Environment variables

SDK configuration (backend)

Variable	Used by	Description
`AUTTER_RUNTIME_KEY`	`@autter/runtime-node`, any OTel SDK	Secret server key (`autter_rt_…`). Passed as `apiKey` to `initAutterServer` / `createBrowserRelayHandler`, or as the `Authorization: Bearer` header for a raw OTLP exporter.
`AUTTER_ENDPOINT`	`@autter/runtime-node`	Base URL of your ingester. Defaults to Autter’s hosted endpoint (`https://otlp.autter.dev`) if unset.
`GIT_SHA`	any	Convention used in the docs and examples for the `release` field — wire it to your build’s git commit so regressions can be traced to a release.
`NODE_ENV`	`@autter/runtime-node`	Commonly passed through as `environment`.

Frontend equivalents are passed directly to initAutterBrowser({...}) rather than read from process.env (bundlers don’t expose server env vars to the browser by default) — for example import.meta.env.VITE_GIT_SHA for release.

Endpoint regression SDK options

These are JavaScript options for initAutterServer and Next.js registerAutter, not additional environment variables.

release: set this to the deployed commit SHA. GIT_SHA is a convention that your build must supply.
environment: use the actual deployment environment and keep its name stable.
retainTracesAboveMs: a positive duration in milliseconds opts into bounded retention of slow successful server traces in the local process. Use 2000 as a starting example, not a universal threshold. Zero or omission leaves this retention off. Check export volume after enabling it.
instrumentations: add required database or dependency instrumentation to the existing initialization. Do not create a second SDK.

Use Node/Next.js SDK 1.3.0 or later; prefer patch 1.3.1. Self-hosted ingesters require 1.3.1 or later. There is no dedicated retainTracesAboveMs environment variable. See Endpoint regressions for limits and external OTel requirements.

Ingester configuration (self-hosted)

Variable	Required	Description
`CLICKHOUSE_URL`	yes	ClickHouse HTTP(S) endpoint the ingester writes to.
`CLICKHOUSE_PASSWORD`	yes	ClickHouse password for the ingester’s user.
`AUTTER_INGEST_KEYS`	yes, unless using `AUTTER_KEY_VALIDATOR_URL`	JSON array mapping ingest keys to `{orgId, repositoryId}`. See format below.
`AUTTER_KEY_VALIDATOR_URL`	alternative to `AUTTER_INGEST_KEYS`	URL of your own key-validation service, for dynamic key management instead of a static env-var list.
`AUTTER_SINK_URL`	no	Optional webhook. Each ingest batch is POSTed here as normalized occurrences — see Architecture. Delivery is best-effort, fire-and-forget.

`AUTTER_INGEST_KEYS`

A JSON array, one entry per key:

AUTTER_INGEST_KEYS='[
  {"key":"autter_rt_REPLACE_ME","orgId":"my-org","repositoryId":"my-app"},
  {"key":"autter_rtc_REPLACE_ME","orgId":"my-org","repositoryId":"my-app",
   "scope":"client","allowedOrigins":["https://app.example.com"]}
]'

Server keys (autter_rt_…) have no scope field and can send OTLP traces/metrics plus relay browser events.
Client keys (autter_rtc_…) need "scope": "client" and an allowedOrigins list — the ingester rejects /v1/browser requests whose Origin isn’t on the list.
One key pair per app or repository — repositoryId is how data is grouped for analysis.

Generate keys with anything sufficiently random, for example:

echo "autter_rt_$(openssl rand -hex 16)"

Full config reference: packages/otlp-ingester.

Self-hosting

Run the ingester locally, on a single server, or on AWS.

Get started

Guides

Reference

Environment variables

SDK configuration (backend)

Endpoint regression SDK options

Ingester configuration (self-hosted)

`AUTTER_INGEST_KEYS`

Self-hosting

​SDK configuration (backend)

​Endpoint regression SDK options

​Ingester configuration (self-hosted)

​AUTTER_INGEST_KEYS

Self-hosting

SDK configuration (backend)

Endpoint regression SDK options

Ingester configuration (self-hosted)

`AUTTER_INGEST_KEYS`