🎯 Orchestration

Agent Task Delegator

Brain-and-muscle delegation: premium model orchestrates, cheap models execute. Save 70%+ on API costs.

What it does

Agent Task Delegator implements the “one smart brain, many cheap muscles” pattern. A premium model (Opus, GPT-5.4) handles planning, reasoning, and orchestration. Cheap models (Haiku, Gemma, Qwen) handle execution — the grunt work. The delegator classifies each sub-task and routes it to the right model automatically.

It tracks delegation results, measures per-model success rates, and auto-escalates when a cheap model fails. Over time it learns which task types map to which models for your specific workflows.

Use cases

• Multi-agent setups where you want one orchestrator directing many workers
• Reducing API costs on high-volume agent workflows
• Setting up coding, research, and writing pipelines with separate model tiers
• Teams wanting to route 80% of work to cheap models without quality loss

Requirements

• OpenClaw 3.20+
• Python 3.10+
• API keys for at least two model tiers

How to install

After purchase you’ll receive a download link by email. Extract and copy the agent-task-delegator/ folder into your OpenClaw skills/ directory, then restart OpenClaw.