Dispatching Parallel Agents

Coordinate debugging by dispatching dedicated agents to investigate and fix failures concurrently. This parallel orchestration drastically reduces debugging time and improves fault isolation across independent problem domains.

The best way to debug multiple failing tests is assigning one agent per independent module to work concurrently. Agents use structured prompts and success criteria to investigate failures, aggregating summaries for safe integration.

You can use parallel agents to diagnose independent regressions by dispatching multiple agents to independently investigate and fix failures. This applies to scenarios with 3 or more independent failures that have no shared state.

Parallel agent dispatch ensures safe integration by collecting summaries and merging fixes without conflicts. Agents operate on independent failure domains with no shared state, preserving system safety during the merge.

You should avoid using parallel agents for debugging when dealing with fewer than 3 independent failures or when failures share state. This approach requires independent failure domains to ensure safe, conflict-free integration.