identical whether the environment is next to you or across the world.

Your Stack, Your Way

Docker, Kubernetes, any language or framework - werkr workstations come preconfigured with the tools your project needs, so you're productive from the first minute.

For platform teams

When agents scale,
governance has to follow.

When one developer becomes fifty, and each runs AI agents daily, you need a control plane - centralized configs, access controls, audit trails, and policy enforcement for every workstation and every agent.

See it in five minutes.

Install werkr, spin up an AI workstation, and start coding. It's free.