Crate bamboo_engine 

Bamboo engine — runtime, agent loop, and orchestration.

Re-exports

pub use session_repository::SessionRepository;

pub use token_usage_log::TokenUsageRecord;

pub use app_context::AgentSessionContext;

pub use runtime::execution::agent_spawn::read_cached_session;

pub use runtime::execution::agent_spawn::SessionCache;

pub use session_app::child_completion_coordinator::ChildCompletionCoordinator;

pub use runtime::agent::AgentBuilder;

pub use runtime::config::ApprovalDelegate;

pub use runtime::config::AuxiliaryModelConfig;

pub use runtime::config::ChildApprovalOutcome;

pub use runtime::config::ChildApprovalRequest;

pub use runtime::config::GuardianConfig;

pub use runtime::config::GuardianSpawner;

pub use runtime::config::ImageFallbackConfig;

pub use runtime::config::ImageFallbackMode;

pub use runtime::execution::runner_state::AgentRunner;

pub use runtime::execution::runner_state::AgentStatus;

pub use runtime::hooks::HookRunner;

pub use runtime::managers::LifecycleManager;

pub use runtime::managers::LlmManager;

pub use runtime::managers::MemoryManager;

pub use runtime::managers::MiniLoopExecutor;

pub use runtime::managers::PromptManager;

pub use runtime::managers::ToolManager;

pub use runtime::model_roster::ModelRoster;

pub use runtime::model_roster::RoleModel;

pub use runtime::runtime::AgentRuntime;

pub use runtime::runtime::AgentRuntimeBuilder;

pub use runtime::runtime::ExecuteRequest;

pub use runtime::runtime::ExecuteRequestBuilder;

pub use runtime::task_context::TaskLoopContext;

pub use runtime::task_evaluation::evaluate_task_progress;

pub use runtime::task_evaluation::TaskEvaluationResult;

pub use runtime::Agent;

pub use sdk::runner::child_runner;

pub use sdk::runner::ChildRunner;

pub use sdk::runner::RunChildInput;

pub use sdk::spawn::run_child_spawn;

Modules

agent

app_context

Dependency-inversion context for the agent-session-orchestration cluster.

auto_dream

config

context

events

Engine event infrastructure.

execution

external_agents

gardener

Background “gardener”: LLM-driven remediation of memory-quality problems.

gold_auto_answer

hooks

llm_summarizer

LLM-backed conversation summarizer.

managers

message_hooks

Message preflight hooks.

model_areas

Single, explicit boundary between global and session-bound model configuration.

model_config_helper

prompt_defaults

Global prompt template defaults used by new sessions.

runner

runtime

Agent execution runtime: loop, stream handling, task evaluation.

sdk

Ergonomic SDK surface over the canonical child-spawn path.

session_app

Session management application logic.

session_repository

Canonical session coordinator owned by the framework.

stream

task_context

task_evaluation

title_gen

Backend auto-title-generation service.

token_usage_log

Append-only, per-LLM-call token-usage log.

Structs

FunctionCall

FunctionSchema

Function metadata for tool schema definition.

Message

A single message in a conversation.

PromptSnapshot

Structured snapshot of the effective system prompt and its major sections.

Session

A complete conversation session with state management.

TokenUsage

Token consumption statistics for a single LLM call or aggregated period.

ToolCall

ToolExecutionContext

Context passed to tools during execution.

ToolRegistry

Thread-safe tool registry.

ToolResult

Represents the result of executing a tool.

ToolSchema

Schema definition for a tool.

Enums

AgentError

Errors that can occur during agent operations

AgentEvent

Represents events emitted during agent execution.

MessageContent

Message content in a conversation.

MessagePart

MessagePhase

Assistant message phase used by Responses-style models.

Role

Message role in a conversation.

ToolError

Errors that can occur during tool execution

Traits

AgentHook

Trait for agent lifecycle hooks.

RuntimeSessionPersistence

Port for runtime (non-authoritative) session persistence.

Tool

Trait for implementing executable tools.

ToolExecutor

Trait for tool execution backends