The platform, in depth

What each part of the platform does, and how it works.

What you need to run in production

The parts a production app depends on, without hiring a platform team.

Git-based deploys

Every push deploys, on paid plans every pull request gets a preview, and you can roll back to any earlier deployment in one click.

Sign in with GitHub or with a magic link sent to your email. GitHub repositories connect through the GitHub App; GitLab, gitlab.com or self-hosted, connects with an access token you create in GitLab and paste in Settings, plus a webhook for push deploys.

Managed databases

Postgres, Redis, MySQL, and MongoDB, each with its own credentials. Postgres databases get snapshots you can restore.

S3-compatible storage

Buckets with their own scoped access keys, quotas, and signed URLs for temporary access.

Cron jobs

Run a command on a cron schedule. Each run gets its own isolated container, with logs and an exit code.

Preview environments

On Dev and up, pull requests get their own live URL, up to your plan’s preview limit, removed when the pull request closes.

Static-site hosting

Serve SPAs and static builds from nginx, with long-lived caching for hashed assets and HTTPS from the first deploy.

Autoscaling and scale-to-zero

Add replicas as CPU load rises. Idle services can sleep at zero and wake on the next request.

Custom domains, free HTTPS

Point your domain at us. We issue the certificate and renew it, with nothing for you to configure.

Logs and monitoring

Live build and runtime logs, resource usage, and deploy history, on one page per service.

Gheima School

An account type for bootcamps and universities: cohorts, students, resource limits, and graduation.

Everything else the docs cover

Managed services, configuration, and the ways to drive Gheima from code. Each one links to its full guide.

Managed search

Full-text search on Meilisearch, with typo tolerance and faceting, without running a server. Each instance gets an API key that only reaches indexes under your own prefix.

From the Dev plan

Read the guide: Managed search

Job queues

A Redis user set up for BullMQ, with ACLs that keep your keys on a private prefix and deny destructive commands. Link it to a service and every deploy injects REDIS_URL and QUEUE_PREFIX.

From the Dev plan

Read the guide: Job queues

Transactional email

Send receipts, verification and password-reset mail from your own domain. Gheima generates a DKIM key per domain and signs outgoing mail with it, and keeps a delivery log.

From the Dev plan

Read the guide: Transactional email

Persistent volumes

A durable disk attached to a service at a mount path. Files written there survive restarts, redeploys and rescheduling.

From the Dev plan

Read the guide: Persistent volumes

Env groups

A reusable set of environment variables attached to any number of services. A service’s own variables always take precedence over the group’s.

From the Dev plan

Read the guide: Env groups

Release command

Run migrations or any pre-release step once per deploy, in a fresh container from the new image, before traffic switches over. If it fails, the deploy stops and the previous version keeps serving.

On every plan

Webhooks

Send platform events to an endpoint you control, with signed deliveries and a delivery history, from the dashboard or the API.

On every plan

Read the guide: Webhooks

Single sign-on

Your team signs in through your own OIDC identity provider, such as Okta, Microsoft Entra, Google Workspace or Keycloak. People on your claimed domain become members on their first sign-in.

From the Team plan

Read the guide: Single sign-on

REST API

Everything you do in the dashboard, over HTTP with a bearer API key, with a per-key rate limit set by your plan and an OpenAPI spec.

On every plan

Read the guide: REST API

TypeScript SDK

A typed, fetch-based client with retries and pagination, for Node 18+, the browser and edge runtimes.

On every plan

Read the guide: TypeScript SDK

CLI

Deploy, tail logs and manage services, databases and storage from your terminal. Output in --json and stable exit codes make it fit for scripts.

On every plan

Read the guide: CLI

MCP server and AI-assistant docs

Connect Claude, Cursor, VS Code or any MCP client to your account, read-only or read and write. Every docs page is also available as Markdown for coding agents.

On every plan

Read the guide: MCP server and AI-assistant docs

From commit to live in three steps

You don't need a Dockerfile, and there's no server to rent. Push, and Gheima does the rest.

01

Connect your repository

Connect GitHub or GitLab and pick a repository. Gheima detects your stack.

02

We build the image

Every push starts a clean build, and you can watch the logs line by line.

03

Live over HTTPS

Your service gets a gheima.ma subdomain and a free HTTPS certificate. On the Dev, Pro, Team, and Agency plans, you can point your own domain at it.

Questions about the platform

Which languages and frameworks can I deploy?

Anything Nixpacks can build, such as Node.js, Next.js, Python, Django, Go, Ruby and Rust, or any app with a Dockerfile. Automatic PHP builds are paused while the PHP the builder installs is affected by a security flaw, so PHP and Laravel apps deploy from a Dockerfile for now.

Which managed services are there?

PostgreSQL, MySQL, MongoDB and Redis databases, S3-compatible object storage, Meilisearch search, job queues and transactional email. How many of each you get depends on the plan.

Do I get HTTPS and my own domain?

Every service gets HTTPS on a gheima.ma address. On Dev and above you can add your own domain, and Gheima issues and renews its certificate.

Can an AI assistant deploy for me?

Yes. Gheima runs a hosted MCP server at gheima.ma/mcp, so Claude Code, Cursor and other MCP clients can list your services, deploy them and read their logs once you sign in.

Request an invite