cto-subagent-orchestration

Orchestrate parallel workstreams by assigning each subagent narrow ownership boundaries and a clear output. This ensures work stays separated, easy to synthesize, and avoids duplicating effort across independent analyses.

Use autonomous subagents for architecture analysis and verification when expected gains in speed, quality, and risk reduction outweigh the coordination cost. Delegate only independent parts with a clear critical path that benefit from parallel execution.

The best way to split independent tasks for parallel review is to identify a clear critical path. Then, allocate focused ownership boundaries to each subagent and synthesize the returned results concisely without duplicating work.

Delegating tasks to subagents does increase coordination cost, but it speeds up the overall workflow. Weigh the coordination cost against the expected speed and quality gains before deciding to delegate independent workstreams.

Synthesize returned results from multiple subagents by enforcing focused ownership and narrow responsibility boundaries during delegation. This ensures the returned outputs remain distinct, easy to combine, and free of conflicts.

You should not use subagent orchestration when a task lacks a clear critical path or narrow ownership boundaries. If the coordination cost outweighs the expected speed and quality gains, keep the work with the main agent.