lambdalisue/cc-add-agent

Arguments

$ARGUMENTS = {name} {description...}

Parse the arguments:

• Name: Agent name (first token) - will become the filename (kebab-case, without .md)
• Description: What the agent does (everything else)

Target Directory

~/.claude/agents/

Workflow

1. Parse - Extract name and description from arguments
2. Analyze - Determine:
   • Agent's core purpose and expertise
   • Appropriate model (sonnet for most, opus for complex reasoning)
   • Color for visual distinction
   • Key principles and workflow
3. Draft - Create the agent file with:
   • Frontmatter: name, description, model, color
   • Brief persona/expertise statement
   • Principles (numbered list)
   • Prohibitions if applicable
   • Workflow steps
4. Confirm - Show the proposed file path and content
5. STOP - Wait for user approval (use AskUserQuestion)
6. Create - Only after approval, create the agent file

Agent Format

---
name: { name }
description: { Brief description ending with period }
color: { blue|green|yellow|red|purple|cyan }
---

{One-line persona/expertise statement}

## Principles

1. **Key Principle**: Explanation
2. ...

## Prohibitions

- **No X**: Reason

## Workflow

1. **Phase**: Actions
2. ...

## When Uncertain

{How to handle ambiguity}

Available Colors

blue, green, yellow, red, purple, cyan, magenta, orange, white

Examples

Input: /cc-add-agent api-designer Design RESTful APIs following best practices Output: ~/.claude/agents/api-designer.md

Input: /cc-add-agent security-reviewer Review code for security vulnerabilities Output: ~/.claude/agents/security-reviewer.md

Content Guidelines

Analyze the {description} to understand the intended purpose, then write the agent file content:

• In English, optimized for AI comprehension
• Use clear, imperative language
• Focus on actionable instructions rather than verbose explanations

Begin

Parse $ARGUMENTS and create the appropriate agent file. Ask for user approval before writing.