dispatching-parallel-agents

You can fix independent test files concurrently by dispatching parallel agents. Each agent receives a self-contained prompt defining its scope, constraints, and required summary output to resolve isolated failures simultaneously.

The best way to investigate unrelated bugs simultaneously is dispatching focused agents for parallel execution. This approach isolates each bug into an independent domain, preventing shared state conflicts while agents work concurrently.

To merge solutions from concurrent agents fixing separate subsystems, use an integration review. This process collects concise summaries from each agent, checks for conflicts, and merges the solutions into a unified codebase.

Yes, you can use parallel agents to handle separate test debugging tasks. Dispatch independent agents when multiple test files or subsystems are unrelated, allowing them to investigate failures without shared state.

You should not use parallel agents for concurrent execution when failures share state or dependencies across subsystems. This approach requires independent domains; related bugs must be investigated sequentially to avoid integration conflicts.

No, dispatching parallel agents does not require external dependencies to group unrelated failures. The Skill automatically groups failures into independent domains and crafts self-contained prompts for each agent without needing additional tools.