Getting started

Five minutes from an invite to a flag your code reads.

1. Sign in

Flag is in an invite-only beta, so sign-ups are closed. Open the invite link from your email, then sign in with a passkey, a password or (where set up) Google, Microsoft, GitHub, Apple or your company’s SSO. Open the app at /app/. Org admins invite more people under Settings → Members.

2. Create a flag

Each org starts with a default project and two environments: production, marked critical, and development. Under Flags, choose New flag and start from a template:

Or choose the kind and variations yourself. Give it a key: that’s what your code asks for, for example new-checkout. New flags are off in every environment.

The same over the REST API, with an API key from Settings → API keys ($ORG is your org’s slug):

curl -X POST https://flag.nightroll.app/api/v1/orgs/$ORG/projects/default/flags \
  -H "Authorization: Bearer $KEY" -H 'content-type: application/json' \
  -d '{"key": "new-checkout", "template": "release"}'

3. Get an SDK key

Under Environments, pick the environment and choose:

• New server SDK key (srv_...) for back ends. It is secret and shown only once: put it in your secret store. Server SDKs download the environment’s whole configuration, encrypted with this key.
• or copy the client-side ID (cli_...) for browsers and mobile apps. It is public and only sees flags marked available to client-side SDKs, with targeted keys hashed.

Each key belongs to one environment. Rotate keys from the same page: the old ones keep working for as long as you choose.

4. Read the flag from your code

TypeScript (Node, Bun, Deno)

npm install @nightroll/flag

import { FlagClient } from '@nightroll/flag';

const flags = new FlagClient({ sdkKey: process.env.FLAG_SDK_KEY });
await flags.ready();

const user = { kind: 'user', key: 'user-123', email: 'ada@example.com', plan: 'pro' };
if (flags.boolVariation('new-checkout', false, user)) {
  // the new checkout
}

Browsers and React

import { FlagClient } from '@nightroll/flag/web';
import { FlagProvider, useFlag } from '@nightroll/flag/react';

const flags = new FlagClient({ clientId: 'cli_...' });

function App({ user }) {
  return <FlagProvider client={flags} context={{ key: user.id, plan: user.plan }}><Checkout /></FlagProvider>;
}
function Checkout() {
  return useFlag('new-checkout', false) ? <NewCheckout /> : <OldCheckout />;
}

Mark the flag available to client-side SDKs (under Variations & settings) for browsers to see it.

Any other language: OFREP

Any OpenFeature SDK with an OFREP provider works: point it at https://flag.nightroll.app with the header Authorization: Bearer <key>. Or call it directly:

curl https://flag.nightroll.app/ofrep/v1/evaluate/flags/new-checkout \
  -H "Authorization: Bearer $FLAG_SDK_KEY" -H 'content-type: application/json' \
  -d '{"context": {"targetingKey": "user-123", "plan": "pro"}}'

More in SDKs.

5. Turn it on

Open the flag. In Targeting, pick development, tick Targeting is on, and add what you need:

• Individual targets: keys that always get a variation (your own user, say).
• Rules: for example email ends with @acme.com serves On; is in segment beta-testers serves On.
• Default rule: what everyone else gets, one variation or a percentage rollout (10% On, 90% Off).

Choose Review & save: Flag shows a dry run of exactly what changes before anything does. In production you also see the blast radius (recent traffic, dependent flags, connected SDKs) and type production to confirm. Streaming SDKs get the change within about a second.

Check it worked under Evaluate (paste a context, see the variation and every step that led to it) and Insights (evaluations per hour by variation, from your SDKs).

6. Let your agent help

Create an API key, then:

claude mcp add --transport http flag https://flag.nightroll.app/mcp --header "Authorization: Bearer $KEY"

Or add https://flag.nightroll.app/mcp as a custom connector in claude.ai or Claude Desktop and approve it in the browser. The QuickStart walks through it, with prompts to try.

Next

• Protect production: Environments → Settings (critical, require approval).
• Release on a schedule: a flag’s Schedules.
• Clean up: Stale flags and Kill switches.
• Automate: REST API.

This page as Markdown: /docs/getting-started.md · All docs for agents: /llms.txt