---
description: Make all workflow agents project-aware by analyzing your codebase
argument-hint: (none)
---

# /eg:customize-agents

You are tasked with making all workflow agents project-aware by regenerating them with deep understanding of the specific codebase.

Input: `$ARGUMENTS`

## Step 1: Identify Agents to Customize

List all workflow agents:
```
!ls -la .claude/agents/eg/*.md
```

## Step 2: Spawn Parallel Broodmother Instances

Launch multiple broodmother instances to customize different agent groups in parallel:

### Group 1 - Specification Agents
```
Task(
  description="Customize specification agents",
  prompt=`
    First, perform deep project analysis:
    - Read ALL CLAUDE*.md files throughout the project
    - Analyze the codebase structure to assess business logic and entities
    - Identify project-specific requirements and constraints
    
    Then regenerate these agents with project-specific knowledge:
    - eg:spec-creator: Add awareness of requirement patterns, entities, and project constraints
    - eg:spec-reviewer: Include project-specific validation criteria and standards
    
    Read each agent from .claude/agents/eg/, enhance with project knowledge, and save back.
  `,
  subagent_type="broodmother"
)
```

### Group 2 - Design Agents
```
Task(
  description="Customize design agents",
  prompt=`
    First, perform deep project analysis:
    - Read ALL CLAUDE*.md files throughout the project
    - Detect frameworks, libraries, and tech stack
    - Identify authentication patterns and architectural principles
    
    Then regenerate these agents with project-specific knowledge:
    - eg:architect: Include actual architecture patterns and design principles
    - eg:architecture-reviewer: Add project-specific architectural standards
    - eg:planner: Include knowledge of project structure and conventions
    - eg:plan-reviewer: Add project-specific planning criteria
    
    Read each agent from .claude/agents/eg/, enhance with project knowledge, and save back.
  `,
  subagent_type="broodmother"
)
```

### Group 3 - Implementation Agents
```
Task(
  description="Customize implementation agents",
  prompt=`
    First, perform deep project analysis:
    - Read ALL CLAUDE*.md files throughout the project
    - Analyze coding conventions and patterns
    - Understand error handling and testing approaches
    
    Then regenerate these agents with project-specific knowledge:
    - eg:coder: Include coding standards, common imports, and idioms
    - eg:code-reviewer: Add project-specific review criteria and quality standards
    - eg:debugger: Include error patterns and debugging strategies
    
    Read each agent from .claude/agents/eg/, enhance with project knowledge, and save back.
  `,
  subagent_type="broodmother"
)
```

### Group 4 - Quality Agents
```
Task(
  description="Customize quality agents",
  prompt=`
    First, perform deep project analysis:
    - Read ALL CLAUDE*.md files throughout the project
    - Identify testing frameworks and patterns
    - Understand documentation and commit conventions
    
    Then regenerate these agents with project-specific knowledge:
    - eg:tester: Include testing frameworks and strategies
    - eg:summarizer: Add commit conventions and documentation style
    - eg:artificer: Include deep pattern understanding for templates
    
    Read each agent from .claude/agents/eg/, enhance with project knowledge, and save back.
  `,
  subagent_type="broodmother"
)
```

## Step 3: Verify Customizations

After all groups complete:
```
!echo "Agent Customization Complete
Date: $(date)

Customized Agent Groups:
- Specification: spec-creator, spec-reviewer
- Design: architect, architecture-reviewer, planner, plan-reviewer
- Implementation: coder, code-reviewer, debugger
- Quality: tester, summarizer, artificer

All agents now include:
- Project-specific patterns and conventions
- Framework and library awareness
- Architectural principles
- Testing approaches
- Error handling patterns

Agents are ready for project-specific workflows.
" > agent-customization-$(date +%Y%m%d-%H%M%S).log
```

## Step 4: Suggest Next Steps

Recommend testing the customized agents:
- Run a simple workflow to verify agents understand the project
- Check that agents follow project conventions
- Verify framework-specific knowledge is applied

## Benefits

Running agents in parallel:
- Faster customization of all agents
- Specialized focus for each group
- Consistent project knowledge
- Efficient use of resources

The customized agents will work as if built specifically for this codebase.