super-swarm-spark

You can achieve parallel task execution by providing a plan file path to initialize a rolling pool of up to 15 concurrent subagents. The system parses the plan, builds per-task context packs, and executes tasks in parallel until completion.

Plan-aware orchestration parses plan files to build per-task context packs, enforcing canonical paths and explicit file targets. This mechanism reduces drift and ensures deterministic task execution across concurrent subagents.

Progress tracking and context management are handled automatically by validating results, logging outcomes, and committing completed steps to the central plan. This ensures continuous progress and automatic plan updates during parallel execution.

Yes, the system implements a max-concurrency scheduler that coordinates a rolling pool of up to 15 subagents. You can start the pool by providing the plan file path and optional task IDs to manage orchestration scale.

The rolling pool limits concurrency to a maximum of 15 subagents. It also requires explicit targets, deterministic task execution, and strict completion criteria to maintain robust validation and reduce drift across tasks.