Features
Target audiences, change rollouts, run experiments, and evaluate flags in your app.
Choose who sees it
Target people, environments, and rollouts before you ship.
Match rules against user attributes, with an explicit fallback for every read.
Reuse audience definitions and make one flag depend on another.
Set time windows for flags, targeting rules, and segments.
Keep keys and configuration scoped to staging, production, or another environment.
Use public keys in clients and server keys on the backend.
Give each targeting key a stable place in the rollout.
Control a release
Review, apply, read back, and roll back when you need to.
Turn a flag off so evaluations serve its off variation.
Require a separate approver before a proposed change can apply.
List and inspect the configuration produced by each applied change.
Read the provider's live state after applying a change.
Explicitly restore a prior config version; rollback is not automatic.
Check code references, recent use, and rollback coverage before retiring a flag.
Understand each decision
Explain a decision and run experiments on a flag.
Assign users through flags and compare outcomes against defined metrics.
See why a flag returned a value and test changes before applying them.
See evaluation counts and fallback rates for each flag.
Export exposure events as resumable NDJSON for your own analysis.
Operate with code
CLI, HTTP API, and MCP.
Plan, apply, inspect, and roll back with script-friendly JSON output.
Types from your effective config for server or client. --check fails CI when keys or types drift.
Manage flags, keys, versions, and experiments through versioned endpoints.
Use scoped Superflag tools from an agent host through the local CLI.
Inspect recorded changes and compare configuration versions.
Build it into your app
SDKs decide from cached config in Node, React, and React Native.
Cache configuration and evaluate flags inside your server process.
Read client-visible flags with React hooks and a public key.
Evaluate client-visible flags in React Native and Expo apps.
Evaluate against cached configuration instead of calling an API for every check.
Use the OpenFeature provider or the OFREP HTTP protocol.