Semantic Cache for cortex

Provides caching with optional semantic similarity matching for LLM responses. Supports multiple backends:

• In-Memory: Fast LRU cache with TTL support (default)
• Redis/Valkey: Distributed cache for multi-instance deployments

Features

• memory (default): Enable in-memory LRU cache
• redis: Enable Redis/Valkey backend
• full: Enable all backends

Example

use cortexai_cache::{Cache, MemoryCache, CacheConfig};

// In-memory cache
let cache = MemoryCache::new(CacheConfig::default());

// Store a response
cache.store("What is Rust?", "context", "Rust is a systems programming language", vec![]).await?;

// Retrieve (exact match)
if let Some(entry) = cache.get("What is Rust?", "context").await? {
    println!("Cached response: {}", entry.response);
}