Module context 

Session context management for AI agents

This module provides intelligent context management for AI agents, including conversation history, task context, and workspace state. The context system is designed to maximize AI performance while minimizing token usage through intelligent compression and summarization.

Key Features

• Efficient History Management: Real zstd compression for optimized context handling
• Task Context: Structured task and goal tracking for AI agents
• Agent State: Persistent agent memory and decision tracking
• Workspace State: File system and project state awareness
• Smart Compression: Zstd-based compression with message summarization

Examples

Basic Context Management

use ai_session::context::{SessionContext, Message, MessageRole};
use ai_session::SessionId;
use chrono::Utc;

let session_id = SessionId::new();
let mut context = SessionContext::new(session_id);

// Add conversation messages
let user_message = Message {
    role: MessageRole::User,
    content: "Help me implement a REST API".to_string(),
    timestamp: Utc::now(),
    token_count: 7,
};
context.add_message(user_message);

let assistant_message = Message {
    role: MessageRole::Assistant,
    content: "I'll help you create a REST API. Let's start with the basic structure...".to_string(),
    timestamp: Utc::now(),
    token_count: 18,
};
context.add_message(assistant_message);

// Check context stats
println!("Messages: {}", context.get_message_count());
println!("Total tokens: {}", context.get_total_tokens());

Context Compression

use ai_session::context::{SessionContext, Message, MessageRole};
use ai_session::SessionId;

let session_id = SessionId::new();
let mut context = SessionContext::new(session_id);

// Fill context with many messages...
for i in 0..100 {
    let message = Message {
        role: MessageRole::User,
        content: format!("Message {}", i),
        timestamp: chrono::Utc::now(),
        token_count: 5,
    };
    context.add_message(message);
}

println!("Before compression: {} tokens", context.get_total_tokens());

// Compress when approaching token limit
if context.get_total_tokens() > 400 {
    let compressed = context.compress_context().await;
    if compressed {
        println!("After compression: {} tokens", context.get_total_tokens());
    }
}

Structs

AgentState

Agent state

CompressedHistory

Compressed history with zstd compression

CompressionStats

Compression statistics

ContextSummary

Context summary

FileChange

File change event

FileState

File state

Message

A message in the conversation

SessionConfig

Session configuration

SessionContext

Session context containing AI-relevant state

TaskContext

Current task context

TokenEfficientHistory

Token-efficient conversation history with real zstd compression

WorkspaceState

Workspace state

Enums

FileChangeType

Type of file change

MessageRole

Message role

TaskPriority

Task priority