claude-code-specialized-agents

name: claude-code-specialized-agents description: Create and deploy specialized AI agents with dedicated context windows, custom tools, and specific expertise areas. Triggered when needing focused AI assistance for distinct domains like marketing, UX design, or planning. Agents work independently and can run multiple instances simultaneously. 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 Code Specialized Agents

Create dedicated AI agents with their own context windows and tools for specific expertise domains.

Quick Start

Type "use the agent [agent-name]" followed by your specific request to invoke a specialized agent for focused assistance.

Core Workflow

Agent Creation

When you need dedicated expertise for a specific domain:

1. Initialize Agent: Create new agent via interface
2. Configure Tools: Select read-only tools or custom toolset based on needs
3. Set Power Level: Choose Sonnet, Opus, or Haiku based on complexity requirements
4. Assign Identity: Give descriptive name and visual identifier (color)
5. Save Configuration: Agent becomes permanently available

Agent Invocation

When you need specialized assistance within an existing chat:

1. Direct Command: Type "use the agent [agent-name]" + specific task
2. Context Isolation: Agent operates in separate context window (different color UI)
3. Independent Operation: Agent works like "an employee going off and working in its own space"
4. Result Integration: Agent returns findings to main chat thread

Multi-Agent Orchestration

When complex tasks require multiple perspectives:

1. Parallel Deployment: Spin up multiple agents simultaneously
2. Same-Type Scaling: Create multiple instances of same agent type
3. Cross-Domain Coordination: Combine different agent types (UX + marketing + planning)

Techniques

Agent Specialization Patterns

• Domain-Specific Naming: "brand copywriter", "UX UI agent", "planning agent"
• Task-Specific Requests: "generate five headline ideas for a coffee shop website"
• Contextual Handoffs: Use agents mid-conversation without losing main thread context

Access Shortcuts

• Slash Command: Type /agents to access agent directory
• Exit Command: Hit "exit" to return to main chat from agent context
• Agent Discovery: Ask Claude "what kind of agents do you recommend I create?"

Anti-Patterns

• NEVER rely on agents for tasks that don't require specialized context isolation
• NEVER create agents without clear domain boundaries
• AVOID using agents for simple queries that work fine in main context

Edge Cases & Error Handling

Agent Context Management

• Agents maintain separate context windows - information doesn't automatically transfer between main chat and agent
• Multiple same-type agents can run independently without interference
• Agent responses integrate back into main thread while preserving context separation

Power Level Selection

• Choose appropriate model tier (Haiku/Sonnet/Opus) based on task complexity
• Higher tiers consume more resources but provide better specialized reasoning

Bundled Resources Plan

• scripts/agent-templates.md - Pre-configured agent templates for common use cases
• references/agent-best-practices.md - Guidelines for effective agent specialization
• examples/multi-agent-workflows.md - Sample orchestration patterns