Installation

dependencies {
    // The graph builder and the engine. This is all the quick start needs.
    implementation("dev.deeptelar:telar-core:0.1.0-alpha09")

    // Optional modules. Add only the ones you use.
    implementation("dev.deeptelar:telar-serialization:0.1.0-alpha09")      // save @Serializable states
    implementation("dev.deeptelar:telar-checkpoint-file:0.1.0-alpha09")    // save runs as JSON files
    implementation("dev.deeptelar:telar-checkpoint-browser:0.1.0-alpha09") // save runs in a browser's localStorage (JS and Wasm)
    implementation("dev.deeptelar:telar-agent:0.1.0-alpha09")              // chat models, tools and the tool-calling agent
    implementation("dev.deeptelar:telar-anthropic:0.1.0-alpha09")          // call Claude, on every platform
    implementation("dev.deeptelar:telar-openai:0.1.0-alpha09")             // call OpenAI, Ollama and compatible servers, on every platform
    implementation("dev.deeptelar:telar-langchain4j:0.1.0-alpha09")        // call AI models through LangChain4j (JVM)
    implementation("dev.deeptelar:telar-typesafe:0.1.0-alpha09")           // ask Jev, a decision model, on every platform
}

Until 0.1.0-alpha05 the project was called langgraph-kt, and its artifacts were io.github.cuento3yllevo2:langgraph-kt-*. The changelog says how to update.

Module	Targets	Purpose
telar-core	JVM/Android, iOS, macOS, Linux, Windows, JS, Wasm	Graph builder, execution engine, checkpointing interfaces
telar-serialization	same as core	KotlinxStateSerializer for @Serializable states, CheckpointCodec for custom checkpointers
telar-checkpoint-file	same as core (Node.js only for JS/Wasm)	FileCheckpointer, one JSON file per thread
telar-checkpoint-browser	JS and Wasm in a browser	LocalStorageCheckpointer, runs that survive a page reload
telar-agent	same as core	ChatModel, Tool, and the tool-calling agent: toolAgent / toolLoop
telar-anthropic	same as core	AnthropicChatModel, Claude through Ktor
telar-openai	same as core	OpenAiChatModel: OpenAI, Ollama and other servers with the Chat Completions API, through Ktor
telar-langchain4j	JVM (Java 17+)	LangChain4jChatModel and chatNode / chatMessagesNode for LangChain4j 1.x models
telar-typesafe	same as core	TypeSafeDecisionModel: Jev, the decision model of TypeSafe AI, through Ktor
telar-test	same as core	ScriptedChatModel, ScriptedDecisionModel and record(), for tests of a graph with an agent. Not released yet: it comes with the version after 0.1.0-alpha09.

Requires Kotlin 2.x. JVM artifacts target Java 11, except telar-langchain4j, which needs Java 17 because LangChain4j does.

Status

Alpha. 0.1.0-alpha09 is the latest release, and it is on Maven Central. The API may still change before 1.0; the changelog lists what changes in each version, and the roadmap lists what is planned.