March 27, 2026·4 min read

Miget Agent Skills: Manage Infrastructure from Your AI Editor

Miget Team

Miget Agent Skills: Manage Infrastructure from Your AI Editor

Your AI coding assistant just got a new capability - full knowledge of the Miget API. With one command, your editor can deploy apps, provision databases, configure domains, and debug failed deploys without you ever opening a dashboard.

What is agent-skills?

agent-skills is an open source package that teaches AI coding assistants how to work with the Miget platform. It follows the Agent Skills Open Standard - a universal format that works across Claude Code, Cursor, Windsurf, and other AI-powered editors.

The skill file contains structured knowledge of every Miget API endpoint, authentication flow, resource model, and common workflow. When installed, your AI editor can call the Miget API on your behalf using natural language instructions.

Installation

One command in your project root:

npx skills add migetapp/agent-skills

Then set your API token so the editor can authenticate:

export MIGET_API_TOKEN="miget_api_xxxxxxxxxxxxx"

Generate a token at app.miget.com/my_account#api_tokens if you don't have one yet.

What you can do

Once installed, talk to your AI editor like you would talk to a DevOps engineer:

Deploy an app from GitHub:

"Create a new Node.js app called api-service, connect it to my GitHub repo acme/api on the main branch, and deploy it to my production resource."

Add a database:

"Add a PostgreSQL addon to my web app and inject the connection string as DATABASE_URL."

Debug a failed deploy:

"My last deploy on staging failed. Check the build logs and tell me what went wrong."

Scale resources:

"Upgrade my production resource to the pro plan and increase RAM to 2GB."

The editor handles the API calls, waits for async operations to complete, and reports back with results and next steps.

How it works

The skill file is structured around three layers:

Behavioral guidelines - Rules the AI follows before making any API call. It checks for your API token, asks for required fields instead of guessing, and validates dependencies before creating resources.
Core concepts - The Miget resource model: workspaces, resources (compute), projects, apps, services, buckets, and how they relate to each other.
API reference - Every endpoint with method, path, request body, and response format. Authentication, error handling, and common workflows included.

This means the AI doesn't just know the endpoints - it understands when to use them and how to chain them together for real-world tasks.

Works with your editor

The skill is compatible with any editor that supports the Agent Skills standard:

Claude Code - auto-detects installed skills
Cursor - loads skills from project directory
Windsurf - picks up skills via Cascade

No editor-specific configuration needed. Install once, works everywhere.

Open source

The agent-skills repo is fully open source. You can inspect the skill file, suggest improvements, or fork it for your own use case. It lives at github.com/migetapp/agent-skills.

The skill stays in sync with the Miget API - as new features ship, the skill file is updated automatically.

Get started

Run npx skills add migetapp/agent-skills
Set MIGET_API_TOKEN in your shell
Ask your editor to deploy something

For the full guide, visit the Miget x AI page or read the documentation.