Create Agent Task

This task helps you create a new BMAD agent for web platforms (Gemini, ChatGPT, etc.).

Note for User-Created Agents: If you're creating a custom agent for your own use (not part of the core BMAD system), prefix the agent ID with a period (e.g., .data-analyst) to ensure it's gitignored and won't conflict with repository updates.

Process

Follow these steps to create a new agent:

1. Gather Basic Information

Ask the user for:

Agent ID: A short, lowercase identifier (e.g., data-analyst, security-expert)
Agent Name: The character name (e.g., "Elena", "Marcus")
Title: Professional title (e.g., "Data Analyst", "Security Expert")
Description: A brief description of the agent's role and primary focus

2. Define Personality and Expertise

Ask about:

Personality traits: How should this agent behave? (professional, friendly, detail-oriented, etc.)
Communication style: How do they speak? (formal, casual, technical, empathetic)
Expertise areas: What are they exceptionally good at?
Years of experience: How senior are they in their role?
Motivations: What drives them to excel?

3. Identify Capabilities

Determine what the agent can do:

Existing tasks: Which existing tasks from /bmad-core/tasks/ should this agent know?
New tasks needed: Does this agent need any specialized tasks that don't exist yet?
Templates used: Which document templates will this agent work with?
Checklists: Which quality checklists apply to this agent's work?

4. Create the Persona File

Create /bmad-core/personas/{agent-id}.md with this structure: (For user-created agents, use /bmad-core/personas/.{agent-id}.md)

# {Agent Name} - {Title}

## Character Profile

**Name:** {Agent Name}
**Title:** {Title}
**Experience:** {Years} years in {field}

## Personality

{Describe personality traits, communication style, and approach to work}

## Core Expertise

{List main areas of expertise and specialization}

## Responsibilities

{List key responsibilities in bullet points}

## Working Style

{Describe how they approach problems, collaborate, and deliver results}

## Motivations

{What drives them to excel in their role}

## Catchphrases

{Optional: Any signature phrases or ways of speaking}

5. Create the Agent Configuration

Create /agents/{agent-id}.yml with this structure: (For user-created agents, use /agents/.{agent-id}.yml)

agent:
  id: { agent-id }
  name: { Agent Name }
  title: { Title }
  description: >-
    {Full description of the agent's role and value}
  persona: { agent-id }
  customize: >-
    {Any specific behavioral customizations}

dependencies:
  tasks:
    - { list of task IDs }
  templates:
    - { list of template IDs }
  checklists:
    - { list of checklist IDs }
  data:
    - { list of data file IDs }
  utils:
    - template-format

6. Create Any New Tasks

If new tasks were identified, create them in /bmad-core/tasks/{task-name}.md (For user-created tasks, use /bmad-core/tasks/.{task-name}.md)

7. Test and Validate

Run npm run validate to check configuration
Run npm run build:agent -a {agent-id} to build the agent
Review the generated output in /dist/agents/{agent-id}.txt

Example Questions to Ask

"What will this agent be called? (ID like 'data-analyst')"
"What's their character name? (like 'Elena')"
"What's their professional title?"
"Describe their main role in 2-3 sentences."
"What personality traits should they have?"
"How many years of experience do they have?"
"What existing tasks should they know? (e.g., create-doc-from-template, execute-checklist)"
"Do they need any specialized tasks that don't exist yet?"
"Which document templates will they use?"
"What motivates them in their work?"

Important Notes

Keep personas engaging but professional
Ensure all referenced tasks, templates, and checklists exist
Web agents can be more detailed than IDE agents (no size constraints)
Consider how this agent will collaborate with existing team members
Run validation after creating to catch any issues

4.1 KiB Raw Blame History