task-partitioner

Partitioning tasks for multi-agent systems requires generating dependency graphs and safe execution batches. This skill analyzes a task list, computes dependencies, groups parallelizable tasks into batches, and outputs an execution plan with a file ownership map to prevent conflicts.

A file ownership map assigns a disjoint set of files to each task in a multi-agent execution plan. This prevents conflicts and race conditions by ensuring parallel agents operate on strictly separate files during batch execution.

Parallel execution batches are planned by analyzing a task list to compute dependencies and grouping independent tasks into executable groups. The process outputs an execution plan with defined boundaries, enabling safe parallel agent execution across complex projects.

Dependency graphs schedule independent tasks across parallel agents by identifying prerequisites and grouping independent tasks. This allows the system to split large task graphs into executable batches with defined boundaries for safe parallel processing.

Manual dependency definition is not required before generating a multi-agent execution plan. The skill performs automated dependency analysis to discover prerequisites and independent tasks directly from the provided task list.

Multi-agent task partitioning leads to unclear file ownership when tasks lack disjoint file assignments. Generating a per-task file ownership map resolves this by assigning a strictly separate set of files to each task to prevent conflicts.