Dispatching Parallel Agents

Dispatching parallel agents lets you investigate and fix independent failures concurrently instead of sequentially. Assign each agent a narrow scope—one test file or subsystem—so they work without conflicts, drastically reducing total debugging time when you have 3+ unrelated problems.

Use parallel agents when you have three or more independent failures with no shared state or dependencies across different test files or subsystems. This approach eliminates wasted time waiting and context switching, making it ideal for multi-subsystem debugging sessions.

The skill enforces explicit scope boundaries, non-overlapping change zones, and mandatory output summaries per agent. Integrate work progressively using conflict-resolution workflows to verify each agent's changes before merging, maintaining code integrity across concurrent fixes.

No. Parallel agents are designed for independent failures across different test files or subsystems with no shared state. Failures in the same file or with dependencies require sequential investigation to avoid conflicts and ensure correctness.

The skill requires three or more independent failures to be effective. With fewer failures or tightly coupled problems, sequential debugging may be more straightforward than coordinating parallel agent work and conflict resolution.