Fundamentals

Every chain type uses a *Builder struct with method chaining and a .build() that returns Result<Chain, ChainError>. Required fields cause a ChainError::MissingObject if absent.

let chain = LLMChainBuilder::new()
    .prompt(formatter)          // required
    .llm(llm)                   // required
    .options(ChainCallOptions::default())  // optional
    .output_key("answer")       // optional, default "output"
    .build()?;

use langchainx::{
    message_formatter, fmt_message, fmt_template,
    prompt::{HumanMessagePromptTemplate, MessageOrTemplate},
    prompt_args, template_fstring,
    schemas::Message,
};

// Build a multi-message prompt template
let prompt = message_formatter![
    fmt_message!(Message::new_system_message("You are a helpful assistant.")),
    fmt_template!(HumanMessagePromptTemplate::new(
        template_fstring!("{input}", "input")
    )),
];

// Build input variables
let input = prompt_args! {
    "input" => "What is Rust?",
};

use langchainx::{
    chain::{Chain, LLMChainBuilder},
    llm::openai::OpenAI,
    message_formatter, fmt_message, fmt_template,
    prompt::{HumanMessagePromptTemplate, MessageOrTemplate},
    prompt_args, template_fstring,
    schemas::Message,
};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let llm = OpenAI::default();

    let prompt = message_formatter![
        fmt_message!(Message::new_system_message("You are helpful.")),
        fmt_template!(HumanMessagePromptTemplate::new(
            template_fstring!("{input}", "input")
        )),
    ];

    let chain = LLMChainBuilder::new()
        .prompt(prompt)
        .llm(llm)
        .build()?;

    // invoke: returns String
    let output = chain.invoke(prompt_args! { "input" => "Hello" }).await?;
    println!("{output}");

    // call: returns GenerateResult (includes token usage)
    let result = chain.call(prompt_args! { "input" => "Hello" }).await?;
    println!("{}", result.generation);
    println!("{:?}", result.tokens);

    Ok(())
}

#[async_trait]
pub trait Chain: Sync + Send {
    async fn call(&self, input_variables: PromptArgs) -> Result<GenerateResult, ChainError>;
    async fn invoke(&self, input_variables: PromptArgs) -> Result<String, ChainError>;
    async fn execute(&self, input_variables: PromptArgs) -> Result<HashMap<String, Value>, ChainError>;
    async fn stream(&self, input_variables: PromptArgs)
        -> Result<Pin<Box<dyn Stream<Item = Result<StreamData, ChainError>> + Send>>, ChainError>;
    fn get_input_keys(&self) -> Vec<String>;
    fn get_output_keys(&self) -> Vec<String>;
}

• invoke — returns just the generated string (most common)
• call — returns GenerateResult with token usage
• execute — returns HashMap<String, Value> keyed by output key
• stream — returns a Stream of StreamData chunks

prompt_args! keys must exactly match the variable names in template_fstring!.

// WRONG: template uses "input" but args use "query"
let prompt = message_formatter![fmt_template!(
    HumanMessagePromptTemplate::new(template_fstring!("{input}", "input"))
)];
let args = prompt_args! { "query" => "Hello" }; // key mismatch — runtime error

// CORRECT: keys match
let args = prompt_args! { "input" => "Hello" };