Skip to content
excalibase
docs

Excalibase

Excalibase is an open-source platform as a service: databases, APIs and app hosting, self-hostable on Kubernetes. You work in Studio, the web console, where each project gets its own database and the services around it.

Start here

Every project on Excalibase Cloud is reached at https://api.excalibase.io/{projectId}. Exchange a publishable key for a token, send it as Authorization: Bearer, and call GraphQL, REST, auth and functions on that one host. Connect to your project shows each call with a curl that runs as written, and Errors lists what a refused request answers. Every page keeps what only applies when you run Excalibase yourself under its own Self-hosting heading.

The platform

  • Postgres projects: Postgres 15, 16 or 17, with a table editor, SQL, roles, policies, triggers, indexes, extensions, advisors and CSV / TSV / Excel / Google Sheets import.
  • High availability: Free runs one instance; Standard runs three and Enterprise five, with synchronous replication.
  • MySQL projects: MySQL 8.4 with the same generated APIs, end-user login and row permissions, TLS verified against the project's own CA. See MySQL projects.
  • DocumentDB projects: a MongoDB-compatible document database on Postgres, with Mongo users managed in Studio.
  • APIs: GraphQL and REST generated from the schema. Each GraphQL request compiles to a single SQL query.
  • Permissions: per table, role and operation, with row filters and column lists. Nothing is reachable until permitted.
  • Realtime: table changes over WebSocket, filtered by the same permissions.
  • Storage and edge functions (TypeScript on Deno) per project.
  • Containers: deploy any image beside your database, sandboxed with gVisor.
  • Sites: upload a static frontend build and it is live at <name>.excalibase.app, with custom domains. See Sites.
  • Backups: base backups and WAL archiving, point-in-time restore, weekly disaster-recovery drill.

How a project fits together

Each project gets its own database and its own services. The control plane provisions them, backs them up and keeps them running; every API request runs as a role and passes through that project's permissions.

  • Clients: your apps (browser and server, through @excalibase/sdk) and Studio, the console your team signs in to.
  • Project services: the API engine (GraphQL, REST, realtime; permissions on every request), auth (end-user sign-up, sign-in, tokens and API keys), edge functions (TypeScript on Deno, CORS per project), storage (files behind signed upload URLs and rules) and containers (your images, sandboxed in gVisor).
  • Data: Postgres (15, 16 or 17; 1, 3 or 5 instances), DocumentDB (MongoDB-compatible, on Postgres), change capture (row changes stream to realtime) and backups (base backups and WAL, restore to a point in time).
  • Control plane: organizations, projects, provisioning, backups, DR drills, plans and limits.

The SDK (@excalibase/sdk) is a typed client for GraphQL, realtime, auth, storage uploads and functions.

Components documented here

GraphQL API — Stable

The engine behind every project's API. Also runs standalone in front of your own PostgreSQL (or MySQL) database: queries, mutations, subscriptions, filtering, pagination and relationships, with no schema files or resolvers.

→ Quick Start · Introduction

REST API — Stable

Every table gets list, insert, update, replace and delete routes with filter operators, ordering, pagination and an OpenAPI document. Aggregations and nested relationships are GraphQL features.

→ Quick Start · Introduction

Auth, Storage, Edge Functions — Beta

End-user auth for each project, per-project file storage, and TypeScript functions.

→ Auth · Storage · Edge Functions

MCP — Beta

Connect Cursor, Claude Code, Codex or Gemini CLI to a project with one URL and a token: schema, SQL, permissions, functions, deploys and logs.

→ MCP for AI coding tools

Provisioning — Beta

How projects are created on Kubernetes, backed up and restored.

→ Introduction · Backup & Restore

CDC Watcher — Beta

Streams INSERT / UPDATE / DELETE events from PostgreSQL (WAL) or MySQL (binlog) to NATS JetStream. It feeds realtime, and works on its own for data pipelines.

→ Introduction

Self-hosting

Everything above also runs on your own machines: on one Linux host with Docker or rootless Podman, or on Kubernetes with Helm charts (RKE2 with Cilium and gVisor is what production runs). A single host has no high availability, edge functions or realtime. See Self-hosting.

Source

All components are open source under the Apache License 2.0: github.com/excalibase.