Crate rig_memory 

Conversation memory policies for the Rig agent framework.

rig-core provides the ConversationMemory trait and an in-process InMemoryConversationMemory backend. This crate adds reusable, named transformations for shaping loaded history before it is sent to the model:

• NoopMemoryPolicy — identity, returns input unchanged.
• SlidingWindowMemory — retains the most recent N messages.
• TokenWindowMemory — retains messages that fit within a token budget.
• HeuristicTokenCounter — provider-agnostic, zero-dependency TokenCounter that approximates token cost from character lengths.
• DemotionHook + DemotingPolicyMemory — bridge truncated turns from a MemoryPolicy into a long-tail store.
• Compactor + CompactingMemory — replace truncated turns with a derived summary artifact (rolling-summary semantics).
• TemplateCompactor — zero-dependency reference Compactor that produces a textual rollup without calling an LLM.

All sliding policies drop a leading orphan tool-result message when the preceding assistant tool call has been truncated, since most providers reject unpaired tool results.

Example

use rig_memory::{InMemoryConversationMemory, IntoFilter, SlidingWindowMemory};

let memory = InMemoryConversationMemory::new()
    .with_filter(SlidingWindowMemory::last_messages(20).into_filter());

Structs

CompactingMemory

A ConversationMemory adapter that wraps a backend with a MemoryPolicy and a Compactor, replacing truncated turns with a summary artifact spliced at the front of the loaded history.

DemotingPolicyMemory

A ConversationMemory adapter that wraps a backend with a MemoryPolicy and a DemotionHook, so messages truncated by the policy flow into the hook before the active window is returned.

HeuristicTokenCounter

A provider-agnostic TokenCounter that approximates token counts from UTF-8 byte lengths.

InMemoryConversationMemory

Re-exports of the core memory abstractions so callers only need a single dependency on rig-memory for both the trait/backend and the policies. A simple thread-safe in-memory ConversationMemory backed by a HashMap.

NoopDemotionHook

Re-exports of the core memory abstractions so callers only need a single dependency on rig-memory for both the trait/backend and the policies. A DemotionHook that does nothing. Useful as a default when an adapter requires a hook value but the caller has no long-tail store wired up yet.

NoopMemoryPolicy

A MemoryPolicy that returns its input unchanged.

PolicyMemory

Wrap a ConversationMemory backend with a MemoryPolicy, propagating policy errors to the caller as MemoryError::Policy.

SlidingWindowMemory

A MemoryPolicy that retains only the most recent max_messages entries.

TemplateCompactor

A zero-dependency reference Compactor that produces a textual rollup of evicted messages without calling an LLM.

TextSummary

Plain-text artifact produced by TemplateCompactor.

TokenWindowMemory

A MemoryPolicy that retains the most recent messages up to a token budget.

Enums

MemoryError

Re-exports of the core memory abstractions so callers only need a single dependency on rig-memory for both the trait/backend and the policies. Errors produced by a ConversationMemory backend.

Traits

Compactor

Re-exports of the core memory abstractions so callers only need a single dependency on rig-memory for both the trait/backend and the policies. Derives a single Message-shaped artifact from a slice of messages that a memory policy has evicted from the active window.

ConversationMemory

Re-exports of the core memory abstractions so callers only need a single dependency on rig-memory for both the trait/backend and the policies. A persistent conversation history backend.

DemotionHook

Re-exports of the core memory abstractions so callers only need a single dependency on rig-memory for both the trait/backend and the policies. A side-channel for messages that a memory policy or adapter removes from active history during ConversationMemory::load.

IntoFilter

Adapt a MemoryPolicy into a closure suitable for InMemoryConversationMemory::with_filter.

MemoryPolicy

A transformation applied to messages loaded from a ConversationMemory.

TokenCounter

Counts the tokens contributed by a single Message.