Skip to content

README.md

Latest commit

 

History

History
64 lines (44 loc) · 2.6 KB

README.md

File metadata and controls

64 lines (44 loc) · 2.6 KB
Raw

LaunchDarkly Server-Side AI SDK for Java

Circle CI

Status: In active development toward v1.0.0. APIs are not yet stable.

This library provides LaunchDarkly AI Config support for the LaunchDarkly Server-Side SDK for Java. It is built on top of launchdarkly-java-server-sdk and lets you retrieve, interpolate, and track AI Configs (models, providers, messages, agents, and judges) managed in the LaunchDarkly dashboard.

Supported Java versions

This library has a minimum Java version of 8.

Getting started

This module is part of the java-core monorepo and is published to Maven Central as com.launchdarkly:launchdarkly-java-server-sdk-ai.

Construct an LDAIClient from an initialized server-side LDClient, then retrieve a typed config:

LDClient ldClient = new LDClient(sdkKey);
LDAIClient aiClient = new LDAIClientImpl(ldClient);

Map<String, Object> variables = new HashMap<>();
variables.put("username", "Sandy");

AICompletionConfig config = aiClient.completionConfig(
    "my-ai-config-key",
    context,
    AICompletionConfigDefault.disabled(), // fallback when the flag is absent
    variables);

if (config.isEnabled()) {
    // config.getModel(), config.getProvider(), and config.getMessages() (already interpolated)
    // are ready to pass to your model provider.
}

The companion agentConfig/agentConfigs and judgeConfig methods retrieve agent and judge configs respectively. Within a prompt message or agent instruction, the evaluation context is available as {{ldctx}} (for example {{ldctx.key}}).

Metric tracking and manual judge evaluation will be added as the SDK is built out (see epic AIC-2629).

Internal API convention

Public, supported types live directly under com.launchdarkly.sdk.server.ai (and its documented subpackages). Anything under com.launchdarkly.sdk.server.ai.internal is implementation detail: it is not part of the supported API, is excluded from the published Javadoc and sources jars, and may change without notice.

Contributing

We encourage pull requests and other contributions from the community. Check out our contributing guidelines for instructions on how to contribute.

About LaunchDarkly

LaunchDarkly is a feature management platform that serves trillions of feature flags daily to help teams build better software, faster. Get started using LaunchDarkly today!