Crate atomr_agents_cache 

LLM cache.

LlmCache is the trait. Stock backends:

• InMemoryLlmCache (always available)
• SemanticLlmCache<E> — embeds the prompt; returns a cached answer if a previous prompt was within threshold cosine distance. Useful for “near-duplicate” cache hits.
• SqliteLlmCache (feature sqlite) and RedisLlmCache (feature redis): backend stubs whose real wire-up lives in a deployment patch.

Structs

CacheKey

CachedTurn

InMemoryLlmCache

SemanticLlmCache

Traits

LlmCache