SpectraTyped observability for TypeScript apps

Quick example ​

import { defineCatalog, consolePublisher } from '@rachelallyson/spectra'
import { z } from 'zod'

const catalog = defineCatalog({
  'app.started': z.object({ env: z.string(), version: z.string() }),
  'guest.created': z.object({ tenantId: z.string(), guestId: z.string() }),
})

catalog.setPublishers([consolePublisher()])

catalog.emit('app.started', { env: 'production', version: '1.0.0' })
//                            ^ TypeScript and Zod both validate this

Why ​

Most observability is unstructured console.log plus whatever Sentry sees. That works until it doesn't — and when it doesn't, you spend an hour grepping logs for the one signal that would have closed the case in seconds.

The Stratum pattern fixes this by making every observable behavior a named, typed entry in a catalog. The catalog becomes the contract: every emit validates against it, tests assert sequences against it, and a coverage report flags which catalog entries no test ever exercises.

Spectra ships the patterns, not the framework. ~600 lines, no runtime dependencies, full type safety, designed to be read in one sitting.

What's next ​

• Getting Started — install + first events in 10 minutes.
• Concepts — the mental model in five short pieces.
• Recipes — tRPC, Inngest, Vitest setup, vendor adapters.
• API Reference — every export with signatures.

Inspiration ​

• Capital One's Stratum-Observability for the catalog + publisher patterns.
• OpenTelemetry for the request-context discipline.

Spectra deliberately doesn't try to replace OpenTelemetry — it's the typed-events-and-business-state layer that sits above OTel traces and metrics.