Documentation menu

Start

Instrument

Vercel AI SDK Node & OpenTelemetry Browser & feedback Any language (OTLP)

Run the loop

Connect your coding agent Run your first loop Issue handoff

Data & trust

Data scrubbing Limits & ingest health

Reference

Documentation

Ship, watch, fix — documented.

Instrument your app, connect your coding agent, run the loop, and hold the trust bar. Agents read the same pages raw — append .md to any page URL.

Start

One pasted prompt, or a few curls by hand — from zero to a diagnosed telemetry window.

Instrument

One registered integration — every generateText, streamText, and tool call flows.

Node & OpenTelemetry

One-call OTLP wiring for any Node process, or point your existing OTel setup at LoopOps.

Browser & feedback

Product events and user feedback — typed or spoken — with a browser-safe write-only key.

Any language (OTLP)

Python, Go, Rust, anything — plain OTLP/HTTP JSON or protobuf, all three pillars.

Run the loop

Connect your coding agent

One URL — https://mcp.loopops.dev — for read-only production checks or a governed improvement cycle.

Run your first loop

Observe → diagnose → decide → improve → verify, end to end, with the approval gate in place.

One canonical packet hands a diagnosed issue to any coding agent — console button or MCP read.

Data & trust

Default-on PII/credential scrubbing at ingest — Sentry's model — with per-project rules.

Limits & ingest health

Caps, quotas, freshness — and the endpoint that tells you WHY nothing is showing up.

Reference

The full HTTP surface — machine-readable specs and the endpoints that matter.