Deploy from Git, Docker or straight from your coding agent. Apps, databases and workers share one compute plan, with no per-app fees.
Connect your agent
$ npx skills add migetapp/agent-skillsWorks with Claude Code, Codex, Cursor and Copilot.
30 days free on a $29 plan.
A real Heroku bill: ten apps, seven Postgres databases and four Redis instances, $1,321 a month. Almost half of it was add-ons.
See the bill, line by lineThe first replica runs on your resource. Extra replicas are billed per second, only while they run, and the panel shows which is which.
What usage has cost so far this cycle, and where it is heading at the current pace. Set a spend limit and Usage Billing stops right there.
See real setups priced out →Apps, databases and other services share one pool of compute. The Fair Scheduler keeps shifting it toward whatever needs it most, so you never pay for idle headroom.
See how the Fair Scheduler works →With the Miget skill in Claude Code, Codex, Cursor or Copilot, it deploys apps, adds PostgreSQL, MySQL or Valkey, sets variables and domains, and reads logs and metrics.
Your agent can buy a resource or move it to a bigger plan over the API. When a payment needs you, it hands you a checkout link.
Each app, database and pull request preview your agent starts runs on your compute plan, with no per-app fee. A short-lived job can go on Usage Billing instead, billed by the second.
Give it a token scoped to the projects and permissions it needs, and leave out buying if you want. A spend limit caps the cycle's bill.
Connect any MCP client to your workspace, with no skill to install.
An isolated computer where your agent runs and tests code before it ships.
Apps, workers and databases share one resource's CPU and RAM for a flat monthly price. When traffic spikes, autoscaling adds replicas beyond the plan, billed per second only while they run.
See plans →Push to your branch - Miget builds, caches and ships. No CI/CD to configure.
Learn more →Every pull request gets a full-stack preview with its own URL.
Learn more →PostgreSQL, MySQL and Valkey run on your resource, inside its CPU and RAM.
Learn more →Clone an app with its database and data in one click, and test against real conditions.
Learn more →1,000 cron minutes with every paid resource, each run at its own size, with a run history.
Learn more →Migetpacks detect your stack and cache aggressively. Deploys in seconds.
Learn more →Grafana dashboards, metrics and logs, with a free tier in every workspace.
Learn more →A private network for your apps and databases, with internal DNS. Reach it over WireGuard, Tailscale or IPsec.
Learn more →Shipped a bad deploy? Roll back to any previous version in one click.
Learn more →Deploy your existing docker-compose.yml as-is. Multi-service apps, no rewrite.
Learn more →S3-compatible buckets for files, assets and backups, billed for what they hold.
Learn more →Open a shell into a running app over SSH to debug it live, on paid plans.
Learn more →Your traffic reaches Miget over our own network and our own IP addresses, at the nearest of our points of presence.
See the edge network →$ git push origin main miget → Build started (commit a1b2c3d) miget → Detecting stack... Node.js 20 miget → Cache restored (npm, 14s saved) miget → Build complete 8.2s miget → Deploying to production miget → Live at myapp.eu-east-1.migetapp.com $ git push origin feat/auth miget → Preview created myapp-pr-42.eu-east-1.migetapp.com miget → PR #42 updated with preview link
# Build and push your image $ docker build -t ghcr.io/acme/api:latest . $ docker push ghcr.io/acme/api:latest miget → New image detected latest miget → Pulling image... miget → Health check passed miget → Live at api.eu-east-1.migetapp.com # Or declare your image in app.json { "image": "ghcr.io/acme/api:latest", "formation": { "web": { "quantity": 2, "size": "standard-1x" } } }
# compose.yaml services: web: build: . ports: - "5000:5000" environment: DATABASE_URL: ${DATABASE_URL} worker: build: . command: celery -A app worker db: image: postgres:18 x-miget: managed: postgres
# config/deploy.yml service: my-kamal-app-abc123 image: migetabc/my-kamal-app-abc123 servers: web: - tcp11.eu-east-1.onmiget.com proxy: host: my-kamal-app-abc123.eu-east-1.migetapp.com app_port: 5000 ssl: true registry: server: registry.eu-east-1.miget.io username: migetabc password: - MIGET_REGISTRY_PASSWORD ssh: user: deploy port: 50715 builder: local: true arch: amd64 env: clear: PORT: 5000
{
"name": "my-saas-app",
"env": {
"NODE_ENV": { "value": "production" },
"SECRET_KEY": { "generator": "secret" },
"DATABASE_URL": { "required": true }
},
"formation": {
"web": { "quantity": 2, "size": "standard-1x" },
"worker": { "quantity": 1, "size": "standard-2x" }
},
"scripts": {
"postdeploy": "npx prisma migrate deploy"
},
"buildpacks": [
{ "url": "heroku/nodejs" }
]
}# Deploy an app curl -X POST https://app.miget.com/api/v1/apps/$APP_UUID/deploy \ -H "Authorization: Bearer $TOKEN" \ -d '{"branch": "main"}' # Response { "message": "Application is being deployed. We will notify you once the process is completed." }
→ you: Add a Redis cache to my project and connect it to the web service. → copilot: Added valkey service to your stack. Injected CACHE_URL into web environment. Updated app.json - review the diff: + "addons": ["valkey:7"] + "env": + "REDIS_URL": { "required": true } Deploy now? (y/n) ▊
$ git push origin main miget → Build started (commit a1b2c3d) miget → Detecting stack... Node.js 20 miget → Cache restored (npm, 14s saved) miget → Build complete 8.2s miget → Deploying to production miget → Live at myapp.eu-east-1.migetapp.com $ git push origin feat/auth miget → Preview created myapp-pr-42.eu-east-1.migetapp.com miget → PR #42 updated with preview link
Connect your GitHub repo and Miget handles the rest. Every push to main triggers a production deploy. Every pull request spins up a full-stack preview environment with its own URL.
# Build and push your image $ docker build -t ghcr.io/acme/api:latest . $ docker push ghcr.io/acme/api:latest miget → New image detected latest miget → Pulling image... miget → Health check passed miget → Live at api.eu-east-1.migetapp.com # Or declare your image in app.json { "image": "ghcr.io/acme/api:latest", "formation": { "web": { "quantity": 2, "size": "standard-1x" } } }
Push to any Docker registry - Docker Hub, GitHub Container Registry, or your own - and Miget deploys automatically. Full control over your build, zero lock-in.
# compose.yaml services: web: build: . ports: - "5000:5000" environment: DATABASE_URL: ${DATABASE_URL} worker: build: . command: celery -A app worker db: image: postgres:18 x-miget: managed: postgres
Point Miget at a repo with a docker-compose file and it detects every service. Build-backed apps, standalone images, and managed databases or caches are provisioned together and wired up for you. Every push reconciles the stack.
Read the Compose Stacks docs →# config/deploy.yml service: my-kamal-app-abc123 image: migetabc/my-kamal-app-abc123 servers: web: - tcp11.eu-east-1.onmiget.com proxy: host: my-kamal-app-abc123.eu-east-1.migetapp.com app_port: 5000 ssl: true registry: server: registry.eu-east-1.miget.io username: migetabc password: - MIGET_REGISTRY_PASSWORD ssh: user: deploy port: 50715 builder: local: true arch: amd64 env: clear: PORT: 5000
Kamal delivers zero-downtime deploys using Docker and SSHKit. Point it at your Miget server and deploy Rails, Hanami, or any Ruby app with a single command.
{
"name": "my-saas-app",
"env": {
"NODE_ENV": { "value": "production" },
"SECRET_KEY": { "generator": "secret" },
"DATABASE_URL": { "required": true }
},
"formation": {
"web": { "quantity": 2, "size": "standard-1x" },
"worker": { "quantity": 1, "size": "standard-2x" }
},
"scripts": {
"postdeploy": "npx prisma migrate deploy"
},
"buildpacks": [
{ "url": "heroku/nodejs" }
]
}Declare formation, environment, buildpacks, and deploy hooks in one app.json. Fully compatible with Heroku - migrate your existing config as-is. Push it, and Miget provisions everything.
# Deploy an app curl -X POST https://app.miget.com/api/v1/apps/$APP_UUID/deploy \ -H "Authorization: Bearer $TOKEN" \ -d '{"branch": "main"}' # Response { "message": "Application is being deployed. We will notify you once the process is completed." }
Create apps, trigger deploys, manage databases, and query logs - all through a REST API. Build custom tooling, integrate with your CI, or automate your workflow end to end.
→ you: Add a Redis cache to my project and connect it to the web service. → copilot: Added valkey service to your stack. Injected CACHE_URL into web environment. Updated app.json - review the diff: + "addons": ["valkey:7"] + "env": + "REDIS_URL": { "required": true } Deploy now? (y/n) ▊
Install the Miget agent skill with one command and your AI editor learns the full API. Deploy, scale, debug, and configure - all through natural language.
Learn more about Miget x AI →“Finally a PaaS that leaves you in control of your infrastructure. At a quarter of what we were paying on GCP, we deployed our entire stack.”Tomasz, CTORead the story →
“On Vercel I was paying per project and it wasn't flexible enough to run a mixed stack. With Miget I'm paying less and I can easily spin up any combination of services I need.”Konrad, Software EngineerRead the story →
“Lately I barely open it, though, because the Miget skill for Claude handles most of the routine server work for me, and it's very quick.”Mateusz Kowalewski, AI Software Developer @ Locon
“Within minutes we were running a website we can be proud of, and know that it'll run very smoothly.”
“Having deployed services here for the past six months, the stability and error-free performance have far exceeded my expectations.”
“The Miget team was very responsive and worked with me to get things back on track. Since then, I haven't had any issues.”
Teams have moved to Miget from Google Cloud, AWS, Hetzner, OVH and Vercel, some for a fraction of the bill. We plan the move with you and migrate it for you, free.
30 days free on a $29 plan.