@flagify/node
v1.6.0
Published
Official Flagify SDK for feature flag evaluation. TypeScript-first with local caching.
Downloads
315
Maintainers
Readme
Overview
@flagify/node is the official Node.js SDK for Flagify. TypeScript-first, with in-memory caching and sub-millisecond flag evaluation.
- TypeScript-first -- Full type safety with generics support
- In-memory cache -- Sub-millisecond evaluations after initial sync
- Stale-while-revalidate -- Serves cached values while refreshing in the background
- Lightweight -- Zero runtime dependencies (except
dotenv) - Isomorphic -- ESM and CommonJS output
- Runtime-friendly -- Runs on Node.js, modern browsers, edge runtimes, and React Native (Hermes engine)
Table of contents
- Installation
- Quick start
- Supported runtimes
- Configuration
- User context & targeting
- API reference
- How it works
- Environment variables
- Contributing
- License
Installation
# pnpm
pnpm add @flagify/node
# npm
npm install @flagify/node
# yarn
yarn add @flagify/nodeQuick start
import { Flagify } from '@flagify/node'
const flagify = new Flagify({
projectKey: 'proj_xxx',
publicKey: 'pk_xxx',
})
// Boolean flag
if (flagify.isEnabled('new-checkout')) {
showNewCheckout()
}
// Typed value
const limit = flagify.getValue<number>('rate-limit')Supported runtimes
@flagify/node works wherever the standard fetch API and AbortController are available:
- Node.js -- any version with
fetch(Node 18+ natively; Node 16 with a fetch polyfill). No minimumengines.nodeis enforced. - Modern browsers -- evergreen Chromium, Firefox, Safari.
- Edge runtimes -- Cloudflare Workers, Vercel Edge, Deno, Bun.
- React Native (Hermes engine) -- iOS and Android. The SDK polyfills
AbortSignal.timeoutinternally so runtimes that shipAbortControllerwithout the statictimeouthelper (Hermes, Node < 17.3) still get the 10s request deadline.
This is why @flagify/react works on React Native without any extra setup -- it delegates to @flagify/node, which handles the runtime gap transparently.
Configuration
import { Flagify } from '@flagify/node'
const flagify = new Flagify({
// Required
projectKey: 'proj_xxx',
publicKey: 'pk_xxx',
// Optional -- server-side only, never expose in client bundles
secretKey: 'sk_xxx',
options: {
// Custom API endpoint (defaults to https://api.flagify.dev)
apiUrl: 'https://api.flagify.dev',
// Cache TTL in ms (default: 5 minutes)
staleTimeMs: 300_000,
// Real-time updates via SSE — pushes flag changes to the client instantly
realtime: true,
// User context for targeting rules
user: {
id: 'user_123',
email: '[email protected]',
role: 'admin',
group: 'engineering',
geolocation: {
country: 'US',
region: 'CA',
city: 'San Francisco',
},
// Custom attributes
plan: 'enterprise',
companySize: 50,
},
},
})Configuration options
| Option | Type | Required | Default | Description |
|--------|------|----------|---------|-------------|
| projectKey | string | Yes | -- | Project identifier from your Flagify workspace |
| publicKey | string | Yes | -- | Client-safe publishable API key |
| secretKey | string | No | -- | Server-side secret key |
| options.apiUrl | string | No | https://api.flagify.dev | Custom API base URL |
| options.staleTimeMs | number | No | 300000 | Cache staleness threshold in ms |
| options.realtime | boolean | No | false | Enable real-time SSE updates |
| options.pollIntervalMs | number | No | -- | Polling interval in ms for periodic flag sync |
| options.sseIdleTimeoutMs | number | No | 45000 | Silence watchdog for the realtime stream. If no bytes (including heartbeat comments) arrive within this window, the client aborts and reconnects. Must exceed the server heartbeat interval. |
| options.sseReconnectBaseMs | number | No | 1000 | Base delay for the SSE reconnection backoff. Actual delay is jittered to 50–100% of the exponential value and honors the server's retry: field as a floor. |
| options.sseReconnectMaxMs | number | No | 30000 | Maximum delay for the SSE reconnection backoff. |
| options.user | FlagifyUser | No | -- | User context for targeting |
User context & targeting
Targeting rules let a flag return different values per user (admins, paid plans, beta cohorts, geographies, etc.). The targeting rules themselves are configured server-side in the Flagify dashboard or API — the SDK only forwards the user attributes.
Since v1.1.0, the SDK always runs the targeting engine on every sync — even when you don't pass
options.user. Catch-all rules (no segment, no conditions) and rollout rules that don't depend on user identity apply to anonymous callers. Segment/condition rules simply don't match when the context is empty, which is the correct behavior for anonymous. You only need to passoptions.userwhen you have user-specific targeting.
There are two valid patterns, depending on whether your process serves one user or many.
Pattern 1 — long-lived single-user client
Useful for CLIs, edge workers, single-tenant background jobs, or React Server Components that build a fresh client per request. Pass the user once via options.user; the client will fetch all flag values already evaluated for that user at startup. After await ready(), isEnabled(), getValue(), and getVariant() return the targeted values straight from the local cache.
const flagify = new Flagify({
projectKey: 'proj_xxx',
publicKey: 'pk_xxx',
options: {
user: { id: 'u_42', role: 'admin', plan: 'enterprise' },
// For long-lived processes, enable realtime (or set pollIntervalMs) so the
// cache doesn't go stale after the one-shot startup sync.
realtime: true,
},
})
await flagify.ready()
flagify.isEnabled('admin-tools') // true if the targeting rule matches role === 'admin'Pattern 2 — per-request evaluation (Express, Fastify, Next API routes)
In a typical multi-tenant web server, the client is created once at startup with no options.user, and you call await flagify.evaluate(key, user) per request with the request's user. evaluate() calls the API, the server applies the targeting rules, and you get the result back.
Note: even without options.user, isEnabled() / getValue() / getVariant() work against the local cache and reflect catch-all and rollout rules (see the note above). Use flagify.evaluate(key, user) per request only when you need user-specific targeting (segment membership, conditions, etc.).
import express from 'express'
import { Flagify } from '@flagify/node'
const flagify = new Flagify({
projectKey: 'proj_xxx',
publicKey: 'pk_xxx',
options: { realtime: true },
})
await flagify.ready()
const app = express()
app.get('/admin', async (req, res) => {
// evaluate() makes a network call per request — always wrap in try/catch and
// pick a safe fallback so an API blip doesn't 500 your handler.
let allowed = false
try {
const result = await flagify.evaluate('admin-tools', {
id: req.user.id,
role: req.user.role,
email: req.user.email,
})
allowed = result.value === true
} catch (err) {
console.warn('[flagify] evaluate failed, denying access by default', err)
}
if (!allowed) return res.status(403).end()
res.render('admin')
})The user object uses id (not userId) — the SDK serializes it to userId on the wire automatically.
{
id: string // required
email?: string
role?: string
group?: string
geolocation?: { country?: string; region?: string; city?: string }
[key: string]: unknown // any custom attribute
}evaluate() returns { key, value, reason } where reason is one of targeting_rule, rollout, default, or disabled. See the targeting docs for the full list of operators and segment options.
Don't mix the patterns in the React SDK. In
@flagify/react, the user goes in<FlagifyProvider options={{ user }}>once. Do not callclient.evaluate()from a React hook per flag — it bypasses the cache and produces a flash of the wrong value. See@flagify/reactREADME.
API reference
new Flagify(config: FlagifyOptions)
Creates a new Flagify client. Immediately fetches all flags and populates the local cache.
const flagify = new Flagify({
projectKey: 'proj_xxx',
publicKey: 'pk_xxx',
})flagify.isEnabled(flagKey: string): boolean
Evaluates a boolean feature flag.
Returns false when:
- The flag does not exist
- The flag is disabled
- The flag type is not
boolean
if (flagify.isEnabled('dark-mode')) {
applyDarkTheme()
}flagify.getValue<T>(flagKey: string, fallback: T): T
Returns the resolved value of a feature flag with a typed fallback.
// String variant
const variant = flagify.getValue<string>('checkout-flow', 'control')
// Number
const limit = flagify.getValue<number>('rate-limit', 100)
// JSON object
const config = flagify.getValue<{ maxRetries: number; timeout: number }>('api-config', {
maxRetries: 3,
timeout: 5000,
})flagify.getVariant(flagKey: string, fallback: string): string
Returns the string variant of a multivariate flag. Returns the variant with the highest weight, or the fallback if the flag has no variants or is disabled.
const variant = flagify.getVariant('checkout-flow', 'control')flagify.evaluate(flagKey: string, user: FlagifyUser): Promise<EvaluateResult>
Server-side evaluation with user targeting. Calls the Flagify API with user context for targeting rules.
const result = await flagify.evaluate('premium-feature', {
id: 'user_123',
email: '[email protected]',
role: 'admin',
})
// result: { key: 'premium-feature', value: true, reason: 'targeting_rule' }flagify.ready(): Promise<void>
Resolves when the initial flag sync is complete. Useful in server startup sequences.
const flagify = new Flagify({ projectKey: 'proj_xxx', publicKey: 'pk_xxx' })
await flagify.ready()flagify.destroy(): void
Disconnects the realtime listener, stops polling, and cleans up resources.
flagify.destroy()flagify.onFlagChange
Callback invoked when a flag changes via SSE or background refetch.
flagify.onFlagChange = (event) => {
console.log(`Flag ${event.flagKey} was ${event.action}`)
}How it works
Init Evaluate Stale?
---- -------- ------
GET /v1/flags --> Read from cache --> Background refetch
Cache all flags Sub-ms response GET /v1/flags/:key
Return stale value immediately- On initialization, the client syncs all flags from
GET /v1/flags - All evaluations read from the in-memory
Mapcache -- sub-millisecond - When a flag exceeds
staleTimeMs, the stale value is returned immediately while a backgroundGET /v1/flags/:keyrefreshes the cache - If the API is unreachable, the client falls back to cached defaults
Environment variables
| Variable | Description |
|----------|-------------|
| FLAGIFY_API_URL | Override the default API base URL |
| FLAGIFY_DEBUG | Set to 1 to enable verbose debug logs (SSE connect/reconnect/sync, flag-change notifications, idle-timeout warnings). Off by default — production consoles stay quiet. Errors that indicate a real problem (failed evaluation, duplicate connect, missing Provider) always log regardless. |
Debug logging
@flagify/node is silent in normal operation. To diagnose connection issues, opt in with the FLAGIFY_DEBUG env var:
FLAGIFY_DEBUG=1 node app.jsYou'll see entries like:
[Flagify] Realtime connected
[Flagify] Synced 12 flags via SSE
[Flagify] Flag changed: new-checkout (updated)
[Flagify] Realtime reconnectedReal errors (failed post-sync evaluation, duplicate connect() calls, missing configuration) always log with no opt-in needed.
Verifying webhook signatures
Every Flagify webhook delivery includes an X-Flagify-Signature: t=<unix>,v1=<hex> header. The signed string is <unix>.<rawBody> with HMAC-SHA256 keyed on the webhook secret you received when the subscription was created.
Two helpers ship in the server-only @flagify/node/webhooks subpath:
verifyWebhookSignature(rawBody, header, secret, opts?)— throws on failure, returnsvoidon success. Use when you parse the body yourself.constructWebhookEvent(rawBody, header, secret, opts?)— verifies AND parses the JSON, returns a typedWebhookEvent. The convenient default.
Why a subpath? These helpers use
node:crypto, which does not exist in React Native. Keeping them out of the main entry keeps@flagify/node(and therefore@flagify/react) bundleable by Metro/Expo. Since v1.6.0 they are not exported from the package root — import them from@flagify/node/webhooks.
import express from "express";
import {
constructWebhookEvent,
WebhookSignatureError,
} from "@flagify/node/webhooks";
const app = express();
// Webhook signatures sign the *raw bytes*, so disable JSON parsing for this route.
app.post(
"/webhooks/flagify",
express.raw({ type: "application/json" }),
(req, res) => {
try {
const event = constructWebhookEvent(
req.body, // Buffer — the raw bytes
req.header("x-flagify-signature") ?? "",
process.env.FLAGIFY_WEBHOOK_SECRET!,
);
// Idempotency: persist event.id and skip duplicates on retry.
console.log(event.event, event.data.environmentId);
res.sendStatus(200);
} catch (err) {
if (err instanceof WebhookSignatureError) {
console.error("rejected webhook:", err.code);
res.sendStatus(403);
return;
}
throw err;
}
},
);Both helpers accept an optional { tolerance, now }:
tolerance(seconds, default300) — replay-attack window. Set to0to disable timestamp checking.now— override the clock; useful in tests.
WebhookSignatureError.code enumerates the failure modes: MISSING_HEADER, MALFORMED_HEADER, TIMESTAMP_OUT_OF_TOLERANCE, SIGNATURE_MISMATCH, INVALID_PAYLOAD.
Types
All types are exported for convenience:
import type {
FlagifyOptions,
FlagifyUser,
FlagifyFlag,
IFlagifyClient,
EvaluateResult,
FlagChangeEvent,
RealtimeEvents,
RealtimeListener,
} from '@flagify/node'
// Webhook types live behind the server-only subpath:
import type {
WebhookEvent,
WebhookEventType,
WebhookEventData,
VerifyWebhookSignatureOptions,
WebhookSignatureErrorCode,
} from '@flagify/node/webhooks'Contributing
We welcome contributions. Please open an issue first to discuss what you'd like to change.
# Clone
git clone https://github.com/flagifyhq/javascript.git
cd javascript
# Install
pnpm install
# Development
pnpm run dev
# Build
pnpm run buildLicense
MIT -- see LICENSE for details.
