design-an-interface

You can generate multiple API interface designs by spawning 3+ parallel sub-agents that produce radically different module shapes, capturing signatures, usage examples, and hidden internals for side-by-side evaluation.

Comparing module architecture options is best done by generating radically different interface designs in parallel, then performing a structured evaluation and synthesis to choose the best approach.

Yes, you can use multi-agent parallelism to explore API design decisions by coordinating 3+ sub-agents to simultaneously generate diverse interface options and capture what each design hides internally.

You evaluate interface designs by comparing the captured signatures and usage examples from each parallel sub-agent, analyzing what each module shape hides internally to synthesize the best approach.

You should use parallel sub-agents for interface generation when you need to explore diverse API design options rapidly and want a structured evaluation to compare radically different module shapes side-by-side.