deep-research

Evidence-based technology research systematically investigates codebases and architectural patterns to generate documented findings. A wave-based workflow coordinates three parallel agents per wave, enforcing strict wave gating and synthesizing outputs before advancing to the next phase.

To conduct deep research for architectural analysis, initiate a four-wave workflow with three parallel agents per wave. This process evaluates tool suitability and platform decisions, enforcing evidence-first synthesis and explicit contradiction resolution to produce a final report.

Yes, you can use deep research for competitive analysis and complex codebase investigation. The workflow applies strict wave gating across decision domains, ensuring primary sources and reproducible evidence inform the final synthesized recommendations.

The best way to resolve conflicting findings during project research is through explicit contradiction resolution. The workflow enforces evidence-first synthesis during wave gating, requiring three parallel agents to reconcile differences before generating the final report.

The four-wave research process generates per-wave artifacts like wave_01_exploration.md and wave_04_synthesis.md, culminating in a final_report.md. These deliverables document evidence-based findings to support technology decisions and reduce risk.

You should avoid a multi-agent wave-gated approach for technology evaluation when rapid, lightweight assessments are needed. The strict wave gating and three-agent-per-wave requirement add overhead designed for rigorous, deep codebase investigation rather than quick checks.