parallel-task-decomposition

You can break down broad user goals into independent, scoped subtasks executed by parallel agents. Each subtask receives a compact governance contract to guide model behavior, required outputs, and validation.

Task decomposition for parallel agents is the process of splitting broad goals into independent, non-overlapping read-only or write lanes. It enforces route integrity and scoped ownership to enable safer parallel delegation.

You resolve conflicts from parallel codebase investigations by integrating delegated outputs back into the orchestrator thread. The main thread collects results, resolves conflicts, and performs final validation.

Yes, parallel task decomposition supports multi-area code reviews by splitting the review into independent read-only subtasks with non-overlapping scopes. This allows parallel agents to investigate distinct codebase areas safely.

You should avoid parallel subtask decomposition for feature ports when the implementation requires overlapping write scopes across components. Non-overlapping write lanes and route integrity are mandatory, making tightly coupled goals unsuitable for parallel delegation.