Stackship platform
What Stackship is, the concepts everything else builds on — boundaries, resource groups, resources and roles — and where to start.
- Connect an AI client — Connect Claude Code, Claude Desktop, Codex, opencode or another MCP client to the platform, so that an AI assistant can work with your resources as you.
- API reference — Every HTTP endpoint of the kernel module: authentication, IAM action, parameters, bodies and status codes.
- Authenticate — Get an access token as a person or as a service account, and send it with your API calls.
- Error codes — The error codes every platform module can answer with, and those of the platform's core — the status, the cause and what to do.
- Errors — The body a failed API call returns, what each HTTP status code means on the platform, and where to look up an error's code.
- File errors — The error codes the file routes of apps and container instances answer with — the status, the cause and what to do.
- Platform API — Where the platform's HTTP API lives, how its routes are organised, and how to list the routes an installation serves.
- Snapshot errors — Why the platform refuses to take or restore a snapshot of a PostgreSQL cluster, SQL Server instance or container instance, and what to do about each refusal.
- User settings — The per-user settings the platform stores, their defaults, and the routes that read and change them.
- Connect a cluster — Register a Kubernetes cluster so that boundaries can be projected onto it and resources placed there.
- Clusters — How the platform reaches the Kubernetes clusters it deploys onto, what it stores about them, and what the Clusters list shows.
- Manage a cluster — Change a registered cluster, issue a join token, revoke an agent, or remove a cluster from the platform.
- Configuration reference — Configuration keys of the kernel module: sections, environment variables, types and defaults.
- Deployment sources — Connect a boundary to GitHub, GitLab, Azure DevOps or a private container registry, so that its resources can build from your repositories or pull your images.
- Deployment source settings — The installation settings deployment sources depend on — callback addresses, webhook delivery, and provider defaults — and what to check when pushes do not deploy.
- Set up e-mail delivery — Configure the SMTP server the platform sends invitations and notifications through, and send a test message.
- Edit e-mail templates — Change the subject and text of the e-mails the platform sends, preview them, and restore earlier versions.
- Getting started — Sign in for the first time and set the platform up with the first-run wizard — a boundary, a cluster and the first invitations.
- Stackship platform — What Stackship is, the concepts everything else builds on — boundaries, resource groups, resources and roles — and where to start.
- Manage LLM gateways — Register the language-model endpoints that platform services such as Sentinel use, choose which models they may use, and decide which gateway and model each Sentinel stage gets.
- Turn on AI clients — Switch the platform's MCP interface for AI clients on or off, choose which clients may register themselves, and the installation settings behind it.
- Operations — What a platform operation is, the states it passes through, and how to follow one in the portal, with the CLI and through the API.
- Permissions — The platform-core actions — clusters, deployment sources, operations, activity, platform settings — and the built-in roles that hold them.
- Platform Settings — The Platform Settings page, its tabs, and the permission each tab needs.
- Service accounts — Create a service account for CI/CD or another integration, give it roles, get a token with its credentials, and rotate or delete it.
- Single sign-on — Let a tenant's users sign in through their own identity provider — add e-mail domains and OIDC or SAML identity providers, and send users straight to their provider.
- Export platform telemetry — Decide whether the platform sends its own logs, metrics and traces to an external OpenTelemetry backend, and to which one.
- Troubleshooting — A checklist for when something on the platform does not behave as expected — wrong scope, missing permission, failed deployments and data problems.
- Invite and manage users — Invite people to a boundary by e-mail, follow and resend invitations, and find the users who already have access.