Compact Command
Compact Command is a set of best practices for implementing sub-agents within Claude Code that emphasizes efficient context management and API cost optimization. Rather than deploying multiple independent agents with complete system context, this approach uses structured delegation patterns to minimize redundant information processing and reduce API expenses. The methodology recognizes that agentic systems often repeat context across multiple agent instances, creating unnecessary computational overhead and costs.
Core Principles
The framework prioritizes task-specific context engineering, where each sub-agent receives only the information necessary to complete its assigned work. This selective context passing reduces token consumption compared to approaches where all agents maintain identical knowledge bases. Compact Command also emphasizes clear task boundaries and delegation protocols, ensuring that work is routed to the most appropriate agent without excessive back-and-forth communication that would inflate token usage.
Implementation Considerations
Effective implementation requires careful structuring of agent responsibilities and communication patterns. Context should be designed to flow between agents efficiently, with each handoff including only relevant information. This approach works best for workflows with well-defined subtasks where dependencies and information needs can be anticipated, allowing for more predictable token consumption and faster execution times compared to open-ended multi-agent systems.
Source Notes
- 2026-04-14: I Looked At Amazon After They Fired 16,000 Engineers. Their AI Broke Everything.
- 2026-04-07: Demystifying Claude Code Key Concepts for Non Technical Users · ▶ source