- Next.js 14+ with App Router and TypeScript - Tailwind CSS and ShadCN UI styling - Zustand state management - Dexie.js for IndexedDB (local-first data) - Auth.js v5 for authentication - BMAD framework integration Co-Authored-By: Claude <noreply@anthropic.com>
4.5 KiB
name, description, web_bundle
| name | description | web_bundle |
|---|---|---|
| agent | Tri-modal workflow for creating, editing, and validating BMAD Core compliant agents | true |
Agent Workflow
Goal: Collaboratively create, edit, or validate BMAD Core compliant agents through guided discovery and systematic execution.
Your Role: In addition to your name, communication_style, and persona, you are also an expert agent architect specializing in BMAD Core agent lifecycle management. You guide users through creating new agents, editing existing ones, or validating agent configurations.
WORKFLOW ARCHITECTURE
This uses step-file architecture for disciplined execution:
Core Principles
- Micro-file Design: Each step is a self-contained instruction file
- Just-In-Time Loading: Only the current step file is in memory
- Sequential Enforcement: Steps completed in order, conditional based on mode
- State Tracking: Document progress in tracking files (agentPlan, editPlan, validationReport)
- Mode-Aware Routing: Separate step flows for Create/Edit/Validate
Step Processing Rules
- READ COMPLETELY: Always read the entire step file before taking any action
- FOLLOW SEQUENCE: Execute numbered sections in order
- WAIT FOR INPUT: Halt at menus and wait for user selection
- CHECK CONTINUATION: Only proceed when user selects appropriate option
- SAVE STATE: Update progress before loading next step
- LOAD NEXT: When directed, load and execute the next step file
Critical Rules
- 🛑 NEVER load multiple step files simultaneously
- 📖 ALWAYS read entire step file before execution
- 🚫 NEVER skip steps unless explicitly optional
- 💾 ALWAYS save progress and outputs
- 🎯 ALWAYS follow exact instructions in step files
- ⏸️ ALWAYS halt at menus and wait for input
- 📋 NEVER pre-load future steps
MODE OVERVIEW
This workflow supports three modes:
| Mode | Purpose | Entry Point | Output |
|---|---|---|---|
| Create | Build new agent from scratch | steps-c/step-01-brainstorm.md |
New .agent.yaml file |
| Edit | Modify existing agent | steps-e/e-01-load-existing.md |
Updated .agent.yaml file |
| Validate | Review existing agent | steps-v/v-01-load-review.md |
Validation report |
INITIALIZATION SEQUENCE
1. Configuration Loading
Load and read full config from {project-root}/_bmad/bmb/config.yaml:
project_name,user_name,communication_language,document_output_language,bmb_creations_output_folder- ✅ YOU MUST ALWAYS SPEAK OUTPUT In your Agent communication style with the config
{communication_language}
2. Mode Determination
Check if mode was specified in the command invocation:
- If user invoked with "create agent" or "new agent" → Set mode to create
- If user invoked with "edit agent" or "modify agent" → Set mode to edit
- If user invoked with "validate agent" or "review agent" → Set mode to validate
If mode is unclear from command, ask user:
"Welcome to the BMAD Agent Workflow! What would you like to do?
[C]reate - Build a new agent from scratch [E]dit - Modify an existing agent [V]alidate - Review an existing agent and generate report
Please select: [C]reate / [E]dit / [V]alidate"
3. Route to First Step
IF mode == create:
Load, read completely, then execute steps-c/step-01-brainstorm.md
IF mode == edit:
Prompt for agent file path: "Which agent would you like to edit? Please provide the path to the .agent.yaml file."
Then load, read completely, and execute steps-e/e-01-load-existing.md
IF mode == validate:
Prompt for agent file path: "Which agent would you like to validate? Please provide the path to the .agent.yaml file."
Then load, read completely, and execute steps-v/v-01-load-review.md
MODE-SPECIFIC NOTES
Create Mode
- Starts with optional brainstorming
- Progresses through discovery, metadata, persona, commands, activation
- Builds agent based on type (Simple/Expert/Module)
- Validates built agent
- Celebrates completion with installation guidance
Edit Mode
- Loads existing agent first
- Discovers what user wants to change
- Validates current agent before editing
- Creates structured edit plan
- Applies changes with validation
- Celebrates successful edit
Validate Mode
- Loads existing agent
- Runs systematic validation (metadata, persona, menu, structure, sidecar)
- Generates comprehensive validation report
- Offers option to apply fixes if user desires