subagent-delegation

To delegate parallelizable tasks without context bloat, offload independent, research-heavy processes to isolated subagents that return only a final synthesis. This maintains main-thread efficiency by keeping deep dives and inventory scans separated from your primary working context.

The best way to run parallel research investigations for security audits is delegating them to autonomous subagents. This approach executes multiple independent investigations simultaneously, achieving N-way speedups while keeping the main repository stable through git worktree management.

Git worktree management for subagent delegation automatically manages isolated branches for code-modifying tasks. This ensures the main repository remains stable and unaffected until verification loops are complete, providing operational safety and auditability for multi-file refactoring.

Yes, structured JSON return contracts are required for subagent delegation during multi-file refactoring. These contracts ensure operational safety and auditability by enforcing clean verification loops and standardized data returns from isolated subagent processes.

Yes, you can use subagent delegation for context-heavy inventory scans and multi-file refactoring. It offloads these independent tasks to isolated processes, ensuring the main thread remains clean and efficient while requiring structured JSON return contracts for safe verification.

You should not use subagent delegation for tasks that are not independent, context-heavy, or parallelizable. If your workflow lacks clean verification loops or cannot enforce structured JSON return contracts, the overhead of managing isolated subagent processes and git worktrees outweighs the benefits.