INK – Cloud Infrastructure for AI Agents

TL;DR

TL;DR: INK (deployink.com) is cloud infrastructure built specifically for AI agents—deploy services, databases, and templates through the CLI, Skills, or MCP without managing Dockerfiles or CI pipelines.

What Is INK?

INK is a cloud platform designed from the ground up for AI agents. Instead of wrestling with infrastructure configuration, agents can deploy full-stack apps, databases, and APIs through a unified interface using CLI commands, Agent Skills, or MCP (Model Context Protocol).

“Every service built for AI coding agents to operate end to end through CLI and skills.”

INK auto-detects 30+ frameworks including Next.js, React, Node.js, Python (Django, Flask), Go, Vue.js, Rust, and more. Agents push code via INK git or GitHub, and INK handles the rest—SSL certificates, custom domains, logs, and scaling.

Setup Workflow

Prerequisites

• An INK account (get $5 trial credit, no credit card required)
• A supported AI agent client (Claude Code, Cursor, Windsurf, etc.)
• Git installed locally

Step 1: Install the INK CLI

# Install via npm
npm install -g @ink-dev/cli

# Or via curl
curl -fsSL https://get.deployink.com | sh

Step 2: Authenticate

ink login

This opens a browser for OAuth authentication. After login, your API credentials are stored locally.

Step 3: Deploy Your First Service

# Deploy from the current directory
ink deploy

# Or specify a framework explicitly
ink deploy --framework nextjs --region us-east

INK auto-detects the framework, provisions the right runtime, and returns a live URL within seconds.

Step 4: Connect via MCP

INK provides an MCP server that AI agents can use to manage infrastructure:

{
  "mcpServers": {
    "ink": {
      "command": "npx",
      "args": ["@ink-dev/mcp", "server"]
    }
  }
}

Once connected, agents can run commands like fn service_create, fn service_get, and fn template_deploy directly.

Step 5: Deploy a Database

# Deploy a PostgreSQL instance
ink template deploy --name postgresql

# Deploy Redis
ink template deploy --name redis

Available templates include PostgreSQL, Redis, MySQL, and MongoDB.

Supported AI Clients

INK integrates directly with these AI coding agents:

• Claude Code (Anthropic)
• Cursor (via MCP)
• VS Code (Agent Mode)
• Windsurf (via MCP)
• Gemini CLI (Google)
• Goose (Block)
• Kimi Code (Moonshot)
• Open Code / Cline / KiloCode
• Antigravity

Supported Frameworks

Auto-detected frameworks include:

| Category | Frameworks | |---|---| | Full-stack | Next.js, Nuxt, Rails, Laravel, Django, Spring Boot | | Frontend | React, Vue.js, Svelte, Angular, HTML/Static | | API/Backend | Express, FastAPI, Flask, Go, Rust, Elixir, Bun, Deno | | Python ML | Streamlit, Gradio | | Other | Ruby, PHP, .NET/C#, Haskell, Zig |

Pricing

INK charges based on actual compute, storage, and bandwidth usage. A $5 trial credit is available without a credit card. Agents can query their balance and usage through the CLI or MCP to avoid surprise costs.

Source and Accuracy Notes

• Project page: deployink.com
• Short link: ml.ink (redirects to deployink.com)
• HN launch thread: news.ycombinator.com/item?id=47337028 (32 points, Show HN)
• License: Proprietary SaaS (no open-source version detected)
• Source last checked: 2026-08-04

FAQ

Q: Is INK open source? A: No. INK is a proprietary cloud platform. There is no open-source self-hostable version.

Q: Does INK require Docker? A: No. INK auto-detects frameworks and handles containerization internally. No Dockerfiles or Docker knowledge required.

Q: Can I use INK without an AI agent? A: Yes. The INK CLI works entirely via human operators. But the platform is designed with agent-first workflows in mind.

Q: What happens when the $5 trial credit runs out? A: You’ll be charged for actual usage at published rates. Agents can query remaining balance via ink workspace list or the MCP interface.

Q: Does INK support custom domains? A: Yes. Use ink domain add to attach a custom domain with automatic SSL provisioning.

Conclusion

INK fills a specific niche: infrastructure that AI agents can actually operate autonomously. Instead of handing an agent a raw cloud account, INK provides a controlled surface—databases, templates, logs, and billing—that agents can navigate through a clean CLI or MCP interface.

If you run AI coding agents in production and want them to deploy services without human infrastructure toil, INK is worth evaluating. The MCP integration is the key differentiator—agents can manage the full deployment lifecycle without leaving their toolchain.