dispatching-parallel-agents

Fix multiple independent test failures concurrently by dispatching parallel agents, each assigned a distinct failing test file with a self-contained scope to investigate root causes and propose fixes simultaneously.

Parallel agent dispatch for debugging assigns one agent per independent problem domain so investigations run concurrently, requiring each agent to have a self-contained scope, clear constraints, and return a concise summary of root cause and changes.

Use parallel agents to triage codebase failures when three or more unrelated test files or subsystems require concurrent investigation with no shared state, ensuring each problem can be understood and fixed independently.

Design prompts for parallel agents by giving each a self-contained scope, explicit goals, constraints, and expected outputs to avoid drift, then request a concise summary of root cause and proposed fixes for integration.

Integrate fixes after parallel agents investigate independent failures by reviewing each agent's concise summary and proposed fixes, performing conflict checks, and running full-suite verification to ensure no regressions across the codebase.

Limitations of concurrent debugging with parallel agents include the requirement that failures have no shared state, the need for each agent to have a self-contained scope, and the necessity of conflict checks and full-suite verification after integration.