Multi-Agent Research Team

A SOUL.md + AGENTS.md configuration that turns your OpenClaw into a research orchestrator. When you ask a complex research question, the main agent automatically spawns 2-3 sub-agents to research different angles in parallel, then synthesizes their findings.

What's Included

• SOUL.md — Research orchestrator personality with decompose → delegate → synthesize protocol
• AGENTS.md — Sub-agent spawning rules, handoff conventions, and cost management
• openclaw.json snippet — Cheaper model for sub-agents, max 3 concurrent
• README.md — Usage guide with examples

How It Works

1. You ask a complex question
2. Main agent breaks it into 2-3 research angles
3. Sub-agents research each angle in parallel (using cheaper models)
4. Main agent synthesizes findings, flags contradictions, cites sources

Includes smart routing: simple questions answered directly, complex ones get the full research team treatment.

1. Copy SOUL.md and AGENTS.md to ~/.openclaw/workspace/
2. Merge the openclaw.json snippet into your config
3. Ask a complex question and watch the sub-agents work!
4. Research summaries are saved to memory/research/