claude-agents-specialized-subagents - SKILL.md Agent Skill

name: claude-agents-specialized-subagents description: Create specialized AI subagents with dedicated context windows and custom system prompts for focused tasks like marketing copy, UX design, or planning. Trigger when you need autonomous AI employees working independently on specific domains while maintaining separate conversation contexts. Use "agents" slash command or "use the agent [name]" to invoke. 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 Agents: Specialized Subagents

Create autonomous AI employees that work in isolated context windows for specialized tasks.

Quick Start

/agents → create new agent → name: "brand-copywriter" → model: opus → color: purple → save

Then invoke: "use the agent brand copywriter to generate five headline ideas for a coffee shop website"

Core Workflow

When you need specialized expertise isolated from main conversation:

Create Agent: /agents → "Create New Agent"
Configure: Set name, select model (opus recommended), assign color
Save: Agent becomes available across all Claude Code instances
Invoke: Either /agents to browse or directly "use the agent [name] to [task]"
Monitor: Subagent appears in different color, works in separate context window
Scale: Multiple agents can work simultaneously on different tasks

Agent triggers main conversation when complete - think of it as an employee going off to work independently and returning with results.

Techniques

Agent Naming Patterns:

brand-copywriter - Marketing copy generation
ux-ui-agent - User experience design
planning-agent - Project planning and strategy

Multi-Agent Workflows:

Spin up multiple instances of same agent type
Run different agent types simultaneously
Each maintains separate context and memory

Agent Recommendations: Ask Claude: "what kind of agents do you recommend I create?" for domain-specific suggestions.

Anti-Patterns

NEVER expect agents to share context with main conversation - they work in isolation.

NEVER rely on agents remembering previous main chat history - their context window is independent.

Edge Cases & Error Handling

Agent Access: Use /agents if direct invocation fails Context Isolation: Agents cannot access your main conversation history Multiple Instances: Same agent type can run multiple copies simultaneously Exit Shortcut: Hit "exit" to return to main chat from agent view

Bundled Resources Plan

scripts/agent-templates.md - Common agent configurations and naming conventions references/agent-use-cases.md - Domain-specific agent examples and triggers