delegate

Parallel task execution across a plan is coordinated by decomposing the plan into discrete tasks, scheduling workers in dependency-aware waves, and aggregating results into a structured summary. This ensures dependencies are respected before dependent tasks begin.

Wave-based task decomposition is a deterministic workflow mechanism that groups tasks into execution waves based on their dependencies. It schedules workers so that tasks in later waves only start after all dependencies in earlier waves are resolved.

To start delegating tasks, provide a plan file path using the --plan flag and run the delegate command. The system will decompose the plan into tasks and begin scheduling workers in dependency-aware execution waves.

When a worker fails during execution, the system applies structured failure handling by blocking all dependent tasks in subsequent waves. It logs the failure in memory-audit friendly logs to ensure deterministic workflow integrity is maintained.

Yes, parallel task orchestration automatically collects, validates, and reports results from all scheduled workers. After each wave completes, the system aggregates the validated outputs into a structured summary for review.

Dependency-aware waves are used instead of running all tasks at once to enforce deterministic workflow execution. This approach prevents tasks from starting before their required inputs are ready, ensuring results are collected and validated correctly without race conditions.