TokenForensics.ai

Observability & ops plumbing

Internal production-hygiene tooling: error tracking, product analytics, performance metrics, and rate limiting.

Walkthrough — Observability & ops plumbing

What it is

  • Sentry — instrumentation.ts + sentry.{client,server,edge}.config.ts, across all three runtimes.
  • PostHog — src/app/providers.tsx (client, pageviews via the /ingest proxy) + src/lib/posthog-server.ts (lazy server client).
  • Vercel Analytics + Speed Insights — wired in the root layout.
  • Rate limiting — src/lib/ratelimit.ts, an Upstash sliding-window limiter (ingestLimiter), ready to attach to the capture-ingest endpoint when Pillar 1 lands.

The boundary that matters

This is internal ops only

TokenForensics is not an observability product. Observability watches, gateways control — forensics investigates. This plumbing watches our app; it points inward only. Telemetry must never capture customer token content or evidence payloads. Customer-facing forensic capability is a pillar, never built on this internal plumbing.

Why it matters

Indirectly: it keeps the product reliable and fast and surfaces errors to us before users hit them. The user-facing benefit is a stable workbench — within that strict boundary.

Last updated