Created two comprehensive guides for v6 features:
**docs/agent-customization-guide.md**
- Complete guide for customizing agent names, personas, memories, and behaviors
- Update-safe customization via bmad/_cfg/agents/ configuration files
- Real-world examples (TDD setup, multilingual agents, custom workflows)
- Troubleshooting and best practices
**docs/web-bundles-gemini-gpt-guide.md**
- Comprehensive guide for using BMad agents in Gemini Gems and Custom GPTs
- Critical setup rules with exact configuration text required
- Cost-saving strategy: web planning → local implementation (60-80% savings)
- Platform comparison (Gemini Gems strongly recommended over Custom GPTs)
- Complete workflow examples showing full planning-to-implementation cycle
- Team bundle guidance (Gemini 2.5 Pro+ only)
**README.md updates**
- Added prominent links in v6 Core Enhancements section
- Created new "Customization & Sharing" documentation category
- Web Bundles feature highlighted with direct guide link
**Unified output folder structure across all modules:**
**Before (confusing):**
- output_folder: Main docs
- game_design_docs: Separate design folder
- tech_docs: Separate technical folder
- dev_ephemeral_location: Separate ephemeral folder outside docs
**After (simplified):**
- output_folder: Single location for ALL AI-generated artifacts (default: "docs")
- Clearer prompt: "Where should AI Generated Artifacts be saved?"
- sprint_artifacts: Phase 4 ephemeral content now WITHIN output_folder
- BMM: {output_folder}/stories (stories, context, reports)
- BMGD: {output_folder}/sprint-artifacts
- No longer in separate {bmad_folder}-ephemeral location
**Benefits:**
- One clear location for all planning artifacts (PRD, Architecture, UX, etc.)
- Phase 4 ephemeral items logically grouped within output folder
- Eliminated confusing separate folder proliferation
- sprint_artifacts now configurable per module
**Files changed:**
- src/core/_module-installer/install-config.yaml
- src/modules/bmm/_module-installer/install-config.yaml
- src/modules/bmgd/_module-installer/install-config.yaml
**Also cleaned up BMGD config:**
- Renamed: specified_framework → primary_platform (clearer naming)
- Removed: unused data_path variable
Replaced old "project_level" (0-4) system with new "selected_track" terminology:
- **quick-flow**: Bug fixes and small features (replaces Level 0-1)
- **bmad-method**: Full planning track (replaces Level 2-3)
- **enterprise-bmad-method**: Extended planning (replaces Level 4)
**Core workflow updates:**
- solutioning-gate-check: Complete rewrite of validation logic for track-based artifacts
- architecture: Updated context detection, error handling, and messaging for tracks
- workflow-init: Updated artifact detection patterns for track-based paths
- All workflow status paths updated (method-greenfield, method-brownfield, enterprise-*)
Unified variable naming conventions across all workflows:
- {output_folder} → {output-folder} (hyphenated format)
- {dev_ephemeral_location} → {sprint_artifacts} (clearer purpose)
- Hardcoded status file paths → {workflow_status_file} variable
Fixed corrupted variable patterns throughout workflow files:
- {output*folder} → {output-folder}
- {ephemeral*location} → {sprint_artifacts}
- \_prd* → *prd* (escaped underscore artifacts)
- **\*\***\_\_\_**\*\*** → proper field placeholders
Affected patterns included malformed glob patterns, template variables, and markdown formatting artifacts from previous edits.
**Architecture workflow (create-architecture):**
- Fixed: "Decision Architecture" → "Create Architecture" (consistent naming)
- Improved PRD not found handling with exit/continue options
- Better user guidance when running standalone vs. within workflow path
- Removed hardcoded Level checks, now track-aware
- Enhanced validation checklist formatting (□ → - [])
- Typo fixes: "mulitple" → "multiple"
**Solutioning gate check:**
- Complete validation logic rewrite for track-based system
- Removed Level-specific artifact expectations
- Simplified document discovery (track determines what exists)
- Better analysis prompts and user feedback
**Workflow-init:**
- Updated artifact detection patterns for new folder structure
- Fixed corrupt glob patterns throughout
- Better sprint_artifacts location detection
- Improved workflow path assignment logic
**Various workflows:**
- Consistent variable naming across 40+ workflow files
- Improved error messages and user guidance
- Better markdown formatting (checkboxes, lists)
- Removed redundant validation criteria files (now inline)
Removed duplicate BMGD 4-production workflows (12 workflows):
- code-review, correct-course, create-story, dev-story
- epic-tech-context, retrospective, sprint-planning
- story-context, story-done, story-ready
**Why:** BMGD now uses shared BMM Phase 4 implementation workflows
**Benefit:** Single source of truth, no duplication to maintain
Also removed:
- validation-criteria.yaml (validation now inline in instructions)
- architecture-patterns.yaml references (patterns now managed differently)
- AUDIT-REPORT.md files (stale audit artifacts)
**BMB workflows:**
- Updated checklists for workflow and module creation
- Improved agent architecture documentation
- Minor instruction clarifications
**Core brainstorming workflow:**
- Updated README with usage examples
- Enhanced instructions and template clarity
- Better integration with other modules
**BMM installer:**
- Updated for track-based system
- sprint_artifacts configuration
**Tech Writer agent:**
- Minor configuration update for documentation workflows
Removed 200+ files that should not be in repository:
- Installed agent markdown files (analyst, architect, dev, pm, sm, tea, etc.)
- Complete workflow instruction copies
- Documentation duplicates (quick-start, agents-guide, workflows-*)
- Test architecture knowledge base (22 files, 14,000+ lines)
- Configuration files (config.yaml, team definitions)
These are generated during installation and should not be version controlled.
Removed 21 pre-generated XML bundles:
- BMM agents (analyst, architect, dev, pm, sm, tea, tech-writer, ux-designer)
- BMGD agents (game-architect, game-designer, game-dev, game-scrum-master)
- CIS agents (brainstorming-coach, creative-problem-solver, etc.)
- Team bundles (team-fullstack, team-gamedev, creative-squad)
**Why:** Users should generate fresh bundles via `npm run bundle` to get latest changes and customizations.
- **2 new documentation files** (comprehensive guides)
- **98 source files modified** (299 insertions, 6,567 deletions)
- **3 installer config files simplified** (major folder structure improvement)
- **200+ .bmad/ artifacts removed** (should not be in repo)
- **21 web-bundle files removed** (users regenerate as needed)
- **12 duplicate workflows removed** (BMGD consolidation)
- **40+ workflows updated** (track system, variable standardization, corruption fixes)
|
||
|---|---|---|
| .. | ||
| README.md | ||
| agent-architecture.md | ||
| agent-command-patterns.md | ||
| agent-types.md | ||
| brainstorm-context.md | ||
| checklist.md | ||
| communication-styles.md | ||
| instructions.md | ||
| workflow.yaml | ||
README.md
Create Agent Workflow
Interactive agent builder creating BMad Core compliant agents as YAML source files that compile to .md during installation.
Table of Contents
Quick Start
# Direct workflow
workflow create-agent
# Via BMad Builder
*create-agent
Agent Types
Simple Agent
- Self-contained functionality
- Basic command structure
- No external resources
Expert Agent
- Sidecar resources for domain knowledge
- Extended capabilities
- Knowledge base integration
Module Agent
- Full-featured with workflows
- Module-specific commands
- Integrated with module structure
Workflow Phases
Phase 0: Optional Brainstorming
- Creative ideation session
- Explore concepts and personalities
- Generate command ideas
- Output feeds into persona development
Phase 1: Agent Setup
- Choose agent type (Simple/Expert/Module)
- Define identity (name, title, icon, filename)
- Assign to module (if Module agent)
Phase 2: Persona Development
- Define role and responsibilities
- Craft unique identity/backstory
- Select communication style
- Establish guiding principles
- Add critical actions (optional)
Phase 3: Command Building
- Add required commands (*help, *exit)
- Define workflow commands
- Add task commands
- Create action commands
- Configure attributes
Phase 4: Finalization
- Generate .agent.yaml file
- Create customize file (optional)
- Setup sidecar resources (Expert agents)
- Validate and compile
- Provide usage instructions
Output Structure
Generated Files
Standalone Agents:
- Source:
{bmad_folder}/agents/{filename}.agent.yaml - Compiled:
{bmad_folder}/agents/{filename}.md
Module Agents:
- Source:
src/modules/{module}/agents/{filename}.agent.yaml - Compiled:
{bmad_folder}/{module}/agents/{filename}.md
YAML Structure
agent:
metadata:
id: {bmad_folder}/{module}/agents/{filename}.md
name: Agent Name
title: Agent Title
icon: 🤖
module: module-name
persona:
role: '...'
identity: '...'
communication_style: '...'
principles: ['...', '...']
menu:
- trigger: command-name
workflow: path/to/workflow.yaml
description: Command description
Optional Customize File
Location: {bmad_folder}/_cfg/agents/{module}-{filename}.customize.yaml
Allows persona and menu overrides that persist through updates.
Installation
Compilation Methods
Quick Rebuild:
bmad compile-agents
During Module Install: Automatic compilation when installing modules
Manual Compilation:
node tools/cli/bmad-cli.js compile-agents
Examples
Creating a Code Review Agent
User: I need a code review agent
Builder: Let's brainstorm first...
[Brainstorming generates ideas for strict vs friendly reviewer]
Builder: Now let's build your agent:
- Type: Simple
- Name: Code Reviewer
- Role: Senior developer conducting thorough reviews
- Style: Professional but approachable
- Commands:
- *review-pr: Review pull request
- *review-file: Review single file
- *review-standards: Check coding standards
Creating a Domain Expert
Type: Expert
Name: Legal Advisor
Sidecar: legal-knowledge/
Commands:
- *contract-review
- *compliance-check
- *risk-assessment
Workflow Files
create-agent/
├── workflow.yaml # Configuration
├── instructions.md # Step guide
├── checklist.md # Validation
├── README.md # This file
├── agent-types.md # Type details
├── agent-architecture.md # Patterns
├── agent-command-patterns.md # Commands
└── communication-styles.md # Styles
Best Practices
- Use brainstorming for complex agents
- Start simple - Add commands incrementally
- Test commands before finalizing
- Document thoroughly in descriptions
- Follow naming conventions consistently