Spala Catalogue

*.example.spala.ai

Fifty backends.
All of them running.

Forty-eight complete products you can sign into. One API surfaces already serving public endpoints you can call from your terminal — the rest still provisioning, and labeled that way. Not a gallery of mockups: a provisioning report.

Provisioning state
Product, liveDegradedAPI surface
  Hover a tile — the feed follows

Forty-eight products, turned inside out

Switch any row between the shipped interface, the schema underneath it, and the routes that enforce its rules. That triple is the whole point of Spala.

Two API surfaces, one serving endpoints

Pick one. The request and the real response appear on the right, fetched live from that backend — nothing cached, nothing staged. Surfaces still provisioning are marked “endpoints pending”.

What every published surface quietly proves

01

Auth is never an afterthought

Every published backend ships signup, sessions, and owner-scoped reads and writes — verified by behavioral probes before it may publish.

02

Isolation is tested, not assumed

Two-account probes check that one user can never read another's rows, on every owned model of every published project.

03

Contracts are generated, not drawn

Each surface publishes OpenAPI straight from its live definition — the same source the runtime executes.

04

Agents built all of it

Models, endpoints, and logic were drafted by coding agents over Spala MCP, validated deterministically, then published.

Entry fifty-one is yours.

Point your coding agent at Spala's MCP. It builds the database, auth, APIs and logic in a backend you can read, verify and export.

$npx @spala-ai/mcp-install init --client claude-code