Prompt chaining¶

Original tutorial: Prompt chaining.

Prompt chaining runs a sequence of LLM calls where each step uses the result of the previous step. This workflow generates a joke, checks whether it has a punchline, and only improves it when needed.

Setup¶

Set OPENAI_API_KEY before running the model calls.

In [1]:

var userHomeDir = System.getProperty("user.home");
var localRepoUrl = "file://" + userHomeDir + "/.m2/repository/";
var langchain4jVersion = "1.19.0";
var langgraph4jVersion = "1.9.0-beta6";

var userHomeDir = System.getProperty("user.home"); var localRepoUrl = "file://" + userHomeDir + "/.m2/repository/"; var langchain4jVersion = "1.19.0"; var langgraph4jVersion = "1.9.0-beta6";

In [2]:

%dependency /add-repo local \{localRepoUrl} release|never snapshot|always
%dependency /add org.bsc.langgraph4j:langgraph4j-core:\{langgraph4jVersion}
%dependency /add dev.langchain4j:langchain4j-open-ai:\{langchain4jVersion}
%dependency /resolve

%dependency /add-repo local \{localRepoUrl} release|never snapshot|always %dependency /add org.bsc.langgraph4j:langgraph4j-core:\{langgraph4jVersion} %dependency /add dev.langchain4j:langchain4j-open-ai:\{langchain4jVersion} %dependency /resolve

Repository local url: file:///Users/bsorrentino/.m2/repository/ added.
Adding dependency org.bsc.langgraph4j:langgraph4j-core:1.9.0-beta6
Adding dependency dev.langchain4j:langchain4j-open-ai:1.19.0
Solving dependencies
Resolved artifacts count: 12
Add to classpath: /Users/bsorrentino/Library/Jupyter/kernels/rapaio-jupyter-kernel/mima_cache/org/bsc/langgraph4j/langgraph4j-core/1.9.0-beta6/langgraph4j-core-1.9.0-beta6.jar
Add to classpath: /Users/bsorrentino/Library/Jupyter/kernels/rapaio-jupyter-kernel/mima_cache/org/bsc/async/async-generator/5.0.0/async-generator-5.0.0.jar
Add to classpath: /Users/bsorrentino/Library/Jupyter/kernels/rapaio-jupyter-kernel/mima_cache/org/slf4j/slf4j-api/2.0.9/slf4j-api-2.0.9.jar
Add to classpath: /Users/bsorrentino/Library/Jupyter/kernels/rapaio-jupyter-kernel/mima_cache/dev/langchain4j/langchain4j-open-ai/1.19.0/langchain4j-open-ai-1.19.0.jar
Add to classpath: /Users/bsorrentino/Library/Jupyter/kernels/rapaio-jupyter-kernel/mima_cache/dev/langchain4j/langchain4j-core/1.19.0/langchain4j-core-1.19.0.jar
Add to classpath: /Users/bsorrentino/Library/Jupyter/kernels/rapaio-jupyter-kernel/mima_cache/org/jspecify/jspecify/1.0.0/jspecify-1.0.0.jar
Add to classpath: /Users/bsorrentino/Library/Jupyter/kernels/rapaio-jupyter-kernel/mima_cache/dev/langchain4j/langchain4j-http-client/1.19.0/langchain4j-http-client-1.19.0.jar
Add to classpath: /Users/bsorrentino/Library/Jupyter/kernels/rapaio-jupyter-kernel/mima_cache/dev/langchain4j/langchain4j-http-client-jdk/1.19.0/langchain4j-http-client-jdk-1.19.0.jar
Add to classpath: /Users/bsorrentino/Library/Jupyter/kernels/rapaio-jupyter-kernel/mima_cache/com/fasterxml/jackson/core/jackson-annotations/2.22/jackson-annotations-2.22.jar
Add to classpath: /Users/bsorrentino/Library/Jupyter/kernels/rapaio-jupyter-kernel/mima_cache/com/fasterxml/jackson/core/jackson-core/2.22.1/jackson-core-2.22.1.jar
Add to classpath: /Users/bsorrentino/Library/Jupyter/kernels/rapaio-jupyter-kernel/mima_cache/com/fasterxml/jackson/core/jackson-databind/2.22.1/jackson-databind-2.22.1.jar
Add to classpath: /Users/bsorrentino/Library/Jupyter/kernels/rapaio-jupyter-kernel/mima_cache/com/knuddels/jtokkit/1.1.0/jtokkit-1.1.0.jar

In [3]:

import dev.langchain4j.model.openai.OpenAiChatModel;

import java.util.Objects;

var apiKey = Objects.requireNonNull(
        System.getenv("OPENAI_API_KEY"),
        "Set OPENAI_API_KEY before running this notebook."
);

var model = OpenAiChatModel.builder()
        .apiKey(apiKey)
        .modelName("gpt-4o-mini")
        .temperature(0.0)
        .build();

import dev.langchain4j.model.openai.OpenAiChatModel; import java.util.Objects; var apiKey = Objects.requireNonNull( System.getenv("OPENAI_API_KEY"), "Set OPENAI_API_KEY before running this notebook." ); var model = OpenAiChatModel.builder() .apiKey(apiKey) .modelName("gpt-4o-mini") .temperature(0.0) .build();

SLF4J: No SLF4J providers were found.
SLF4J: Defaulting to no-operation (NOP) logger implementation
SLF4J: See https://www.slf4j.org/codes.html#noProviders for further details.

Define the state¶

The state keeps the topic and each version of the joke. Each node updates only the field it produces.

In [4]:

import org.bsc.langgraph4j.state.AgentState;
import org.bsc.langgraph4j.state.Channel;
import org.bsc.langgraph4j.state.Channels;

import java.util.Map;

class JokeState extends AgentState {
    static final String TOPIC = "topic";
    static final String JOKE = "joke";
    static final String IMPROVED_JOKE = "improved_joke";
    static final String FINAL_JOKE = "final_joke";

    static final Map<String, Channel<?>> SCHEMA = Map.of(
            TOPIC, Channels.base(() -> ""),
            JOKE, Channels.base(() -> ""),
            IMPROVED_JOKE, Channels.base(() -> ""),
            FINAL_JOKE, Channels.base(() -> "")
    );

    JokeState(Map<String, Object> initData) {
        super(initData);
    }

    String topic() { return this.<String>value(TOPIC).orElse(""); }
    String joke() { return this.<String>value(JOKE).orElse(""); }
    String improvedJoke() { return this.<String>value(IMPROVED_JOKE).orElse(""); }
    String finalJoke() { return this.<String>value(FINAL_JOKE).orElse(""); }
}

import org.bsc.langgraph4j.state.AgentState; import org.bsc.langgraph4j.state.Channel; import org.bsc.langgraph4j.state.Channels; import java.util.Map; class JokeState extends AgentState { static final String TOPIC = "topic"; static final String JOKE = "joke"; static final String IMPROVED_JOKE = "improved_joke"; static final String FINAL_JOKE = "final_joke"; static final Map<String, Channel<?>> SCHEMA = Map.of( TOPIC, Channels.base(() -> ""), JOKE, Channels.base(() -> ""), IMPROVED_JOKE, Channels.base(() -> ""), FINAL_JOKE, Channels.base(() -> "") ); JokeState(Map<String, Object> initData) { super(initData); } String topic() { return this.<String>value(TOPIC).orElse(""); } String joke() { return this.<String>value(JOKE).orElse(""); } String improvedJoke() { return this.<String>value(IMPROVED_JOKE).orElse(""); } String finalJoke() { return this.<String>value(FINAL_JOKE).orElse(""); } }

Define the nodes and routing¶

In [5]:

import dev.langchain4j.data.message.UserMessage;
import org.bsc.langgraph4j.action.EdgeAction;
import org.bsc.langgraph4j.action.NodeAction;

NodeAction<JokeState> generateJoke = state -> {
    var response = model.chat(UserMessage.from("Write a short joke about " + state.topic()));
    return Map.of(JokeState.JOKE, response.aiMessage().text());
};

EdgeAction<JokeState> checkPunchline = state ->
        state.joke().contains("?") || state.joke().contains("!") ? "Pass" : "Fail";

NodeAction<JokeState> improveJoke = state -> {
    var response = model.chat(UserMessage.from("Make this joke funnier by adding wordplay: " + state.joke()));
    return Map.of(JokeState.IMPROVED_JOKE, response.aiMessage().text());
};

NodeAction<JokeState> polishJoke = state -> {
    var response = model.chat(UserMessage.from("Add a surprising twist to this joke: " + state.improvedJoke()));
    return Map.of(JokeState.FINAL_JOKE, response.aiMessage().text());
};

import dev.langchain4j.data.message.UserMessage; import org.bsc.langgraph4j.action.EdgeAction; import org.bsc.langgraph4j.action.NodeAction; NodeAction<JokeState> generateJoke = state -> { var response = model.chat(UserMessage.from("Write a short joke about " + state.topic())); return Map.of(JokeState.JOKE, response.aiMessage().text()); }; EdgeAction<JokeState> checkPunchline = state -> state.joke().contains("?") || state.joke().contains("!") ? "Pass" : "Fail"; NodeAction<JokeState> improveJoke = state -> { var response = model.chat(UserMessage.from("Make this joke funnier by adding wordplay: " + state.joke())); return Map.of(JokeState.IMPROVED_JOKE, response.aiMessage().text()); }; NodeAction<JokeState> polishJoke = state -> { var response = model.chat(UserMessage.from("Add a surprising twist to this joke: " + state.improvedJoke())); return Map.of(JokeState.FINAL_JOKE, response.aiMessage().text()); };

Build and run the workflow¶

In [6]:

import org.bsc.langgraph4j.StateGraph;

import static org.bsc.langgraph4j.StateGraph.END;
import static org.bsc.langgraph4j.StateGraph.START;
import static org.bsc.langgraph4j.action.AsyncEdgeAction.edge_async;
import static org.bsc.langgraph4j.action.AsyncNodeAction.node_async;

var workflow = new StateGraph<>(JokeState.SCHEMA, JokeState::new)
        .addNode("generate_joke", node_async(generateJoke))
        .addNode("improve_joke", node_async(improveJoke))
        .addNode("polish_joke", node_async(polishJoke))
        .addEdge(START, "generate_joke")
        .addConditionalEdges("generate_joke", edge_async(checkPunchline), Map.of(
                "Pass", END,
                "Fail", "improve_joke"
        ))
        .addEdge("improve_joke", "polish_joke")
        .addEdge("polish_joke", END);

var chain = workflow.compile();

var result = chain.invoke(Map.of(JokeState.TOPIC, "cats"))
        .orElseThrow();

System.out.println("Initial joke:\n" + result.joke());
if (result.improvedJoke().isBlank()) {
    System.out.println("\nFinal joke:\n" + result.joke());
} else {
    System.out.println("\nImproved joke:\n" + result.improvedJoke());
    System.out.println("\nFinal joke:\n" + result.finalJoke());
}

import org.bsc.langgraph4j.StateGraph; import static org.bsc.langgraph4j.StateGraph.END; import static org.bsc.langgraph4j.StateGraph.START; import static org.bsc.langgraph4j.action.AsyncEdgeAction.edge_async; import static org.bsc.langgraph4j.action.AsyncNodeAction.node_async; var workflow = new StateGraph<>(JokeState.SCHEMA, JokeState::new) .addNode("generate_joke", node_async(generateJoke)) .addNode("improve_joke", node_async(improveJoke)) .addNode("polish_joke", node_async(polishJoke)) .addEdge(START, "generate_joke") .addConditionalEdges("generate_joke", edge_async(checkPunchline), Map.of( "Pass", END, "Fail", "improve_joke" )) .addEdge("improve_joke", "polish_joke") .addEdge("polish_joke", END); var chain = workflow.compile(); var result = chain.invoke(Map.of(JokeState.TOPIC, "cats")) .orElseThrow(); System.out.println("Initial joke:\n" + result.joke()); if (result.improvedJoke().isBlank()) { System.out.println("\nFinal joke:\n" + result.joke()); } else { System.out.println("\nImproved joke:\n" + result.improvedJoke()); System.out.println("\nFinal joke:\n" + result.finalJoke()); }

Initial joke:
Why was the cat sitting on the computer? 

Because it wanted to keep an eye on the mouse!

Final joke:
Why was the cat sitting on the computer? 

Because it wanted to keep an eye on the mouse!

In [ ]: