By name: claude-agent-delegation-system description: Creates and manages specialized Claude agents for delegated tasks with their own context windows. Triggered when needing domain-specific expertise (marketing, UX, planning, etc.) or when wanting to parallelize different types of work. Use "use the agent [name]" to invoke agents that work independently like employees with custom instructions. source: 10 Claude Code tips I wish I knew from the start + https://youtu.be/BEIulrjHzMI?si=JiZ_XdfyLiw9nCGP generated_by: SkillJack (skilljacked.com)
Claude Agent Delegation System
Create specialized Claude agents that work independently in their own context windows for domain-specific tasks.
Quick Start
use the agent brand copywriter to generate five headline ideas for a coffee shop website
Core Workflow
When you need specialized expertise or want to delegate specific tasks:
- Access agents menu → Navigate to agents section in Claude interface
- Invoke agent → Use "use the agent [agent-name]" command in chat
- Agent activates → Subagent appears in different color, working in isolated context
- Receive results → Agent returns with specialized output and reasoning
- Exit back → Hit 'exit' to return to main chat context
When creating new agents:
- Ask Claude: "what kind of agents do you recommend I create?"
- Create agents for specific domains (UX/UI, planning, marketing, etc.)
Techniques
Agent Invocation Patterns
- Direct command: "use the agent [name] to [specific task]"
- Implicit invocation: Natural language that implies need for specialized agent
- Parallel processing: Spin up multiple instances of same agent type simultaneously
Agent Specialization Areas
- Brand copywriter (headlines, marketing copy)
- UX/UI designer (interface design, user experience)
- Planning agent (project organization, strategy)
- Technical specialist agents (domain-specific coding, architecture)
Context Management
- Each agent operates in isolated context window
- Main chat remains separate from agent work
- Agents can maintain their own conversation threads
Anti-Patterns
NEVER try to manage complex multi-domain tasks in single context when agents could specialize NEVER forget that agents work independently - they don't share context with main chat NEVER assume agents remember previous sessions - each invocation is fresh context
Edge Cases & Error Handling
- Agent not found: Create agent first or verify agent name spelling
- Context switching: Use 'exit' command to cleanly return to main chat
- Multiple agent coordination: Results need manual integration in main context
- Agent specialization overlap: Define clear boundaries between agent roles