Module patterns 

Coordination Patterns

Reusable multi-agent patterns for common coordination scenarios. These patterns abstract away the mechanics of distributing work, collecting results, and handling failures across multiple fighters.

Structs

ChainHandler

A handler in the chain of responsibility.

MapReduceConfig

Configuration for a MapReduce operation.

MapReduceResult

Result of a MapReduce operation.

ScatterResponse

A response from a scatter-gather participant.

SupervisedWorker

State of a supervised worker.

SupervisorConfig

Configuration for the supervisor pattern.

Functions

auction_filter_bids

Filter bids to only include those that meet a minimum confidence threshold.

auction_select_winner

Run an auction: collect bids from fighters and select the winner.

chain_find_handler

Determine which handler in the chain should handle a task.

chain_walk

Walk the chain: each handler decides if it can handle, else passes along.

execute_auction

Execute the auction pattern: announce a task to all capable fighters, collect bids, and award the task to the best bidder.

execute_chain_of_responsibility

Execute the chain of responsibility pattern by sending the task through handlers until one processes it. Each handler is asked via messaging whether it can handle the task (based on capabilities). The first capable handler processes it.

execute_map_reduce

Execute a MapReduce operation with provided results.

execute_map_reduce_distributed

Execute a MapReduce operation by actually distributing work via the message router. Splits the input, sends chunks to workers, and merges results.

execute_scatter_gather

Execute the scatter-gather pattern: send a task to all capable fighters, wait for responses (with configurable timeout), and select the best result based on SelectionCriteria.

map_reduce_merge

Reduce (merge) results from multiple map workers.

map_split

Split input into chunks for map workers.

scatter_select

Select the best response from scatter-gather results.

supervisor_handle_failure

Handle a worker failure according to the supervisor strategy.

supervisor_monitor_health

Monitor worker health and handle failures via the message router. Sends heartbeat checks and restarts workers that fail to respond.