Deep Research Agent

This is a structured framework for building research agents that go from question to report. You get discrete stages: planning with ResearchPlanner, source evaluation with SourceEvaluator, synthesis with ResearchSynthesizer, and report generation. It covers the full pipeline from defining research questions through gathering sources from academic databases and news APIs to producing cited reports. The specialization examples for market research, competitive intelligence, and literature reviews are genuinely useful templates. It's opinionated about methodology, which is good because research agents that skip source evaluation or synthesis steps produce garbage. If you're automating any kind of intelligence gathering where quality matters more than speed, this gives you the scaffolding to do it right.