Struct ChatSession 

pub struct ChatSession { /* private fields */ }

A chat session that manages conversation history with context window limits.

ChatSession maintains a list of messages and automatically trims old messages when the total token count exceeds the configured maximum context size. The system prompt (first message) is always preserved.

Examples

use mojentic::llm::{ChatSession, LlmBroker};
use mojentic::llm::gateways::OllamaGateway;
use std::sync::Arc;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let gateway = Arc::new(OllamaGateway::default());
    let broker = LlmBroker::new("qwen3:32b", gateway);
    let mut session = ChatSession::new(broker);

    let response = session.send("What is Rust?").await?;
    println!("Response: {}", response);

    Ok(())
}

Implementations

impl ChatSession

pub fn new(broker: LlmBroker) -> Self

Create a new chat session with default settings.

Arguments

• broker - The LLM broker to use for generating responses

Examples

use mojentic::llm::{ChatSession, LlmBroker};
use mojentic::llm::gateways::OllamaGateway;
use std::sync::Arc;

let gateway = Arc::new(OllamaGateway::default());
let broker = LlmBroker::new("qwen3:32b", gateway);
let session = ChatSession::new(broker);

pub fn builder(broker: LlmBroker) -> ChatSessionBuilder

Create a chat session builder for custom configuration.

Arguments

• broker - The LLM broker to use for generating responses

Examples

use mojentic::llm::ChatSession;

let session = ChatSession::builder(broker)
    .system_prompt("You are a helpful coding assistant.")
    .temperature(0.7)
    .max_context(16384)
    .build();

pub async fn send(&mut self, query: &str) -> Result<String>

Send a message to the LLM and get a response.

This method:

1. Adds the user message to the conversation history
2. Generates a response using the LLM
3. Adds the assistant’s response to the history
4. Automatically trims old messages if context window is exceeded

Arguments

• query - The user’s message

Returns

The LLM’s response as a string

Examples

let response = session.send("What is 2 + 2?").await?;
println!("Answer: {}", response);

pub fn send_stream<'a>( &'a mut self, query: &str, ) -> Pin<Box<dyn Stream<Item = Result<String>> + 'a>>

Send a message to the LLM and get a streaming response.

This method:

1. Adds the user message to the conversation history
2. Streams the response from the LLM, yielding chunks as they arrive
3. After the stream is fully consumed, adds the assembled response to history
4. Automatically trims old messages if context window is exceeded

Arguments

• query - The user’s message

Returns

A stream of string chunks from the LLM response

Examples

use futures::stream::StreamExt;

let mut stream = session.send_stream("Tell me a story");
while let Some(result) = stream.next().await {
    print!("{}", result?);
}

pub fn insert_message(&mut self, message: LlmMessage)

Insert a message into the conversation history.

If the total token count exceeds max_context, the oldest messages are removed until the total is under the limit. The system prompt (index 0) is always preserved.

Arguments

• message - The message to add

pub fn messages(&self) -> &[SizedLlmMessage]

Get the current conversation history

pub fn total_tokens(&self) -> usize

Get the total token count of the current conversation