Crate brainwires_agents 

Brainwires Agents - Agent orchestration, coordination, and lifecycle management

This crate provides the multi-agent infrastructure for autonomous task execution:

Core Components

• CommunicationHub - Inter-agent messaging bus with 50+ message types
• FileLockManager - File access coordination with deadlock detection
• ResourceLockManager - Scoped resource locking with heartbeat-based liveness
• OperationTracker - Operation tracking with heartbeat-based liveness checking
• ValidationLoop - Quality checks before agent completion (Bug #5 prevention)
• TaskManager - Hierarchical task decomposition and dependency tracking
• TaskQueue - Priority-based task scheduling with dependency awareness

Coordination Patterns

• ContractNet - Bidding protocol for agent negotiation
• Saga - Compensating transactions for distributed operations
• OptimisticConcurrency - Optimistic locking with version-based conflict detection
• WaitQueue - Queue-based coordination primitives
• MarketAllocation - Market-based task allocation
• ThreeStateModel - State snapshots for rollback support

Analysis & Validation

• ResourceChecker - Conflict detection and resolution
• ValidationAgent - Rule-based validation
• Confidence - Response confidence scoring
• WorktreeManager - Git worktree management for agent isolation

Feature Flags

• tools - Enable validation tool integration (check_duplicates, verify_build, check_syntax)

Re-exports

pub use agent_hooks::AgentLifecycleHooks;

pub use agent_hooks::ConversationView;

pub use agent_hooks::DefaultDelegationHandler;

pub use agent_hooks::DelegationRequest;

pub use agent_hooks::DelegationResult;

pub use agent_hooks::IterationContext;

pub use agent_hooks::IterationDecision;

pub use agent_hooks::ToolDecision;

pub use runtime::AgentExecutionResult;

pub use runtime::AgentRuntime;

pub use runtime::run_agent_loop;

pub use communication::AgentMessage;

pub use communication::CommunicationHub;

pub use communication::ConflictInfo;

pub use communication::ConflictType;

pub use communication::GitOperationType;

pub use confidence::ConfidenceFactors;

pub use confidence::ResponseConfidence;

pub use confidence::extract_confidence;

pub use confidence::quick_confidence_check;

pub use file_locks::FileLockManager;

pub use file_locks::LockType;

pub use operation_tracker::OperationTracker;

pub use resource_checker::ConflictCheck;

pub use resource_checker::ResourceChecker;

pub use resource_locks::ResourceLockGuard;

pub use resource_locks::ResourceLockManager;

pub use resource_locks::ResourceScope;

pub use resource_locks::ResourceType as ResourceLockType;

pub use task_manager::TaskManager;

pub use task_manager::format_duration_secs;

pub use task_queue::TaskQueue;

pub use worktree::WorktreeManager;

pub use access_control::AccessControlManager;

pub use access_control::ContentionStrategy;

pub use access_control::LockBundle;

pub use access_control::LockPersistence;

pub use git_coordination::GitCoordinator;

pub use git_coordination::GitLockRequirements;

pub use git_coordination::GitOperationLocks;

pub use git_coordination::GitOperationRunner;

pub use git_coordination::get_lock_requirements;

pub use git_coordination::git_tools;

pub use plan_executor::ExecutionApprovalMode;

pub use plan_executor::ExecutionProgress;

pub use plan_executor::PlanExecutionConfig;

pub use plan_executor::PlanExecutionStatus;

pub use plan_executor::PlanExecutorAgent;

pub use task_orchestrator::FailurePolicy;

pub use task_orchestrator::OrchestrationResult;

pub use task_orchestrator::TaskOrchestrator;

pub use task_orchestrator::TaskOrchestratorConfig;

pub use task_orchestrator::TaskSpec;

pub use workflow::WorkflowBuilder;

pub use workflow::WorkflowContext;

pub use workflow::WorkflowResult;

pub use contract_net::ContractNetManager;

pub use market_allocation::MarketAllocator;

pub use optimistic::OptimisticController;

pub use saga::SagaExecutor;

pub use state_model::StateModelProposedOperation;

pub use state_model::StateSnapshot;

pub use state_model::ThreeStateModel;

pub use wait_queue::WaitQueue;

pub use context::AgentContext;

pub use execution_graph::ExecutionGraph;

pub use execution_graph::RunTelemetry;

pub use execution_graph::StepNode;

pub use execution_graph::ToolCallRecord;

pub use pool::AgentPool;

pub use pool::AgentPoolStats;

pub use system_prompts::judge_agent_prompt;

pub use system_prompts::planner_agent_prompt;

pub use system_prompts::reasoning_agent_prompt;

pub use system_prompts::simple_agent_prompt;

pub use cycle_orchestrator::CycleOrchestrator;

pub use cycle_orchestrator::CycleOrchestratorConfig;

pub use cycle_orchestrator::CycleOrchestratorResult;

pub use cycle_orchestrator::CycleRecord;

pub use cycle_orchestrator::MergeStrategy;

pub use judge_agent::JudgeAgent;

pub use judge_agent::JudgeAgentConfig;

pub use judge_agent::JudgeContext;

pub use judge_agent::JudgeVerdict;

pub use judge_agent::MergeStatus;

pub use judge_agent::WorkerResult;

pub use planner_agent::DynamicTaskPriority;

pub use planner_agent::DynamicTaskSpec;

pub use planner_agent::PlannerAgent;

pub use planner_agent::PlannerAgentConfig;

pub use planner_agent::PlannerOutput;

pub use planner_agent::SubPlannerRequest;

pub use task_agent::FailureCategory;

pub use task_agent::TaskAgent;

pub use task_agent::TaskAgentConfig;

pub use task_agent::TaskAgentResult;

pub use task_agent::TaskAgentStatus;

pub use task_agent::spawn_task_agent;

pub use validator_agent::ValidatorAgent;

pub use validator_agent::ValidatorAgentConfig;

pub use validator_agent::ValidatorAgentResult;

pub use validator_agent::ValidatorAgentStatus;

pub use validator_agent::spawn_validator_agent;

pub use brainwires_core;

pub use brainwires_tool_system;

pub use validation_loop::*;

Modules

access_control

Unified access control manager for inter-agent coordination

agent_hooks

Agent Loop Hooks - Granular control over the agent execution loop

communication

Inter-agent communication hub and message types

confidence

Response Confidence Extraction

context

Agent Context - environment for autonomous task execution

contract_net

Contract-Net Protocol for Task Allocation

cycle_orchestrator

Cycle Orchestrator - Plan→Work→Judge loop

execution_graph

Execution DAG and telemetry for TaskAgent runs

file_locks

File locking system for multi-agent coordination

git_coordination

Git operation coordination for multi-agent systems

judge_agent

Judge Agent - LLM-powered cycle evaluator

market_allocation

Market-Based Resource Allocation with Priority Bidding

operation_tracker

Operation tracker with heartbeat-based liveness checking

optimistic

Optimistic Concurrency with Conflict Resolution

plan_executor

Plan Executor Agent - Executes plans by orchestrating task execution

planner_agent

Planner Agent - LLM-powered dynamic task planner

pool

Agent Pool - Manages a pool of background task agents

prelude

Prelude module for convenient imports

resource_checker

Cross-resource conflict detection

resource_locks

Resource locking system for build/test/git coordination

runtime

Agent Runtime - Generic execution loop for autonomous agents

saga

Saga-Style Compensating Transactions

state_model

Three-State Model for comprehensive state tracking

system_prompts

System prompts for task agents with structured reasoning

task_agent

TaskAgent - Autonomous agent that executes a task in a loop using AI + tools

task_manager

Task Manager - Manages a tree-structured task list

task_orchestrator

Task Orchestrator - Bridges TaskManager and AgentPool

task_queue

Priority-based task queue for agent scheduling

validation_agent

Independent Validation Agent for operation verification

validation_loop

Validation Loop - Enforces quality checks before agent completion

validator_agent

ValidatorAgent - Standalone read-only agent that runs external validators

wait_queue

Wait queue implementation for resource coordination

workflow

Workflow Graph Builder — Declarative DAG-based workflow pipelines

worktree

Git Worktree Management for Agent Isolation

Enums

PreHookDecision

Decision returned by a ToolPreHook before a tool call.

Traits

ToolPreHook

Pluggable pre-execution hook for semantic tool validation.