Skip to main content
Krilo Labs
How Krilo Ship works

From AI prompt to live URL in 90 seconds.

A complete walkthrough of how Krilo Ship turns AI-generated code into production deployments.

The 4-step flow

Each step is a single MCP tool call. The agent never touches infrastructure directly.

1

1. AI calls krilo_ship

Your agent calls a single MCP tool with the target domain, the build artifact, the region, and (optionally) any secrets your app needs at runtime. No clicking, no copying API keys, no DNS dashboard. The agent passes intent — Krilo translates intent into infrastructure changes. Tool schema is typed, idempotent, and AI-readable: failure modes are returned as structured errors the agent can interpret and retry.

2

2. Krilo provisions

In parallel: the domain is registered (or transferred), a VPS is provisioned in your chosen region, an SSL certificate is requested via Let's Encrypt, DNS records are written through Cloudflare, and a fresh deploy user is created with a single-use SSH key. Every credential is generated for this deployment and rotated when it's done. Nothing is reused, nothing leaks.

3

3. Code ships through the bastion

Your build artifact is uploaded through a Krilo bastion via mTLS — no exposed SSH ports on your VPS, no static credentials anywhere. The deploy user copies, unpacks, and starts your app under a hardened systemd service. Health checks run before traffic is switched. If anything fails, the previous version keeps serving traffic.

4

4. Live URL returned to the AI

The MCP tool returns: live URL, deployment ID, health status, and a rollback handle. Your agent can immediately verify the deploy with an HTTP check, link the URL into a ticket, or kick off the next iteration. No human in the loop unless something asks for one.

What you get with each deploy

  • Hetzner or DigitalOcean VPS (your size choice)
  • Custom domain registration (.com, .dev, .io, .ai, .app, and 30+ more)
  • Auto SSL via Let's Encrypt with auto-renewal
  • Cloudflare DNS management + DDoS protection
  • Automatic backups (last 7 days, daily snapshot)
  • Real-time deployment logs streamed back to the agent
  • One-click rollback to any prior version
  • Multi-region support (Singapore, Frankfurt, US-East, more on request)

Honest, transparent costs

Krilo Ship is a reseller model. You pay actual infrastructure costs (VPS, domains) at small markup, plus a service fee per deployment. No hidden fees. No subscription if you don't need one.

Want exact numbers for your use case? Contact us →

Already have your own VPS or domain?

Krilo Ship works with your existing infrastructure. Connect your Hetzner account, your Namecheap domain, your Cloudflare DNS — pay only the deployment service fee.

This is your infrastructure. We just make it agent-friendly.

Frequently asked questions

Vercel and Netlify are managed PaaS — you pay them to manage your code on their infrastructure. Krilo Ship is a reseller — you own the VPS, the domain, the deployment. We just make it AI-callable. Move-out is free, transparent, and supported.
No. Krilo Ship handles all DevOps automatically. Your AI agent calls a single function, and we handle provisioning, SSL, DNS, deployment, and monitoring. You see the live URL.
Your sites stay up. Krilo Ship is a control plane — once your code is deployed to your VPS, it runs independently. We monitor uptime, but your site doesn't depend on Krilo being online.
Yes. We support full data export, domain transfer, and VPS migration. Your infrastructure is yours — we just made it easier to deploy to.
Any AI tool that supports MCP (Model Context Protocol). This currently includes Claude Code, Cursor, Bolt, Replit, and any custom agent built on the MCP SDK. We're adding more integrations as the standard grows.
Yes. Your secrets are passed via single-use secure URLs and never stored on Krilo's servers. AI agents communicate with deployment-specific scoped APIs, not raw infrastructure access.
Typical first-time deployment: 60-120 seconds. Subsequent deployments to the same site: 15-30 seconds. We optimize aggressively for AI-driven workflows where deploys happen frequently.
Yes. We provide a CLI (krilo ship) and a web dashboard for direct human use. The MCP integration is the killer feature, but it's not required.

Get notified when we launch

We're shipping in Q3 2026. Join the waitlist for early access and a 50% discount on your first 3 months.