Skip to app details
Back to app library
APPNEURAL APP FAMILYAutomation & Integrations

Feature Flags

Ship to a few, then to everyone

Turn features on for some people first, ramp them up, and switch them off at once if something breaks.

01 / OVERVIEW

See where it fits in your day.

Each domain is one project with one live configuration: there are no environments, and segments and rollouts take the place of Dev and Staging. Flags have a key, a type (boolean, string, number or JSON) with 2 to 20 variations, a kill switch (new flags start off), an off variation, a default rule, tags, an owner and a temporary or permanent mark. Targeting has individual targets and ordered rules on key, email, email domain, country, plan, name or custom attributes, with list, text, number, date and semver operators (each also negated) and reusable segments. Percentage rollouts place people by a stable hash of the flag key and context key, so a ramp from 10% to 50% keeps the first 10% in. Scheduled changes turn a flag on or off at a time or ramp it in steps, applied by a cron every 5 minutes. POST /api/v1/evaluate (secret key) answers servers with the value, variation and reason; GET /api/v1/flags and the /sdk.js browser SDK serve client-side flags only, from a cache that every change refreshes. Daily evaluation counts per variation, a history with field diffs, stale-flag digests and webhooks.

A LOOK INSIDEExample layout · sample content
Flags and what they serve
FlagServingStatus
new-checkoutfeature flagsEXAMPLE
dark-modefeature togglesEXAMPLE
hindi-interfacerolloutEXAMPLE
beta-dashboardkill switchEXAMPLE
02 / CAPABILITIES

Built around the work you do.

What Feature Flags does for you.

01

Targeting

Rules on email, country, plan or any attribute, reusable segments and individual targets.

02

Rollouts

Stable percentage rollouts and scheduled ramps, with no environments to keep in sync.

03

Evaluation

A server API that gives its reasons, and a small browser SDK that serves client-side flags only.

03 / GETTING STARTED

Start small. Build from there.

A suggested path for Feature Flags. What you can do depends on your access in the app.

01

Create a flag

Choose its type and variations; new flags start off.

02

Target and roll out

Add rules, segments and a percentage.

03

Watch and clean up

See evaluations per variation and remove stale flags.

YOUR NEXT MOVE

Make Feature Flags part of your workflow.

The app opens in a new tab, so your ecosystem stays close.

Open app
05 / CATALOG DETAILS

The registry record.

Live host, inventory, governance, health checks, verification, dependencies and history.

Catalog surfaceDEPLOYEDHEALTHY
InventoryACTIVE1 items
P2VERIFIEDPUBLISHED
Overview

Purpose and boundaries

Each domain is one project with one live configuration: there are no environments, and segments and rollouts take the place of Dev and Staging. Flags have a key, a type (boolean, string, number or JSON) with 2 to 20 variations, a kill switch (new flags start off), an off variation, a default rule, tags, an owner and a temporary or permanent mark. Targeting has individual targets and ordered rules on key, email, email domain, country, plan, name or custom attributes, with list, text, number, date and semver operators (each also negated) and reusable segments. Percentage rollouts place people by a stable hash of the flag key and context key, so a ramp from 10% to 50% keeps the first 10% in. Scheduled changes turn a flag on or off at a time or ramp it in steps, applied by a cron every 5 minutes. POST /api/v1/evaluate (secret key) answers servers with the value, variation and reason; GET /api/v1/flags and the /sdk.js browser SDK serve client-side flags only, from a cache that every change refreshes. Daily evaluation counts per variation, a history with field diffs, stale-flag digests and webhooks.

Gradual rollout for a connected domain's apps: flags with variations, targeting rules and segments, percentage rollouts, a kill switch, scheduled changes and a browser SDK.

A/B tests of pages belong to Landing Pages and Campaigns; there are no environments (one live host).

Priority
P2
Portfolio
Automation & Integrations
Capability
Automation & Integrations
Technical owner
Not configured
Runtime
Cloudflare Workers / OpenNext
Last verified
2026-11-18 14:03:13
Semantic inventory

Every number says what it counts

1Catalog itemstrackedEntityCount · catalog-api:live
Production-ready gate

Governance evidence required

5 of 18 required controls are configured. Catalog deployment and inventory maturity remain independent until every required control is present.

MissingBusiness owner
MissingProduct owner
MissingTechnical owner
MissingMaintaining team
MissingGitHub repository
ConfiguredProduction URL
ConfiguredHealth endpoint
MissingDocumentation
MissingArchitecture document
MissingCurrent security review
ConfiguredData classification
MissingDependencies
MissingConsumers
MissingSLA / SLO
MissingOn-call
MissingLast release
ConfiguredLast verification
ConfiguredDeprecation status
Catalog details

flag schema v1

Catalog-specific metadata is not configured.
Live host

Where it runs

DNS & routing

Confirmed infrastructure only

Provider inspection appears after Cloudflare credentials are configured. Mutations require dry run, idempotency and explicit confirmation.
Repository & architecture

Technical context

APPNEURAL-Catalogs/flag-catalogLOCAL · NOT CONFIGURED

Documentation

Documentation and architecture evidence are not configured.
Dependencies

Registered relationships

No dependency edges are registered. Nothing has been inferred.
Operations

Checks and verification

No health checks have been recorded.
Security

Review posture

Security review not configured. This does not mean the catalog is secure or insecure.
Automatic reconciliation

GitHub · Cloudflare · DNS · catalog APIs

No synchronization run has completed yet. The scheduled Worker records missing credentials as Not configured, never as success.

Drift findings

No drift has been recorded.
Evidence & history

Trust record

Verification is verified. Run verification to store check-level evidence, confidence, expiry and remediation.

AI workspace

Human-reviewed drafts

AI outputs are saved as drafts with evidence references. They never replace verified catalog fields automatically.

App libraryOpen app