Skip to content
FlagPilot
Menu

Product

The platform

Four parts, one evaluation engine. The dashboard and the REST API write flag configuration; the delivery service hands immutable snapshots to SDKs; the SDK evaluates locally.

The dashboard

Sign in and pick an organization, a project and an environment. From there you create and edit flags, build segments, manage environments and SDK keys, and run the evaluation debugger against sample users.

Roles are Owner, Admin, Developer and Viewer, per organization. The dashboard hides what your role cannot do; the API enforces every rule on its own.

The management API

Everything the dashboard does goes through a REST API with a committed OpenAPI document, so scripts and CI can do the same. Configuration writes carry the version they expect, so a concurrent edit is refused instead of silently overwritten.

The delivery service

SDKs fetch the current snapshot of one environment with a server key. Polling uses ETags, so an unchanged environment costs a 304. Browsers use a separate client key that only returns evaluated values for flags marked as client-enabled, never rules or segments.

The Node SDK

One client per process downloads the snapshot, validates it and evaluates flags in memory. It never throws into your code: every failure returns your default value with a reason.

Defaults at a glance

Default environments
development, staging, production
SDK poll interval
30 s (configurable)
Rollout precision
Basis points (0.01%)
Key revocation
Effective within 30 s
Delivery rate limit
60 requests/s per key, burst 120
Invitation lifetime
7 days, single use

Run it next to your app

Sign in to the dashboard, or clone the repository and start the whole stack locally with Docker and pnpm.