@quanticjs/feature-flags

Unleash-backed @FeatureFlag pipeline behavior.

Configuration

QuanticFeatureFlagsModule.forRoot(options) forwards UnleashModuleOptions to the Unleash client:

QuanticFeatureFlagsModule.forRoot({
  url: 'https://unleash.example.com/api',     // or UNLEASH_URL
  appName: 'my-app',                          // or UNLEASH_APP_NAME; default 'quanticjs-app'
  customHeaders: { Authorization: token },    // or UNLEASH_API_TOKEN
  refreshInterval: 10000,
  disableMetrics: false,
  environment: 'production',
  // Last-known-good resilience:
  bootstrap: { filePath: '/etc/flags/bootstrap.json' },
  backupPath: '/var/lib/unleash',
  // storageProvider: customStore,
});

Production fail-fast: with NODE_ENV=production and the default insecure token (*:*.unleash-insecure-api-token) or a http://localhost URL, forRoot() throws at boot. Outside production it logs a single warning. Note the check keys on NODE_ENV === 'production' — ops must set it.

The client has 'error' and 'warn' listeners attached before forRoot() returns, so an Unleash outage cannot crash the process via an unhandled EventEmitter 'error'.

Outage semantics — choose a fallback per flag

When Unleash is unreachable and no bootstrap/backup exists, every isEnabled() returns false, so @FeatureFlag('x') with the default fallback: 'throw' returns Result.failure(Forbidden) for ALL guarded commands — i.e. fail-closed.

• 'throw' (default) — kill-switch semantics: a provider outage disables the feature. Correct for risky features.
• 'skip' / 'default' — the command still executes: correct for launched features where a provider outage must not take the feature down.
• Additionally set bootstrap and/or backupPath so restarts during an outage evaluate from last-known-good values instead of all-false.