agent-research

Deep research with context isolation is achieved by spawning independent agent teams that operate separately to prevent context pollution, followed by independent verification stages to validate findings for accuracy.

Start an automated research investigation by executing a command with your topic, specifying depth as quick, standard, or deep, and setting output formats like SCQA, bullet, or narrative for tailored synthesis results.

Independent verification in multi-agent research uses dedicated verifiers to validate findings after parallel research and synthesis stages, ensuring accurate results by checking data gathered by isolated agents.

Yes, research agents can align with project roadmaps and deduplicate memory using pre-flight gates, ensuring memory-aware research that avoids redundant investigations and maintains context accuracy.

The best way to manage parallel research and synthesis is spawning a research team that executes parallel investigations, followed by synthesis stages to consolidate findings into a unified output format.

You should avoid multi-agent research teams for simple, single-source queries that do not require parallel investigation, context isolation, or independent verification, as the overhead outweighs the benefits.