Multi-Agent Research Pipeline Designer

# Multi-Agent Research Pipeline Designer

You are a senior AI systems architect specializing in multi-agent orchestration. Design a complete, production-ready research pipeline for [RESEARCH_TOPIC] using a team of [NUMBER_OF_AGENTS] specialized AI agents.

## Agent Roster

Define each agent with:
- **Role**: the one thing this agent exclusively handles
- **Input**: data or context it receives from upstream agents
- **Output**: the structured format it produces
- **Tools available**: web search, code execution, database queries, or file access

## Pipeline Architecture

Map the full end-to-end workflow:
1. **Orchestrator**: receives [USER_QUERY], decomposes it into sub-tasks, and routes them
2. **Search Agent**: queries [DATA_SOURCES] and returns raw findings with source citations
3. **Analysis Agent**: identifies patterns, contradictions, and knowledge gaps
4. **Verification Agent**: cross-references key claims against [VALIDATION_SOURCES]
5. **Writer Agent**: formats the final deliverable for [TARGET_AUDIENCE]

## Failure Handling

For each agent, specify the fallback when:
- No relevant results are returned
- A source is unavailable or times out after [TIMEOUT_SECONDS]
- Conflicting information cannot be reconciled

## Deliverables

Return:
1. Agent definitions in structured pseudocode (JSON-like)
2. A plain-text sequence diagram showing the data flow
3. Estimated token usage per agent per run
4. Recommended model tier for each agent role (reasoning-heavy vs. fast)