deep-researchlisted

# Deep Research (Agent Teams) Conduct deep, multi-source research using a coordinated Agent Team. Multiple researcher agents work in parallel on different sub-topics, share discoveries via messaging, and the team lead synthesizes findings into a structured report. ## Architecture ``` Team Lead (you) ├── researcher-1 (sub-topic A) ├── researcher-2 (sub-topic B) ├── researcher-3 (sub-topic C) └── ...up to researcher-6 ``` - **Team Lead**: Decomposes topic, creates tasks, spawns researchers, synthesizes final report. - **Researchers**: `general-purpose` agents with WebSearch/WebFetch. Each owns one sub-topic task. Reports findings back via SendMessage. ## Workflow ### Step 1: Decompose the Research Topic Break the user's topic into 3-6 independent sub-questions. Select from these angles based on relevance: | Angle | Example sub-question | |-------|---------------------| | Background | What is X? History and origin. | | Current State | Latest developments, versions, adoption. | | Comparisons | Alternatives, trade-offs, benchmarks. | | Best Practices | Recommended approaches, common pitfalls. | | Ecosystem | Community, tooling, integrations. | | Future | Roadmap, predictions, emerging trends. | If the topic is too broad, ask the user to narrow it first. ### Step 2: Create the Research Team ``` TeamCreate: team_name: "deep-research-{topic-slug}" description: "Deep research on {topic}" agent_type: "research-lead" ``` ### Step 3: Create Tasks for Each Sub-questio