@gitbook/adaptive
GitBook's adaptive content feature allows you to personalize your documentation based on visitor data. The @gitbook/adaptive SDK provides utilities to help facilitate passing data to GitBook for adapting your content.
This SDK includes both feature flag helpers for popular providers like LaunchDarkly and Reflag, as well as generic utilities for writing custom visitor data.
Prerequisites
Make sure you have:
- A GitBook site with adaptive content enabled
- An active project on LaunchDarkly or Reflag
- A frontend project (React-based) where feature flags are available client-side
Feature Flag Helpers
The SDK provides convenient helpers for popular feature flag providers to automatically sync flag values with GitBook's adaptive content system.
LaunchDarkly Integration
1. Install the GitBook + LaunchDarkly Integration
In GitBook, install the LaunchDarkly integration for your site.
2. Add Your Access Credentials
In the GitBook integration settings, provide:
- Your project key
- A service access token from your LaunchDarkly account
3. Install the GitBook Helper
npm install @gitbook/adaptive
4. Configure the Client
import { render } from 'react-dom';
import { withLaunchDarkly } from '@gitbook/adaptive';
import { asyncWithLDProvider, useLDClient } from 'launchdarkly-react-client-sdk';
import MyApplication from './MyApplication';
function PassFeatureFlagsToGitBookSite() {
const ldClient = useLDClient();
React.useEffect(() => {
if (!ldClient) return;
return withLaunchDarkly(ldClient);
}, [ldClient]);
return null;
}
(async () => {
const LDProvider = await asyncWithLDProvider({
clientSideID: 'client-side-id-123abc',
context: {
kind: 'user',
key: 'user-key-123abc',
name: 'Sandy Smith',
email: 'sandy@example.com',
},
options: {}
});
render(
<LDProvider>
<PassFeatureFlagsToGitBookSite />
<MyApplication />
</LDProvider>,
document.getElementById('reactDiv')
);
})();
Visitor Schema Output
Once connected, feature flag values will be available in GitBook under:
unsigned.launchdarkly.flags
Reflag Integration
1. Install the GitBook + Reflag Integration
In GitBook, enable the Reflag integration for your site.
2. Add Your Secret Key
In the GitBook integration settings, provide your Reflag secret key.
3. Install the GitBook Helper
npm install @gitbook/adaptive
4. Configure the Client
import { withReflag } from '@gitbook/adaptive';
import { ReflagProvider, useClient } from '@reflag/react-sdk';
import MyApplication from './MyApplication';
function PassFeatureFlagsToGitBookSite() {
const client = useClient();
React.useEffect(() => {
if (!client) return;
return withReflag(client);
}, [client]);
return null;
}
export function Application() {
const currentUser = useLoggedInUser();
const appConfig = useAppConfig();
return (
<ReflagProvider
publishableKey={appConfig.reflag.publishableKey}
user={{
id: currentUser.uid,
email: currentUser.email ?? undefined,
name: currentUser.displayName ?? '',
}}
company={{
id: currentUser.company.id,
}}
>
<PassFeatureFlagsToGitBookSite />
<MyApplication />
</ReflagProvider>
);
}
Visitor Schema Output
Once connected, feature flag values will be available in GitBook under:
unsigned.reflag.flags
Generic Utilities
If you don't use LaunchDarkly or Reflag, or need to pass custom data beyond feature flags, the SDK provides generic utilities for writing visitor data.
writeGitBookVisitorCookie
The writeGitBookVisitorCookie function allows you to write custom visitor data that will be available in GitBook's adaptive content system.
import { writeGitBookVisitorCookie } from '@gitbook/adaptive';
writeGitBookVisitorCookie('userPlan', 'premium');
writeGitBookVisitorCookie('userRole', 'admin');
writeGitBookVisitorCookie('preferences', {
theme: 'dark',
language: 'en'
});
The data will be available in GitBook under:
unsigned.{cookieName}
For example, the above code would create:
unsigned.userPlan with value "premium"
unsigned.userRole with value "admin"
unsigned.preferences with value { theme: "dark", language: "en" }
Important Note
All visitor data is evaluated on the client side and stored in cookies. Do not pass sensitive or security-critical data through this method.