Module memory 

Memory: manages the agent’s context.

The inference loop uses Memory to store conversation history and to retrieve the full context for each request. This module provides two ready-to-use implementations:

• InMemoryMemory — stores all messages verbatim, never trims;
• WindowMemory — trims to a recent window against a budget, with pluggable token counting and trim strategies.

Compaction, retrieval, persistence and similar strategies plug in by implementing Memory (or injecting a TrimStrategy into WindowMemory) without touching the inference loop. A built-in summarization strategy is provided as SummarizeStrategy.

Structs

Budget

Window budget: token and round limits (both optional; when both are set, the smaller window wins).

CharTokenCounter

Default counter: CJK characters count as 1 token each, other characters count as 1 token per 4 characters (rounded up).

InMemoryMemory

In-memory implementation: stores all messages verbatim.

SummarizeStrategy

Summarization trim strategy: over-budget old messages → summarizer → one System summary message.

TrimResult

Output of a trim strategy: the trimmed message sequence + how the result is handled.

WindowDrop

Default trim strategy: drops the earliest complete rounds until the remaining sequence fits the budget.

WindowMemory

Window memory: stores all messages; context() trims to the budget on retrieval.

Enums

MemoryError

Context access failed.

Traits

Memory

Manages the agent’s context: decides which messages the model sees on each turn.

TokenCounter

Counts the token number of a text.

TrimStrategy

Trim strategy: decides “how to trim” — window dropping, summarization, LLM compaction, etc. Users can inject custom implementations.