Module brain 

Brain module — LLM provider abstraction and interaction.

Defines the LlmProvider trait for model-agnostic LLM interactions, and provides an OpenAI-compatible implementation with streaming support.

Structs

Brain

The Brain wraps an LLM provider and adds higher-level logic: prompt construction, cost tracking, and model selection.

MockLlmProvider

A mock LLM provider for testing and development.

TokenBudgetManager

Tracks token usage against configurable budgets and predicts costs before execution. Can warn or halt when budgets are exceeded.

TokenCounter

Token counter using tiktoken-rs for accurate BPE tokenization.

Enums

BudgetCheckResult

The result of a pre-call budget check.

Constants

DEFAULT_SYSTEM_PROMPT

The system prompt used by default for the Rustant agent.

Traits

LlmProvider

Trait for LLM providers, supporting both full and streaming completions.

Functions

sanitize_tool_sequence

Sanitize tool_call → tool_result ordering in a message sequence.