2985 lines
116 KiB
JavaScript
Raw Normal View History

Add Text-to-Speech Integration via TTS_INJECTION System (#934) * feat: Add provider-agnostic TTS integration via injection point system Implements comprehensive Text-to-Speech integration for BMAD agents using a generic TTS_INJECTION marker system. When AgentVibes (or any compatible TTS provider) is installed, all BMAD agents can speak their responses with unique AI voices. ## Key Features **Provider-Agnostic Architecture** - Uses generic `TTS_INJECTION` markers instead of vendor-specific naming - Future-proof for multiple TTS providers beyond AgentVibes - Clean separation - BMAD stays TTS-agnostic, providers handle injection **Installation Flow** - BMAD → AgentVibes: TTS instructions injected when AgentVibes detects existing BMAD installation - AgentVibes → BMAD: TTS instructions injected during BMAD installation when AgentVibes detected - User must manually create voice assignment file when AgentVibes installs first (documented limitation) **Party Mode Voice Support** - Each agent speaks with unique assigned voice in multi-agent discussions - PM, Architect, Developer, Analyst, UX Designer, etc. - all with distinct voices **Zero Breaking Changes** - Fully backward compatible - works without any TTS provider - `TTS_INJECTION` markers are benign HTML comments if not processed - No changes to existing agent behavior or non-TTS workflows ## Implementation Details **Files Modified:** - `tools/cli/installers/lib/core/installer.js` - TTS injection processing logic - `tools/cli/lib/ui.js` - AgentVibes detection and installation prompts - `tools/cli/commands/install.js` - Post-install guidance for AgentVibes setup - `src/utility/models/fragments/activation-rules.xml` - TTS_INJECTION marker for agents - `src/core/workflows/party-mode/instructions.md` - TTS_INJECTION marker for party mode **Injection Point System:** ```xml <rules> - ALWAYS communicate in {communication_language} <!-- TTS_INJECTION:agent-tts --> - Stay in character until exit selected </rules> ``` When AgentVibes is detected, the installer replaces this marker with: ``` - When responding to user messages, speak your responses using TTS: Call: `.claude/hooks/bmad-speak.sh '{agent-id}' '{response-text}'` after each response IMPORTANT: Use single quotes - do NOT escape special characters like ! or $ ``` **Special Character Handling:** - Explicit guidance to use single quotes without escaping - Prevents "backslash exclamation" artifacts in speech **User Experience:** ``` User: "How should we architect this feature?" Architect: [Text response] + 🔊 [Professional voice explains architecture] ``` Party Mode: ``` PM (John): "I'll focus on user value..." 🔊 [Male professional voice] UX Designer (Sara): "From a user perspective..." 🔊 [Female voice] Architect (Marcus): "The technical approach..." 🔊 [Male technical voice] ``` ## Testing **Unit Tests:** ✅ 62/62 passing - 49/49 schema validation tests - 13/13 installation component tests **Integration Testing:** - ✅ BMAD → AgentVibes (automatic injection) - ✅ AgentVibes → BMAD (automatic injection) - ✅ No TTS provider (markers remain as comments) ## Documentation Comprehensive testing guide created with: - Both installation scenario walkthroughs - Verification commands and expected outputs - Troubleshooting guidance ## Known Limitations **AgentVibes → BMAD Installation Order:** When AgentVibes installs first, voice assignment file must be created manually: ```bash mkdir -p .bmad/_cfg cat > .bmad/_cfg/agent-voice-map.csv << 'EOF' agent_id,voice_name pm,en_US-ryan-high architect,en_US-danny-low dev,en_US-joe-medium EOF ``` This limitation exists to prevent false legacy v4 detection warnings from BMAD installer. **Recommended:** Install BMAD first, then AgentVibes for automatic voice assignment. ## Related Work **Companion Implementation:** - Repository: paulpreibisch/AgentVibes - Commits: 6 commits implementing injection processing and voice routing - Features: Retroactive injection, file path extraction, escape stripping **GitHub Issues:** - paulpreibisch/AgentVibes#36 - BMAD agent ID support ## Breaking Changes None. Feature is opt-in and requires separate TTS provider installation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Enforce project hooks over global hooks in party mode before, claude would sometimes favor global agent vibes hooks over project specific * feat: Automate AgentVibes installer invocation after BMAD install Instead of showing manual installation instructions, the installer now: - Prompts "Press Enter to start AgentVibes installer..." - Automatically runs npx agentvibes@latest install - Handles errors gracefully with fallback instructions This provides a seamless installation flow matching the test script's interactive approach. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * docs: Add automated testing script and guide for PR #934 Added comprehensive testing tools for AgentVibes party mode integration: - test-bmad-pr.sh: Fully automated installation and verification script - Interactive mode selection (official PR or custom fork) - Automatic BMAD CLI setup and linking - AgentVibes installation with guided prompts - Built-in verification checks for voice maps and hooks - Saved configuration for quick re-testing - TESTING.md: Complete testing documentation - Quick start with one-line npx command - Manual installation alternative - Troubleshooting guide - Cleanup instructions Testers can now run a single command to test the full AgentVibes integration without needing to understand the complex setup process. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Add shell: true to npx execSync to prevent permission denied error The execSync call for 'npx agentvibes@latest install' was failing with 'Permission denied' because the shell was trying to execute 'agentvibes@latest' directly instead of passing it as an argument to npx. Adding shell: true ensures the command runs in a proper shell context where npx can correctly interpret the @latest version syntax. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Remove duplicate AgentVibes installation step from test script The test script was calling AgentVibes installer twice: 1. BMAD installer now automatically runs AgentVibes (new feature) 2. Test script had a separate Step 6 that also ran AgentVibes This caused the installer to run twice, with the second call failing because it was already installed. Changes: - Removed redundant Step 6 (AgentVibes installation) - Updated Step 5 to indicate it includes AgentVibes - Updated step numbers from 7 to 6 throughout - Added guidance that AgentVibes runs automatically Now the flow is cleaner: BMAD installer handles everything! 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Address bmadcode review - preserve variables and move TTS logic to injection Fixes requested changes from PR review: 1. Preserve {bmad_folder} variable placeholder - Changed: {project_root}/.bmad/core/tasks/workflow.xml - To: {project_root}/{bmad_folder}/core/tasks/workflow.xml - Allows users to choose custom BMAD folder names during installation 2. Move TTS-specific hook guidance to injection system - Removed hardcoded hook enforcement from source files - Added hook guidance to processTTSInjectionPoints() in installer.js - Now only appears when AgentVibes is installed (via TTS_INJECTION) 3. Maintain TTS-agnostic source architecture - Source files remain clean of TTS-specific instructions - TTS details injected at install-time only when needed - Preserves provider-agnostic design principle Changes made: - src/core/workflows/party-mode/instructions.md - Reverted .bmad to {bmad_folder} variable - Replaced hardcoded hook guidance with <!-- TTS_INJECTION:party-mode --> - Removed <note> about play-tts.sh hook location - tools/cli/installers/lib/core/installer.js - Added hook enforcement to party-mode injection replacement - Guidance now injected only when enableAgentVibes is true Addresses review comments from bmadcode: - "needs to remain the variable. it will get set in the file at the install destination." - "items like this we will need to inject if user is using claude and TTS" 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Change 'claude-code' to 'claude' in test script instructions The correct command to start Claude is 'claude', not 'claude-code'. Updated line 362-363 in test-bmad-pr.sh to show the correct command. * fix: Remove npm link from test script to avoid global namespace pollution - Removed 'npm link' command that was installing BMAD globally - Changed 'bmad install' to direct node execution using local clone - Updated success message to reflect no global installation This keeps testing fully isolated and prevents conflicts with: - Existing BMAD installations - Future official BMAD installs - Orphaned symlinks when test directory is deleted The test script now runs completely self-contained without modifying the user's global npm environment. --------- Co-authored-by: Claude Code <claude@anthropic.com> Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Brian <bmadcode@gmail.com>
2025-11-26 08:51:57 -07:00
/**
* File: tools/cli/installers/lib/core/installer.js
*
* BMAD Method - Business Model Agile Development Method
* Repository: https://github.com/paulpreibisch/BMAD-METHOD
*
* Copyright (c) 2025 Paul Preibisch
* Licensed under the Apache License, Version 2.0
*
* ---
*
* @fileoverview Core BMAD installation orchestrator with AgentVibes injection point support
* @context Manages complete BMAD installation flow including core agents, modules, IDE configs, and optional TTS integration
* @architecture Orchestrator pattern - coordinates Detector, ModuleManager, IdeManager, and file operations to build complete BMAD installation
* @dependencies fs-extra, ora, chalk, detector.js, module-manager.js, ide-manager.js, config.js
* @entrypoints Called by install.js command via installer.install(config)
* @patterns Injection point processing (AgentVibes), placeholder replacement ({bmad_folder}), module dependency resolution
* @related GitHub AgentVibes#34 (injection points), ui.js (user prompts), copyFileWithPlaceholderReplacement()
*/
const path = require('node:path');
const fs = require('fs-extra');
const chalk = require('chalk');
const ora = require('ora');
const { Detector } = require('./detector');
const { Manifest } = require('./manifest');
const { ModuleManager } = require('../modules/manager');
const { IdeManager } = require('../ide/manager');
const { FileOps } = require('../../../lib/file-ops');
const { Config } = require('../../../lib/config');
const { XmlHandler } = require('../../../lib/xml-handler');
const { DependencyResolver } = require('./dependency-resolver');
const { ConfigCollector } = require('./config-collector');
// processInstallation no longer needed - LLMs understand {project-root}
const { getProjectRoot, getSourcePath, getModulePath } = require('../../../lib/project-root');
const { AgentPartyGenerator } = require('../../../lib/agent-party-generator');
const { CLIUtils } = require('../../../lib/cli-utils');
const { ManifestGenerator } = require('./manifest-generator');
2025-10-26 17:04:27 -05:00
const { IdeConfigManager } = require('./ide-config-manager');
class Installer {
constructor() {
this.detector = new Detector();
this.manifest = new Manifest();
this.moduleManager = new ModuleManager();
this.ideManager = new IdeManager();
this.fileOps = new FileOps();
this.config = new Config();
this.xmlHandler = new XmlHandler();
this.dependencyResolver = new DependencyResolver();
this.configCollector = new ConfigCollector();
2025-10-26 17:04:27 -05:00
this.ideConfigManager = new IdeConfigManager();
this.installedFiles = []; // Track all installed files
feat(installer): Enhanced TTS injection summary with tracking and documentation (#1037) ## Summary - Track all files with TTS injection applied during installation - Display informative summary explaining what TTS injection does - Show backup location and restore command for recovery ## What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ## Changes - **installer.js**: Track files in `processAgentFiles()`, `buildStandaloneAgents()`, and `rebuildAgentFiles()` when TTS markers are processed - **compiler.js**: Add TTS injection support for custom agent compilation - **ui.js**: Enhanced installation summary showing: - Explanation of what TTS injection is with example - List of all files with TTS injection applied (grouped by type) - Backup location (~/.bmad-tts-backups/) - Restore command for recovery ## Example Output ``` ═══════════════════════════════════════════════════ AgentVibes TTS Injection Summary ═══════════════════════════════════════════════════ What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ✅ TTS injection applied to 11 file(s): Party Mode (multi-agent conversations): • .bmad/core/workflows/party-mode/instructions.md Agent TTS (individual agent voices): • .bmad/bmm/agents/analyst.md • .bmad/bmm/agents/architect.md ... Backups & Recovery: Pre-injection backups are stored in: ~/.bmad-tts-backups/ To restore original files (removes TTS instructions): bmad-tts-injector.sh --restore /path/to/.bmad ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Claude <noreply@anthropic.com>
2025-12-05 17:54:03 -07:00
this.ttsInjectedFiles = []; // Track files with TTS injection applied
}
/**
* Find the bmad installation directory in a project
* V6+ installations can use ANY folder name but ALWAYS have _cfg/manifest.yaml
* @param {string} projectDir - Project directory
* @returns {Promise<string>} Path to bmad directory
*/
async findBmadDir(projectDir) {
// Check if project directory exists
if (!(await fs.pathExists(projectDir))) {
// Project doesn't exist yet, return default
return path.join(projectDir, 'bmad');
}
// V6+ strategy: Look for ANY directory with _cfg/manifest.yaml
// This is the definitive marker of a V6+ installation
try {
const entries = await fs.readdir(projectDir, { withFileTypes: true });
for (const entry of entries) {
if (entry.isDirectory()) {
const manifestPath = path.join(projectDir, entry.name, '_cfg', 'manifest.yaml');
if (await fs.pathExists(manifestPath)) {
// Found a V6+ installation
return path.join(projectDir, entry.name);
}
}
}
} catch {
// Ignore errors, fall through to default
}
// No V6+ installation found, return default
// This will be used for new installations
return path.join(projectDir, 'bmad');
}
/**
Add Text-to-Speech Integration via TTS_INJECTION System (#934) * feat: Add provider-agnostic TTS integration via injection point system Implements comprehensive Text-to-Speech integration for BMAD agents using a generic TTS_INJECTION marker system. When AgentVibes (or any compatible TTS provider) is installed, all BMAD agents can speak their responses with unique AI voices. ## Key Features **Provider-Agnostic Architecture** - Uses generic `TTS_INJECTION` markers instead of vendor-specific naming - Future-proof for multiple TTS providers beyond AgentVibes - Clean separation - BMAD stays TTS-agnostic, providers handle injection **Installation Flow** - BMAD → AgentVibes: TTS instructions injected when AgentVibes detects existing BMAD installation - AgentVibes → BMAD: TTS instructions injected during BMAD installation when AgentVibes detected - User must manually create voice assignment file when AgentVibes installs first (documented limitation) **Party Mode Voice Support** - Each agent speaks with unique assigned voice in multi-agent discussions - PM, Architect, Developer, Analyst, UX Designer, etc. - all with distinct voices **Zero Breaking Changes** - Fully backward compatible - works without any TTS provider - `TTS_INJECTION` markers are benign HTML comments if not processed - No changes to existing agent behavior or non-TTS workflows ## Implementation Details **Files Modified:** - `tools/cli/installers/lib/core/installer.js` - TTS injection processing logic - `tools/cli/lib/ui.js` - AgentVibes detection and installation prompts - `tools/cli/commands/install.js` - Post-install guidance for AgentVibes setup - `src/utility/models/fragments/activation-rules.xml` - TTS_INJECTION marker for agents - `src/core/workflows/party-mode/instructions.md` - TTS_INJECTION marker for party mode **Injection Point System:** ```xml <rules> - ALWAYS communicate in {communication_language} <!-- TTS_INJECTION:agent-tts --> - Stay in character until exit selected </rules> ``` When AgentVibes is detected, the installer replaces this marker with: ``` - When responding to user messages, speak your responses using TTS: Call: `.claude/hooks/bmad-speak.sh '{agent-id}' '{response-text}'` after each response IMPORTANT: Use single quotes - do NOT escape special characters like ! or $ ``` **Special Character Handling:** - Explicit guidance to use single quotes without escaping - Prevents "backslash exclamation" artifacts in speech **User Experience:** ``` User: "How should we architect this feature?" Architect: [Text response] + 🔊 [Professional voice explains architecture] ``` Party Mode: ``` PM (John): "I'll focus on user value..." 🔊 [Male professional voice] UX Designer (Sara): "From a user perspective..." 🔊 [Female voice] Architect (Marcus): "The technical approach..." 🔊 [Male technical voice] ``` ## Testing **Unit Tests:** ✅ 62/62 passing - 49/49 schema validation tests - 13/13 installation component tests **Integration Testing:** - ✅ BMAD → AgentVibes (automatic injection) - ✅ AgentVibes → BMAD (automatic injection) - ✅ No TTS provider (markers remain as comments) ## Documentation Comprehensive testing guide created with: - Both installation scenario walkthroughs - Verification commands and expected outputs - Troubleshooting guidance ## Known Limitations **AgentVibes → BMAD Installation Order:** When AgentVibes installs first, voice assignment file must be created manually: ```bash mkdir -p .bmad/_cfg cat > .bmad/_cfg/agent-voice-map.csv << 'EOF' agent_id,voice_name pm,en_US-ryan-high architect,en_US-danny-low dev,en_US-joe-medium EOF ``` This limitation exists to prevent false legacy v4 detection warnings from BMAD installer. **Recommended:** Install BMAD first, then AgentVibes for automatic voice assignment. ## Related Work **Companion Implementation:** - Repository: paulpreibisch/AgentVibes - Commits: 6 commits implementing injection processing and voice routing - Features: Retroactive injection, file path extraction, escape stripping **GitHub Issues:** - paulpreibisch/AgentVibes#36 - BMAD agent ID support ## Breaking Changes None. Feature is opt-in and requires separate TTS provider installation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Enforce project hooks over global hooks in party mode before, claude would sometimes favor global agent vibes hooks over project specific * feat: Automate AgentVibes installer invocation after BMAD install Instead of showing manual installation instructions, the installer now: - Prompts "Press Enter to start AgentVibes installer..." - Automatically runs npx agentvibes@latest install - Handles errors gracefully with fallback instructions This provides a seamless installation flow matching the test script's interactive approach. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * docs: Add automated testing script and guide for PR #934 Added comprehensive testing tools for AgentVibes party mode integration: - test-bmad-pr.sh: Fully automated installation and verification script - Interactive mode selection (official PR or custom fork) - Automatic BMAD CLI setup and linking - AgentVibes installation with guided prompts - Built-in verification checks for voice maps and hooks - Saved configuration for quick re-testing - TESTING.md: Complete testing documentation - Quick start with one-line npx command - Manual installation alternative - Troubleshooting guide - Cleanup instructions Testers can now run a single command to test the full AgentVibes integration without needing to understand the complex setup process. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Add shell: true to npx execSync to prevent permission denied error The execSync call for 'npx agentvibes@latest install' was failing with 'Permission denied' because the shell was trying to execute 'agentvibes@latest' directly instead of passing it as an argument to npx. Adding shell: true ensures the command runs in a proper shell context where npx can correctly interpret the @latest version syntax. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Remove duplicate AgentVibes installation step from test script The test script was calling AgentVibes installer twice: 1. BMAD installer now automatically runs AgentVibes (new feature) 2. Test script had a separate Step 6 that also ran AgentVibes This caused the installer to run twice, with the second call failing because it was already installed. Changes: - Removed redundant Step 6 (AgentVibes installation) - Updated Step 5 to indicate it includes AgentVibes - Updated step numbers from 7 to 6 throughout - Added guidance that AgentVibes runs automatically Now the flow is cleaner: BMAD installer handles everything! 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Address bmadcode review - preserve variables and move TTS logic to injection Fixes requested changes from PR review: 1. Preserve {bmad_folder} variable placeholder - Changed: {project_root}/.bmad/core/tasks/workflow.xml - To: {project_root}/{bmad_folder}/core/tasks/workflow.xml - Allows users to choose custom BMAD folder names during installation 2. Move TTS-specific hook guidance to injection system - Removed hardcoded hook enforcement from source files - Added hook guidance to processTTSInjectionPoints() in installer.js - Now only appears when AgentVibes is installed (via TTS_INJECTION) 3. Maintain TTS-agnostic source architecture - Source files remain clean of TTS-specific instructions - TTS details injected at install-time only when needed - Preserves provider-agnostic design principle Changes made: - src/core/workflows/party-mode/instructions.md - Reverted .bmad to {bmad_folder} variable - Replaced hardcoded hook guidance with <!-- TTS_INJECTION:party-mode --> - Removed <note> about play-tts.sh hook location - tools/cli/installers/lib/core/installer.js - Added hook enforcement to party-mode injection replacement - Guidance now injected only when enableAgentVibes is true Addresses review comments from bmadcode: - "needs to remain the variable. it will get set in the file at the install destination." - "items like this we will need to inject if user is using claude and TTS" 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Change 'claude-code' to 'claude' in test script instructions The correct command to start Claude is 'claude', not 'claude-code'. Updated line 362-363 in test-bmad-pr.sh to show the correct command. * fix: Remove npm link from test script to avoid global namespace pollution - Removed 'npm link' command that was installing BMAD globally - Changed 'bmad install' to direct node execution using local clone - Updated success message to reflect no global installation This keeps testing fully isolated and prevents conflicts with: - Existing BMAD installations - Future official BMAD installs - Orphaned symlinks when test directory is deleted The test script now runs completely self-contained without modifying the user's global npm environment. --------- Co-authored-by: Claude Code <claude@anthropic.com> Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Brian <bmadcode@gmail.com>
2025-11-26 08:51:57 -07:00
* @function copyFileWithPlaceholderReplacement
* @intent Copy files from BMAD source to installation directory with dynamic content transformation
* @why Enables installation-time customization: {bmad_folder} replacement + optional AgentVibes TTS injection
* @param {string} sourcePath - Absolute path to source file in BMAD repository
* @param {string} targetPath - Absolute path to destination file in user's project
* @param {string} bmadFolderName - User's chosen bmad folder name (default: 'bmad')
* @returns {Promise<void>} Resolves when file copy and transformation complete
* @sideeffects Writes transformed file to targetPath, creates parent directories if needed
* @edgecases Binary files bypass transformation, falls back to raw copy if UTF-8 read fails
* @calledby installCore(), installModule(), IDE installers during file vendoring
* @calls processTTSInjectionPoints(), fs.readFile(), fs.writeFile(), fs.copy()
*
* AI NOTE: This is the core transformation pipeline for ALL BMAD installation file copies.
* It performs two transformations in sequence:
* 1. {bmad_folder} user's custom folder name (e.g., ".bmad" or "bmad")
* 2. <!-- TTS_INJECTION:* --> TTS bash calls (if enabled) OR stripped (if disabled)
*
* The injection point processing enables loose coupling between BMAD and TTS providers:
* - BMAD source contains injection markers (not actual TTS code)
* - At install-time, markers are replaced OR removed based on user preference
* - Result: Clean installs for users without TTS, working TTS for users with it
*
* PATTERN: Adding New Injection Points
* =====================================
* 1. Add HTML comment marker in BMAD source file:
* <!-- TTS_INJECTION:feature-name -->
*
* 2. Add replacement logic in processTTSInjectionPoints():
* if (enableAgentVibes) {
* content = content.replace(/<!-- TTS_INJECTION:feature-name -->/g, 'actual code');
* } else {
* content = content.replace(/<!-- TTS_INJECTION:feature-name -->\n?/g, '');
* }
*
* 3. Document marker in instructions.md (if applicable)
*/
async copyFileWithPlaceholderReplacement(sourcePath, targetPath, bmadFolderName) {
// List of text file extensions that should have placeholder replacement
const textExtensions = ['.md', '.yaml', '.yml', '.txt', '.json', '.js', '.ts', '.html', '.css', '.sh', '.bat', '.csv'];
const ext = path.extname(sourcePath).toLowerCase();
// Check if this is a text file that might contain placeholders
if (textExtensions.includes(ext)) {
try {
// Read the file content
let content = await fs.readFile(sourcePath, 'utf8');
// Replace {bmad_folder} placeholder with actual folder name
if (content.includes('{bmad_folder}')) {
content = content.replaceAll('{bmad_folder}', bmadFolderName);
}
// Replace escape sequence {*bmad_folder*} with literal {bmad_folder}
if (content.includes('{*bmad_folder*}')) {
content = content.replaceAll('{*bmad_folder*}', '{bmad_folder}');
}
feat(installer): Enhanced TTS injection summary with tracking and documentation (#1037) ## Summary - Track all files with TTS injection applied during installation - Display informative summary explaining what TTS injection does - Show backup location and restore command for recovery ## What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ## Changes - **installer.js**: Track files in `processAgentFiles()`, `buildStandaloneAgents()`, and `rebuildAgentFiles()` when TTS markers are processed - **compiler.js**: Add TTS injection support for custom agent compilation - **ui.js**: Enhanced installation summary showing: - Explanation of what TTS injection is with example - List of all files with TTS injection applied (grouped by type) - Backup location (~/.bmad-tts-backups/) - Restore command for recovery ## Example Output ``` ═══════════════════════════════════════════════════ AgentVibes TTS Injection Summary ═══════════════════════════════════════════════════ What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ✅ TTS injection applied to 11 file(s): Party Mode (multi-agent conversations): • .bmad/core/workflows/party-mode/instructions.md Agent TTS (individual agent voices): • .bmad/bmm/agents/analyst.md • .bmad/bmm/agents/architect.md ... Backups & Recovery: Pre-injection backups are stored in: ~/.bmad-tts-backups/ To restore original files (removes TTS instructions): bmad-tts-injector.sh --restore /path/to/.bmad ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Claude <noreply@anthropic.com>
2025-12-05 17:54:03 -07:00
// Process AgentVibes injection points (pass targetPath for tracking)
content = this.processTTSInjectionPoints(content, targetPath);
Add Text-to-Speech Integration via TTS_INJECTION System (#934) * feat: Add provider-agnostic TTS integration via injection point system Implements comprehensive Text-to-Speech integration for BMAD agents using a generic TTS_INJECTION marker system. When AgentVibes (or any compatible TTS provider) is installed, all BMAD agents can speak their responses with unique AI voices. ## Key Features **Provider-Agnostic Architecture** - Uses generic `TTS_INJECTION` markers instead of vendor-specific naming - Future-proof for multiple TTS providers beyond AgentVibes - Clean separation - BMAD stays TTS-agnostic, providers handle injection **Installation Flow** - BMAD → AgentVibes: TTS instructions injected when AgentVibes detects existing BMAD installation - AgentVibes → BMAD: TTS instructions injected during BMAD installation when AgentVibes detected - User must manually create voice assignment file when AgentVibes installs first (documented limitation) **Party Mode Voice Support** - Each agent speaks with unique assigned voice in multi-agent discussions - PM, Architect, Developer, Analyst, UX Designer, etc. - all with distinct voices **Zero Breaking Changes** - Fully backward compatible - works without any TTS provider - `TTS_INJECTION` markers are benign HTML comments if not processed - No changes to existing agent behavior or non-TTS workflows ## Implementation Details **Files Modified:** - `tools/cli/installers/lib/core/installer.js` - TTS injection processing logic - `tools/cli/lib/ui.js` - AgentVibes detection and installation prompts - `tools/cli/commands/install.js` - Post-install guidance for AgentVibes setup - `src/utility/models/fragments/activation-rules.xml` - TTS_INJECTION marker for agents - `src/core/workflows/party-mode/instructions.md` - TTS_INJECTION marker for party mode **Injection Point System:** ```xml <rules> - ALWAYS communicate in {communication_language} <!-- TTS_INJECTION:agent-tts --> - Stay in character until exit selected </rules> ``` When AgentVibes is detected, the installer replaces this marker with: ``` - When responding to user messages, speak your responses using TTS: Call: `.claude/hooks/bmad-speak.sh '{agent-id}' '{response-text}'` after each response IMPORTANT: Use single quotes - do NOT escape special characters like ! or $ ``` **Special Character Handling:** - Explicit guidance to use single quotes without escaping - Prevents "backslash exclamation" artifacts in speech **User Experience:** ``` User: "How should we architect this feature?" Architect: [Text response] + 🔊 [Professional voice explains architecture] ``` Party Mode: ``` PM (John): "I'll focus on user value..." 🔊 [Male professional voice] UX Designer (Sara): "From a user perspective..." 🔊 [Female voice] Architect (Marcus): "The technical approach..." 🔊 [Male technical voice] ``` ## Testing **Unit Tests:** ✅ 62/62 passing - 49/49 schema validation tests - 13/13 installation component tests **Integration Testing:** - ✅ BMAD → AgentVibes (automatic injection) - ✅ AgentVibes → BMAD (automatic injection) - ✅ No TTS provider (markers remain as comments) ## Documentation Comprehensive testing guide created with: - Both installation scenario walkthroughs - Verification commands and expected outputs - Troubleshooting guidance ## Known Limitations **AgentVibes → BMAD Installation Order:** When AgentVibes installs first, voice assignment file must be created manually: ```bash mkdir -p .bmad/_cfg cat > .bmad/_cfg/agent-voice-map.csv << 'EOF' agent_id,voice_name pm,en_US-ryan-high architect,en_US-danny-low dev,en_US-joe-medium EOF ``` This limitation exists to prevent false legacy v4 detection warnings from BMAD installer. **Recommended:** Install BMAD first, then AgentVibes for automatic voice assignment. ## Related Work **Companion Implementation:** - Repository: paulpreibisch/AgentVibes - Commits: 6 commits implementing injection processing and voice routing - Features: Retroactive injection, file path extraction, escape stripping **GitHub Issues:** - paulpreibisch/AgentVibes#36 - BMAD agent ID support ## Breaking Changes None. Feature is opt-in and requires separate TTS provider installation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Enforce project hooks over global hooks in party mode before, claude would sometimes favor global agent vibes hooks over project specific * feat: Automate AgentVibes installer invocation after BMAD install Instead of showing manual installation instructions, the installer now: - Prompts "Press Enter to start AgentVibes installer..." - Automatically runs npx agentvibes@latest install - Handles errors gracefully with fallback instructions This provides a seamless installation flow matching the test script's interactive approach. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * docs: Add automated testing script and guide for PR #934 Added comprehensive testing tools for AgentVibes party mode integration: - test-bmad-pr.sh: Fully automated installation and verification script - Interactive mode selection (official PR or custom fork) - Automatic BMAD CLI setup and linking - AgentVibes installation with guided prompts - Built-in verification checks for voice maps and hooks - Saved configuration for quick re-testing - TESTING.md: Complete testing documentation - Quick start with one-line npx command - Manual installation alternative - Troubleshooting guide - Cleanup instructions Testers can now run a single command to test the full AgentVibes integration without needing to understand the complex setup process. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Add shell: true to npx execSync to prevent permission denied error The execSync call for 'npx agentvibes@latest install' was failing with 'Permission denied' because the shell was trying to execute 'agentvibes@latest' directly instead of passing it as an argument to npx. Adding shell: true ensures the command runs in a proper shell context where npx can correctly interpret the @latest version syntax. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Remove duplicate AgentVibes installation step from test script The test script was calling AgentVibes installer twice: 1. BMAD installer now automatically runs AgentVibes (new feature) 2. Test script had a separate Step 6 that also ran AgentVibes This caused the installer to run twice, with the second call failing because it was already installed. Changes: - Removed redundant Step 6 (AgentVibes installation) - Updated Step 5 to indicate it includes AgentVibes - Updated step numbers from 7 to 6 throughout - Added guidance that AgentVibes runs automatically Now the flow is cleaner: BMAD installer handles everything! 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Address bmadcode review - preserve variables and move TTS logic to injection Fixes requested changes from PR review: 1. Preserve {bmad_folder} variable placeholder - Changed: {project_root}/.bmad/core/tasks/workflow.xml - To: {project_root}/{bmad_folder}/core/tasks/workflow.xml - Allows users to choose custom BMAD folder names during installation 2. Move TTS-specific hook guidance to injection system - Removed hardcoded hook enforcement from source files - Added hook guidance to processTTSInjectionPoints() in installer.js - Now only appears when AgentVibes is installed (via TTS_INJECTION) 3. Maintain TTS-agnostic source architecture - Source files remain clean of TTS-specific instructions - TTS details injected at install-time only when needed - Preserves provider-agnostic design principle Changes made: - src/core/workflows/party-mode/instructions.md - Reverted .bmad to {bmad_folder} variable - Replaced hardcoded hook guidance with <!-- TTS_INJECTION:party-mode --> - Removed <note> about play-tts.sh hook location - tools/cli/installers/lib/core/installer.js - Added hook enforcement to party-mode injection replacement - Guidance now injected only when enableAgentVibes is true Addresses review comments from bmadcode: - "needs to remain the variable. it will get set in the file at the install destination." - "items like this we will need to inject if user is using claude and TTS" 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Change 'claude-code' to 'claude' in test script instructions The correct command to start Claude is 'claude', not 'claude-code'. Updated line 362-363 in test-bmad-pr.sh to show the correct command. * fix: Remove npm link from test script to avoid global namespace pollution - Removed 'npm link' command that was installing BMAD globally - Changed 'bmad install' to direct node execution using local clone - Updated success message to reflect no global installation This keeps testing fully isolated and prevents conflicts with: - Existing BMAD installations - Future official BMAD installs - Orphaned symlinks when test directory is deleted The test script now runs completely self-contained without modifying the user's global npm environment. --------- Co-authored-by: Claude Code <claude@anthropic.com> Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Brian <bmadcode@gmail.com>
2025-11-26 08:51:57 -07:00
// Write to target with replaced content
await fs.ensureDir(path.dirname(targetPath));
await fs.writeFile(targetPath, content, 'utf8');
} catch {
// If reading as text fails (might be binary despite extension), fall back to regular copy
await fs.copy(sourcePath, targetPath, { overwrite: true });
}
} else {
// Binary file or other file type - just copy directly
await fs.copy(sourcePath, targetPath, { overwrite: true });
}
}
Add Text-to-Speech Integration via TTS_INJECTION System (#934) * feat: Add provider-agnostic TTS integration via injection point system Implements comprehensive Text-to-Speech integration for BMAD agents using a generic TTS_INJECTION marker system. When AgentVibes (or any compatible TTS provider) is installed, all BMAD agents can speak their responses with unique AI voices. ## Key Features **Provider-Agnostic Architecture** - Uses generic `TTS_INJECTION` markers instead of vendor-specific naming - Future-proof for multiple TTS providers beyond AgentVibes - Clean separation - BMAD stays TTS-agnostic, providers handle injection **Installation Flow** - BMAD → AgentVibes: TTS instructions injected when AgentVibes detects existing BMAD installation - AgentVibes → BMAD: TTS instructions injected during BMAD installation when AgentVibes detected - User must manually create voice assignment file when AgentVibes installs first (documented limitation) **Party Mode Voice Support** - Each agent speaks with unique assigned voice in multi-agent discussions - PM, Architect, Developer, Analyst, UX Designer, etc. - all with distinct voices **Zero Breaking Changes** - Fully backward compatible - works without any TTS provider - `TTS_INJECTION` markers are benign HTML comments if not processed - No changes to existing agent behavior or non-TTS workflows ## Implementation Details **Files Modified:** - `tools/cli/installers/lib/core/installer.js` - TTS injection processing logic - `tools/cli/lib/ui.js` - AgentVibes detection and installation prompts - `tools/cli/commands/install.js` - Post-install guidance for AgentVibes setup - `src/utility/models/fragments/activation-rules.xml` - TTS_INJECTION marker for agents - `src/core/workflows/party-mode/instructions.md` - TTS_INJECTION marker for party mode **Injection Point System:** ```xml <rules> - ALWAYS communicate in {communication_language} <!-- TTS_INJECTION:agent-tts --> - Stay in character until exit selected </rules> ``` When AgentVibes is detected, the installer replaces this marker with: ``` - When responding to user messages, speak your responses using TTS: Call: `.claude/hooks/bmad-speak.sh '{agent-id}' '{response-text}'` after each response IMPORTANT: Use single quotes - do NOT escape special characters like ! or $ ``` **Special Character Handling:** - Explicit guidance to use single quotes without escaping - Prevents "backslash exclamation" artifacts in speech **User Experience:** ``` User: "How should we architect this feature?" Architect: [Text response] + 🔊 [Professional voice explains architecture] ``` Party Mode: ``` PM (John): "I'll focus on user value..." 🔊 [Male professional voice] UX Designer (Sara): "From a user perspective..." 🔊 [Female voice] Architect (Marcus): "The technical approach..." 🔊 [Male technical voice] ``` ## Testing **Unit Tests:** ✅ 62/62 passing - 49/49 schema validation tests - 13/13 installation component tests **Integration Testing:** - ✅ BMAD → AgentVibes (automatic injection) - ✅ AgentVibes → BMAD (automatic injection) - ✅ No TTS provider (markers remain as comments) ## Documentation Comprehensive testing guide created with: - Both installation scenario walkthroughs - Verification commands and expected outputs - Troubleshooting guidance ## Known Limitations **AgentVibes → BMAD Installation Order:** When AgentVibes installs first, voice assignment file must be created manually: ```bash mkdir -p .bmad/_cfg cat > .bmad/_cfg/agent-voice-map.csv << 'EOF' agent_id,voice_name pm,en_US-ryan-high architect,en_US-danny-low dev,en_US-joe-medium EOF ``` This limitation exists to prevent false legacy v4 detection warnings from BMAD installer. **Recommended:** Install BMAD first, then AgentVibes for automatic voice assignment. ## Related Work **Companion Implementation:** - Repository: paulpreibisch/AgentVibes - Commits: 6 commits implementing injection processing and voice routing - Features: Retroactive injection, file path extraction, escape stripping **GitHub Issues:** - paulpreibisch/AgentVibes#36 - BMAD agent ID support ## Breaking Changes None. Feature is opt-in and requires separate TTS provider installation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Enforce project hooks over global hooks in party mode before, claude would sometimes favor global agent vibes hooks over project specific * feat: Automate AgentVibes installer invocation after BMAD install Instead of showing manual installation instructions, the installer now: - Prompts "Press Enter to start AgentVibes installer..." - Automatically runs npx agentvibes@latest install - Handles errors gracefully with fallback instructions This provides a seamless installation flow matching the test script's interactive approach. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * docs: Add automated testing script and guide for PR #934 Added comprehensive testing tools for AgentVibes party mode integration: - test-bmad-pr.sh: Fully automated installation and verification script - Interactive mode selection (official PR or custom fork) - Automatic BMAD CLI setup and linking - AgentVibes installation with guided prompts - Built-in verification checks for voice maps and hooks - Saved configuration for quick re-testing - TESTING.md: Complete testing documentation - Quick start with one-line npx command - Manual installation alternative - Troubleshooting guide - Cleanup instructions Testers can now run a single command to test the full AgentVibes integration without needing to understand the complex setup process. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Add shell: true to npx execSync to prevent permission denied error The execSync call for 'npx agentvibes@latest install' was failing with 'Permission denied' because the shell was trying to execute 'agentvibes@latest' directly instead of passing it as an argument to npx. Adding shell: true ensures the command runs in a proper shell context where npx can correctly interpret the @latest version syntax. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Remove duplicate AgentVibes installation step from test script The test script was calling AgentVibes installer twice: 1. BMAD installer now automatically runs AgentVibes (new feature) 2. Test script had a separate Step 6 that also ran AgentVibes This caused the installer to run twice, with the second call failing because it was already installed. Changes: - Removed redundant Step 6 (AgentVibes installation) - Updated Step 5 to indicate it includes AgentVibes - Updated step numbers from 7 to 6 throughout - Added guidance that AgentVibes runs automatically Now the flow is cleaner: BMAD installer handles everything! 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Address bmadcode review - preserve variables and move TTS logic to injection Fixes requested changes from PR review: 1. Preserve {bmad_folder} variable placeholder - Changed: {project_root}/.bmad/core/tasks/workflow.xml - To: {project_root}/{bmad_folder}/core/tasks/workflow.xml - Allows users to choose custom BMAD folder names during installation 2. Move TTS-specific hook guidance to injection system - Removed hardcoded hook enforcement from source files - Added hook guidance to processTTSInjectionPoints() in installer.js - Now only appears when AgentVibes is installed (via TTS_INJECTION) 3. Maintain TTS-agnostic source architecture - Source files remain clean of TTS-specific instructions - TTS details injected at install-time only when needed - Preserves provider-agnostic design principle Changes made: - src/core/workflows/party-mode/instructions.md - Reverted .bmad to {bmad_folder} variable - Replaced hardcoded hook guidance with <!-- TTS_INJECTION:party-mode --> - Removed <note> about play-tts.sh hook location - tools/cli/installers/lib/core/installer.js - Added hook enforcement to party-mode injection replacement - Guidance now injected only when enableAgentVibes is true Addresses review comments from bmadcode: - "needs to remain the variable. it will get set in the file at the install destination." - "items like this we will need to inject if user is using claude and TTS" 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Change 'claude-code' to 'claude' in test script instructions The correct command to start Claude is 'claude', not 'claude-code'. Updated line 362-363 in test-bmad-pr.sh to show the correct command. * fix: Remove npm link from test script to avoid global namespace pollution - Removed 'npm link' command that was installing BMAD globally - Changed 'bmad install' to direct node execution using local clone - Updated success message to reflect no global installation This keeps testing fully isolated and prevents conflicts with: - Existing BMAD installations - Future official BMAD installs - Orphaned symlinks when test directory is deleted The test script now runs completely self-contained without modifying the user's global npm environment. --------- Co-authored-by: Claude Code <claude@anthropic.com> Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Brian <bmadcode@gmail.com>
2025-11-26 08:51:57 -07:00
/**
* @function processTTSInjectionPoints
* @intent Transform TTS injection markers based on user's installation choice
* @why Enables optional TTS integration without tight coupling between BMAD and TTS providers
* @param {string} content - Raw file content containing potential injection markers
* @returns {string} Transformed content with markers replaced (if enabled) or stripped (if disabled)
* @sideeffects None - pure transformation function
* @edgecases Returns content unchanged if no markers present, safe to call on all files
* @calledby copyFileWithPlaceholderReplacement() during every file copy operation
* @calls String.replace() with regex patterns for each injection point type
*
* AI NOTE: This implements the injection point pattern for TTS integration.
* Key architectural decisions:
*
* 1. **Why Injection Points vs Direct Integration?**
* - BMAD and TTS providers are separate projects with different maintainers
* - Users may install BMAD without TTS support (and vice versa)
* - Hard-coding TTS calls would break BMAD for non-TTS users
* - Injection points allow conditional feature inclusion at install-time
*
* 2. **How It Works:**
* - BMAD source contains markers: <!-- TTS_INJECTION:feature-name -->
* - During installation, user is prompted: "Enable AgentVibes TTS?"
* - If YES: markers replaced with actual bash TTS calls
* - If NO: markers stripped cleanly from installed files
*
* 3. **State Management:**
* - this.enableAgentVibes set in install() method from config.enableAgentVibes
* - config.enableAgentVibes comes from ui.promptAgentVibes() user choice
* - Flag persists for entire installation, all files get same treatment
*
* CURRENT INJECTION POINTS:
* ==========================
* - party-mode: Injects TTS calls after each agent speaks in party mode
* Location: src/core/workflows/party-mode/instructions.md
* Marker: <!-- TTS_INJECTION:party-mode -->
* Replacement: Bash call to .claude/hooks/bmad-speak.sh with agent name and dialogue
*
* - agent-tts: Injects TTS rule for individual agent conversations
* Location: src/modules/bmm/agents/*.md (all agent files)
* Marker: <!-- TTS_INJECTION:agent-tts -->
* Replacement: Rule instructing agent to call bmad-speak.sh with agent ID and response
*
* ADDING NEW INJECTION POINTS:
* =============================
* 1. Add new case in this function:
* content = content.replace(
* /<!-- TTS_INJECTION:new-feature -->/g,
* `code to inject when enabled`
* );
*
* 2. Add marker to BMAD source file at injection location
*
* 3. Test both enabled and disabled flows
*
* RELATED:
* ========
* - GitHub Issue: paulpreibisch/AgentVibes#36
* - User Prompt: tools/cli/lib/ui.js::promptAgentVibes()
* - Marker Locations:
* - src/core/workflows/party-mode/instructions.md:101
* - src/modules/bmm/agents/*.md (rules sections)
* - TTS Hook: .claude/hooks/bmad-speak.sh (in AgentVibes repo)
*/
feat(installer): Enhanced TTS injection summary with tracking and documentation (#1037) ## Summary - Track all files with TTS injection applied during installation - Display informative summary explaining what TTS injection does - Show backup location and restore command for recovery ## What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ## Changes - **installer.js**: Track files in `processAgentFiles()`, `buildStandaloneAgents()`, and `rebuildAgentFiles()` when TTS markers are processed - **compiler.js**: Add TTS injection support for custom agent compilation - **ui.js**: Enhanced installation summary showing: - Explanation of what TTS injection is with example - List of all files with TTS injection applied (grouped by type) - Backup location (~/.bmad-tts-backups/) - Restore command for recovery ## Example Output ``` ═══════════════════════════════════════════════════ AgentVibes TTS Injection Summary ═══════════════════════════════════════════════════ What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ✅ TTS injection applied to 11 file(s): Party Mode (multi-agent conversations): • .bmad/core/workflows/party-mode/instructions.md Agent TTS (individual agent voices): • .bmad/bmm/agents/analyst.md • .bmad/bmm/agents/architect.md ... Backups & Recovery: Pre-injection backups are stored in: ~/.bmad-tts-backups/ To restore original files (removes TTS instructions): bmad-tts-injector.sh --restore /path/to/.bmad ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Claude <noreply@anthropic.com>
2025-12-05 17:54:03 -07:00
processTTSInjectionPoints(content, targetPath = null) {
Add Text-to-Speech Integration via TTS_INJECTION System (#934) * feat: Add provider-agnostic TTS integration via injection point system Implements comprehensive Text-to-Speech integration for BMAD agents using a generic TTS_INJECTION marker system. When AgentVibes (or any compatible TTS provider) is installed, all BMAD agents can speak their responses with unique AI voices. ## Key Features **Provider-Agnostic Architecture** - Uses generic `TTS_INJECTION` markers instead of vendor-specific naming - Future-proof for multiple TTS providers beyond AgentVibes - Clean separation - BMAD stays TTS-agnostic, providers handle injection **Installation Flow** - BMAD → AgentVibes: TTS instructions injected when AgentVibes detects existing BMAD installation - AgentVibes → BMAD: TTS instructions injected during BMAD installation when AgentVibes detected - User must manually create voice assignment file when AgentVibes installs first (documented limitation) **Party Mode Voice Support** - Each agent speaks with unique assigned voice in multi-agent discussions - PM, Architect, Developer, Analyst, UX Designer, etc. - all with distinct voices **Zero Breaking Changes** - Fully backward compatible - works without any TTS provider - `TTS_INJECTION` markers are benign HTML comments if not processed - No changes to existing agent behavior or non-TTS workflows ## Implementation Details **Files Modified:** - `tools/cli/installers/lib/core/installer.js` - TTS injection processing logic - `tools/cli/lib/ui.js` - AgentVibes detection and installation prompts - `tools/cli/commands/install.js` - Post-install guidance for AgentVibes setup - `src/utility/models/fragments/activation-rules.xml` - TTS_INJECTION marker for agents - `src/core/workflows/party-mode/instructions.md` - TTS_INJECTION marker for party mode **Injection Point System:** ```xml <rules> - ALWAYS communicate in {communication_language} <!-- TTS_INJECTION:agent-tts --> - Stay in character until exit selected </rules> ``` When AgentVibes is detected, the installer replaces this marker with: ``` - When responding to user messages, speak your responses using TTS: Call: `.claude/hooks/bmad-speak.sh '{agent-id}' '{response-text}'` after each response IMPORTANT: Use single quotes - do NOT escape special characters like ! or $ ``` **Special Character Handling:** - Explicit guidance to use single quotes without escaping - Prevents "backslash exclamation" artifacts in speech **User Experience:** ``` User: "How should we architect this feature?" Architect: [Text response] + 🔊 [Professional voice explains architecture] ``` Party Mode: ``` PM (John): "I'll focus on user value..." 🔊 [Male professional voice] UX Designer (Sara): "From a user perspective..." 🔊 [Female voice] Architect (Marcus): "The technical approach..." 🔊 [Male technical voice] ``` ## Testing **Unit Tests:** ✅ 62/62 passing - 49/49 schema validation tests - 13/13 installation component tests **Integration Testing:** - ✅ BMAD → AgentVibes (automatic injection) - ✅ AgentVibes → BMAD (automatic injection) - ✅ No TTS provider (markers remain as comments) ## Documentation Comprehensive testing guide created with: - Both installation scenario walkthroughs - Verification commands and expected outputs - Troubleshooting guidance ## Known Limitations **AgentVibes → BMAD Installation Order:** When AgentVibes installs first, voice assignment file must be created manually: ```bash mkdir -p .bmad/_cfg cat > .bmad/_cfg/agent-voice-map.csv << 'EOF' agent_id,voice_name pm,en_US-ryan-high architect,en_US-danny-low dev,en_US-joe-medium EOF ``` This limitation exists to prevent false legacy v4 detection warnings from BMAD installer. **Recommended:** Install BMAD first, then AgentVibes for automatic voice assignment. ## Related Work **Companion Implementation:** - Repository: paulpreibisch/AgentVibes - Commits: 6 commits implementing injection processing and voice routing - Features: Retroactive injection, file path extraction, escape stripping **GitHub Issues:** - paulpreibisch/AgentVibes#36 - BMAD agent ID support ## Breaking Changes None. Feature is opt-in and requires separate TTS provider installation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Enforce project hooks over global hooks in party mode before, claude would sometimes favor global agent vibes hooks over project specific * feat: Automate AgentVibes installer invocation after BMAD install Instead of showing manual installation instructions, the installer now: - Prompts "Press Enter to start AgentVibes installer..." - Automatically runs npx agentvibes@latest install - Handles errors gracefully with fallback instructions This provides a seamless installation flow matching the test script's interactive approach. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * docs: Add automated testing script and guide for PR #934 Added comprehensive testing tools for AgentVibes party mode integration: - test-bmad-pr.sh: Fully automated installation and verification script - Interactive mode selection (official PR or custom fork) - Automatic BMAD CLI setup and linking - AgentVibes installation with guided prompts - Built-in verification checks for voice maps and hooks - Saved configuration for quick re-testing - TESTING.md: Complete testing documentation - Quick start with one-line npx command - Manual installation alternative - Troubleshooting guide - Cleanup instructions Testers can now run a single command to test the full AgentVibes integration without needing to understand the complex setup process. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Add shell: true to npx execSync to prevent permission denied error The execSync call for 'npx agentvibes@latest install' was failing with 'Permission denied' because the shell was trying to execute 'agentvibes@latest' directly instead of passing it as an argument to npx. Adding shell: true ensures the command runs in a proper shell context where npx can correctly interpret the @latest version syntax. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Remove duplicate AgentVibes installation step from test script The test script was calling AgentVibes installer twice: 1. BMAD installer now automatically runs AgentVibes (new feature) 2. Test script had a separate Step 6 that also ran AgentVibes This caused the installer to run twice, with the second call failing because it was already installed. Changes: - Removed redundant Step 6 (AgentVibes installation) - Updated Step 5 to indicate it includes AgentVibes - Updated step numbers from 7 to 6 throughout - Added guidance that AgentVibes runs automatically Now the flow is cleaner: BMAD installer handles everything! 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Address bmadcode review - preserve variables and move TTS logic to injection Fixes requested changes from PR review: 1. Preserve {bmad_folder} variable placeholder - Changed: {project_root}/.bmad/core/tasks/workflow.xml - To: {project_root}/{bmad_folder}/core/tasks/workflow.xml - Allows users to choose custom BMAD folder names during installation 2. Move TTS-specific hook guidance to injection system - Removed hardcoded hook enforcement from source files - Added hook guidance to processTTSInjectionPoints() in installer.js - Now only appears when AgentVibes is installed (via TTS_INJECTION) 3. Maintain TTS-agnostic source architecture - Source files remain clean of TTS-specific instructions - TTS details injected at install-time only when needed - Preserves provider-agnostic design principle Changes made: - src/core/workflows/party-mode/instructions.md - Reverted .bmad to {bmad_folder} variable - Replaced hardcoded hook guidance with <!-- TTS_INJECTION:party-mode --> - Removed <note> about play-tts.sh hook location - tools/cli/installers/lib/core/installer.js - Added hook enforcement to party-mode injection replacement - Guidance now injected only when enableAgentVibes is true Addresses review comments from bmadcode: - "needs to remain the variable. it will get set in the file at the install destination." - "items like this we will need to inject if user is using claude and TTS" 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Change 'claude-code' to 'claude' in test script instructions The correct command to start Claude is 'claude', not 'claude-code'. Updated line 362-363 in test-bmad-pr.sh to show the correct command. * fix: Remove npm link from test script to avoid global namespace pollution - Removed 'npm link' command that was installing BMAD globally - Changed 'bmad install' to direct node execution using local clone - Updated success message to reflect no global installation This keeps testing fully isolated and prevents conflicts with: - Existing BMAD installations - Future official BMAD installs - Orphaned symlinks when test directory is deleted The test script now runs completely self-contained without modifying the user's global npm environment. --------- Co-authored-by: Claude Code <claude@anthropic.com> Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Brian <bmadcode@gmail.com>
2025-11-26 08:51:57 -07:00
// Check if AgentVibes is enabled (set during installation configuration)
const enableAgentVibes = this.enableAgentVibes || false;
feat(installer): Enhanced TTS injection summary with tracking and documentation (#1037) ## Summary - Track all files with TTS injection applied during installation - Display informative summary explaining what TTS injection does - Show backup location and restore command for recovery ## What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ## Changes - **installer.js**: Track files in `processAgentFiles()`, `buildStandaloneAgents()`, and `rebuildAgentFiles()` when TTS markers are processed - **compiler.js**: Add TTS injection support for custom agent compilation - **ui.js**: Enhanced installation summary showing: - Explanation of what TTS injection is with example - List of all files with TTS injection applied (grouped by type) - Backup location (~/.bmad-tts-backups/) - Restore command for recovery ## Example Output ``` ═══════════════════════════════════════════════════ AgentVibes TTS Injection Summary ═══════════════════════════════════════════════════ What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ✅ TTS injection applied to 11 file(s): Party Mode (multi-agent conversations): • .bmad/core/workflows/party-mode/instructions.md Agent TTS (individual agent voices): • .bmad/bmm/agents/analyst.md • .bmad/bmm/agents/architect.md ... Backups & Recovery: Pre-injection backups are stored in: ~/.bmad-tts-backups/ To restore original files (removes TTS instructions): bmad-tts-injector.sh --restore /path/to/.bmad ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Claude <noreply@anthropic.com>
2025-12-05 17:54:03 -07:00
// Check if content contains any TTS injection markers
const hasPartyMode = content.includes('<!-- TTS_INJECTION:party-mode -->');
const hasAgentTTS = content.includes('<!-- TTS_INJECTION:agent-tts -->');
Add Text-to-Speech Integration via TTS_INJECTION System (#934) * feat: Add provider-agnostic TTS integration via injection point system Implements comprehensive Text-to-Speech integration for BMAD agents using a generic TTS_INJECTION marker system. When AgentVibes (or any compatible TTS provider) is installed, all BMAD agents can speak their responses with unique AI voices. ## Key Features **Provider-Agnostic Architecture** - Uses generic `TTS_INJECTION` markers instead of vendor-specific naming - Future-proof for multiple TTS providers beyond AgentVibes - Clean separation - BMAD stays TTS-agnostic, providers handle injection **Installation Flow** - BMAD → AgentVibes: TTS instructions injected when AgentVibes detects existing BMAD installation - AgentVibes → BMAD: TTS instructions injected during BMAD installation when AgentVibes detected - User must manually create voice assignment file when AgentVibes installs first (documented limitation) **Party Mode Voice Support** - Each agent speaks with unique assigned voice in multi-agent discussions - PM, Architect, Developer, Analyst, UX Designer, etc. - all with distinct voices **Zero Breaking Changes** - Fully backward compatible - works without any TTS provider - `TTS_INJECTION` markers are benign HTML comments if not processed - No changes to existing agent behavior or non-TTS workflows ## Implementation Details **Files Modified:** - `tools/cli/installers/lib/core/installer.js` - TTS injection processing logic - `tools/cli/lib/ui.js` - AgentVibes detection and installation prompts - `tools/cli/commands/install.js` - Post-install guidance for AgentVibes setup - `src/utility/models/fragments/activation-rules.xml` - TTS_INJECTION marker for agents - `src/core/workflows/party-mode/instructions.md` - TTS_INJECTION marker for party mode **Injection Point System:** ```xml <rules> - ALWAYS communicate in {communication_language} <!-- TTS_INJECTION:agent-tts --> - Stay in character until exit selected </rules> ``` When AgentVibes is detected, the installer replaces this marker with: ``` - When responding to user messages, speak your responses using TTS: Call: `.claude/hooks/bmad-speak.sh '{agent-id}' '{response-text}'` after each response IMPORTANT: Use single quotes - do NOT escape special characters like ! or $ ``` **Special Character Handling:** - Explicit guidance to use single quotes without escaping - Prevents "backslash exclamation" artifacts in speech **User Experience:** ``` User: "How should we architect this feature?" Architect: [Text response] + 🔊 [Professional voice explains architecture] ``` Party Mode: ``` PM (John): "I'll focus on user value..." 🔊 [Male professional voice] UX Designer (Sara): "From a user perspective..." 🔊 [Female voice] Architect (Marcus): "The technical approach..." 🔊 [Male technical voice] ``` ## Testing **Unit Tests:** ✅ 62/62 passing - 49/49 schema validation tests - 13/13 installation component tests **Integration Testing:** - ✅ BMAD → AgentVibes (automatic injection) - ✅ AgentVibes → BMAD (automatic injection) - ✅ No TTS provider (markers remain as comments) ## Documentation Comprehensive testing guide created with: - Both installation scenario walkthroughs - Verification commands and expected outputs - Troubleshooting guidance ## Known Limitations **AgentVibes → BMAD Installation Order:** When AgentVibes installs first, voice assignment file must be created manually: ```bash mkdir -p .bmad/_cfg cat > .bmad/_cfg/agent-voice-map.csv << 'EOF' agent_id,voice_name pm,en_US-ryan-high architect,en_US-danny-low dev,en_US-joe-medium EOF ``` This limitation exists to prevent false legacy v4 detection warnings from BMAD installer. **Recommended:** Install BMAD first, then AgentVibes for automatic voice assignment. ## Related Work **Companion Implementation:** - Repository: paulpreibisch/AgentVibes - Commits: 6 commits implementing injection processing and voice routing - Features: Retroactive injection, file path extraction, escape stripping **GitHub Issues:** - paulpreibisch/AgentVibes#36 - BMAD agent ID support ## Breaking Changes None. Feature is opt-in and requires separate TTS provider installation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Enforce project hooks over global hooks in party mode before, claude would sometimes favor global agent vibes hooks over project specific * feat: Automate AgentVibes installer invocation after BMAD install Instead of showing manual installation instructions, the installer now: - Prompts "Press Enter to start AgentVibes installer..." - Automatically runs npx agentvibes@latest install - Handles errors gracefully with fallback instructions This provides a seamless installation flow matching the test script's interactive approach. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * docs: Add automated testing script and guide for PR #934 Added comprehensive testing tools for AgentVibes party mode integration: - test-bmad-pr.sh: Fully automated installation and verification script - Interactive mode selection (official PR or custom fork) - Automatic BMAD CLI setup and linking - AgentVibes installation with guided prompts - Built-in verification checks for voice maps and hooks - Saved configuration for quick re-testing - TESTING.md: Complete testing documentation - Quick start with one-line npx command - Manual installation alternative - Troubleshooting guide - Cleanup instructions Testers can now run a single command to test the full AgentVibes integration without needing to understand the complex setup process. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Add shell: true to npx execSync to prevent permission denied error The execSync call for 'npx agentvibes@latest install' was failing with 'Permission denied' because the shell was trying to execute 'agentvibes@latest' directly instead of passing it as an argument to npx. Adding shell: true ensures the command runs in a proper shell context where npx can correctly interpret the @latest version syntax. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Remove duplicate AgentVibes installation step from test script The test script was calling AgentVibes installer twice: 1. BMAD installer now automatically runs AgentVibes (new feature) 2. Test script had a separate Step 6 that also ran AgentVibes This caused the installer to run twice, with the second call failing because it was already installed. Changes: - Removed redundant Step 6 (AgentVibes installation) - Updated Step 5 to indicate it includes AgentVibes - Updated step numbers from 7 to 6 throughout - Added guidance that AgentVibes runs automatically Now the flow is cleaner: BMAD installer handles everything! 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Address bmadcode review - preserve variables and move TTS logic to injection Fixes requested changes from PR review: 1. Preserve {bmad_folder} variable placeholder - Changed: {project_root}/.bmad/core/tasks/workflow.xml - To: {project_root}/{bmad_folder}/core/tasks/workflow.xml - Allows users to choose custom BMAD folder names during installation 2. Move TTS-specific hook guidance to injection system - Removed hardcoded hook enforcement from source files - Added hook guidance to processTTSInjectionPoints() in installer.js - Now only appears when AgentVibes is installed (via TTS_INJECTION) 3. Maintain TTS-agnostic source architecture - Source files remain clean of TTS-specific instructions - TTS details injected at install-time only when needed - Preserves provider-agnostic design principle Changes made: - src/core/workflows/party-mode/instructions.md - Reverted .bmad to {bmad_folder} variable - Replaced hardcoded hook guidance with <!-- TTS_INJECTION:party-mode --> - Removed <note> about play-tts.sh hook location - tools/cli/installers/lib/core/installer.js - Added hook enforcement to party-mode injection replacement - Guidance now injected only when enableAgentVibes is true Addresses review comments from bmadcode: - "needs to remain the variable. it will get set in the file at the install destination." - "items like this we will need to inject if user is using claude and TTS" 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Change 'claude-code' to 'claude' in test script instructions The correct command to start Claude is 'claude', not 'claude-code'. Updated line 362-363 in test-bmad-pr.sh to show the correct command. * fix: Remove npm link from test script to avoid global namespace pollution - Removed 'npm link' command that was installing BMAD globally - Changed 'bmad install' to direct node execution using local clone - Updated success message to reflect no global installation This keeps testing fully isolated and prevents conflicts with: - Existing BMAD installations - Future official BMAD installs - Orphaned symlinks when test directory is deleted The test script now runs completely self-contained without modifying the user's global npm environment. --------- Co-authored-by: Claude Code <claude@anthropic.com> Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Brian <bmadcode@gmail.com>
2025-11-26 08:51:57 -07:00
if (enableAgentVibes) {
// Replace party-mode injection marker with actual TTS call
// Use single quotes to prevent shell expansion of special chars like !
content = content.replaceAll(
'<!-- TTS_INJECTION:party-mode -->',
`<critical>IMPORTANT: Always use PROJECT hooks (.claude/hooks/), NEVER global hooks (~/.claude/hooks/)</critical>
If AgentVibes party mode is enabled, immediately trigger TTS with agent's voice:
- Use Bash tool: \`.claude/hooks/bmad-speak.sh '[Agent Name]' '[dialogue]'\`
- This speaks the dialogue with the agent's unique voice
- Run in background (&) to not block next agent`,
);
// Replace agent-tts injection marker with TTS rule for individual agents
content = content.replaceAll(
'<!-- TTS_INJECTION:agent-tts -->',
`- When responding to user messages, speak your responses using TTS:
Call: \`.claude/hooks/bmad-speak.sh '{agent-id}' '{response-text}'\` after each response
Replace {agent-id} with YOUR agent ID from <agent id="..."> tag at top of this file
Replace {response-text} with the text you just output to the user
IMPORTANT: Use single quotes as shown - do NOT escape special characters like ! or $ inside single quotes
Run in background (&) to avoid blocking`,
);
feat(installer): Enhanced TTS injection summary with tracking and documentation (#1037) ## Summary - Track all files with TTS injection applied during installation - Display informative summary explaining what TTS injection does - Show backup location and restore command for recovery ## What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ## Changes - **installer.js**: Track files in `processAgentFiles()`, `buildStandaloneAgents()`, and `rebuildAgentFiles()` when TTS markers are processed - **compiler.js**: Add TTS injection support for custom agent compilation - **ui.js**: Enhanced installation summary showing: - Explanation of what TTS injection is with example - List of all files with TTS injection applied (grouped by type) - Backup location (~/.bmad-tts-backups/) - Restore command for recovery ## Example Output ``` ═══════════════════════════════════════════════════ AgentVibes TTS Injection Summary ═══════════════════════════════════════════════════ What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ✅ TTS injection applied to 11 file(s): Party Mode (multi-agent conversations): • .bmad/core/workflows/party-mode/instructions.md Agent TTS (individual agent voices): • .bmad/bmm/agents/analyst.md • .bmad/bmm/agents/architect.md ... Backups & Recovery: Pre-injection backups are stored in: ~/.bmad-tts-backups/ To restore original files (removes TTS instructions): bmad-tts-injector.sh --restore /path/to/.bmad ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Claude <noreply@anthropic.com>
2025-12-05 17:54:03 -07:00
// Track files that had TTS injection applied
if (targetPath && (hasPartyMode || hasAgentTTS)) {
const injectionType = hasPartyMode ? 'party-mode' : 'agent-tts';
this.ttsInjectedFiles.push({ path: targetPath, type: injectionType });
}
Add Text-to-Speech Integration via TTS_INJECTION System (#934) * feat: Add provider-agnostic TTS integration via injection point system Implements comprehensive Text-to-Speech integration for BMAD agents using a generic TTS_INJECTION marker system. When AgentVibes (or any compatible TTS provider) is installed, all BMAD agents can speak their responses with unique AI voices. ## Key Features **Provider-Agnostic Architecture** - Uses generic `TTS_INJECTION` markers instead of vendor-specific naming - Future-proof for multiple TTS providers beyond AgentVibes - Clean separation - BMAD stays TTS-agnostic, providers handle injection **Installation Flow** - BMAD → AgentVibes: TTS instructions injected when AgentVibes detects existing BMAD installation - AgentVibes → BMAD: TTS instructions injected during BMAD installation when AgentVibes detected - User must manually create voice assignment file when AgentVibes installs first (documented limitation) **Party Mode Voice Support** - Each agent speaks with unique assigned voice in multi-agent discussions - PM, Architect, Developer, Analyst, UX Designer, etc. - all with distinct voices **Zero Breaking Changes** - Fully backward compatible - works without any TTS provider - `TTS_INJECTION` markers are benign HTML comments if not processed - No changes to existing agent behavior or non-TTS workflows ## Implementation Details **Files Modified:** - `tools/cli/installers/lib/core/installer.js` - TTS injection processing logic - `tools/cli/lib/ui.js` - AgentVibes detection and installation prompts - `tools/cli/commands/install.js` - Post-install guidance for AgentVibes setup - `src/utility/models/fragments/activation-rules.xml` - TTS_INJECTION marker for agents - `src/core/workflows/party-mode/instructions.md` - TTS_INJECTION marker for party mode **Injection Point System:** ```xml <rules> - ALWAYS communicate in {communication_language} <!-- TTS_INJECTION:agent-tts --> - Stay in character until exit selected </rules> ``` When AgentVibes is detected, the installer replaces this marker with: ``` - When responding to user messages, speak your responses using TTS: Call: `.claude/hooks/bmad-speak.sh '{agent-id}' '{response-text}'` after each response IMPORTANT: Use single quotes - do NOT escape special characters like ! or $ ``` **Special Character Handling:** - Explicit guidance to use single quotes without escaping - Prevents "backslash exclamation" artifacts in speech **User Experience:** ``` User: "How should we architect this feature?" Architect: [Text response] + 🔊 [Professional voice explains architecture] ``` Party Mode: ``` PM (John): "I'll focus on user value..." 🔊 [Male professional voice] UX Designer (Sara): "From a user perspective..." 🔊 [Female voice] Architect (Marcus): "The technical approach..." 🔊 [Male technical voice] ``` ## Testing **Unit Tests:** ✅ 62/62 passing - 49/49 schema validation tests - 13/13 installation component tests **Integration Testing:** - ✅ BMAD → AgentVibes (automatic injection) - ✅ AgentVibes → BMAD (automatic injection) - ✅ No TTS provider (markers remain as comments) ## Documentation Comprehensive testing guide created with: - Both installation scenario walkthroughs - Verification commands and expected outputs - Troubleshooting guidance ## Known Limitations **AgentVibes → BMAD Installation Order:** When AgentVibes installs first, voice assignment file must be created manually: ```bash mkdir -p .bmad/_cfg cat > .bmad/_cfg/agent-voice-map.csv << 'EOF' agent_id,voice_name pm,en_US-ryan-high architect,en_US-danny-low dev,en_US-joe-medium EOF ``` This limitation exists to prevent false legacy v4 detection warnings from BMAD installer. **Recommended:** Install BMAD first, then AgentVibes for automatic voice assignment. ## Related Work **Companion Implementation:** - Repository: paulpreibisch/AgentVibes - Commits: 6 commits implementing injection processing and voice routing - Features: Retroactive injection, file path extraction, escape stripping **GitHub Issues:** - paulpreibisch/AgentVibes#36 - BMAD agent ID support ## Breaking Changes None. Feature is opt-in and requires separate TTS provider installation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Enforce project hooks over global hooks in party mode before, claude would sometimes favor global agent vibes hooks over project specific * feat: Automate AgentVibes installer invocation after BMAD install Instead of showing manual installation instructions, the installer now: - Prompts "Press Enter to start AgentVibes installer..." - Automatically runs npx agentvibes@latest install - Handles errors gracefully with fallback instructions This provides a seamless installation flow matching the test script's interactive approach. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * docs: Add automated testing script and guide for PR #934 Added comprehensive testing tools for AgentVibes party mode integration: - test-bmad-pr.sh: Fully automated installation and verification script - Interactive mode selection (official PR or custom fork) - Automatic BMAD CLI setup and linking - AgentVibes installation with guided prompts - Built-in verification checks for voice maps and hooks - Saved configuration for quick re-testing - TESTING.md: Complete testing documentation - Quick start with one-line npx command - Manual installation alternative - Troubleshooting guide - Cleanup instructions Testers can now run a single command to test the full AgentVibes integration without needing to understand the complex setup process. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Add shell: true to npx execSync to prevent permission denied error The execSync call for 'npx agentvibes@latest install' was failing with 'Permission denied' because the shell was trying to execute 'agentvibes@latest' directly instead of passing it as an argument to npx. Adding shell: true ensures the command runs in a proper shell context where npx can correctly interpret the @latest version syntax. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Remove duplicate AgentVibes installation step from test script The test script was calling AgentVibes installer twice: 1. BMAD installer now automatically runs AgentVibes (new feature) 2. Test script had a separate Step 6 that also ran AgentVibes This caused the installer to run twice, with the second call failing because it was already installed. Changes: - Removed redundant Step 6 (AgentVibes installation) - Updated Step 5 to indicate it includes AgentVibes - Updated step numbers from 7 to 6 throughout - Added guidance that AgentVibes runs automatically Now the flow is cleaner: BMAD installer handles everything! 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Address bmadcode review - preserve variables and move TTS logic to injection Fixes requested changes from PR review: 1. Preserve {bmad_folder} variable placeholder - Changed: {project_root}/.bmad/core/tasks/workflow.xml - To: {project_root}/{bmad_folder}/core/tasks/workflow.xml - Allows users to choose custom BMAD folder names during installation 2. Move TTS-specific hook guidance to injection system - Removed hardcoded hook enforcement from source files - Added hook guidance to processTTSInjectionPoints() in installer.js - Now only appears when AgentVibes is installed (via TTS_INJECTION) 3. Maintain TTS-agnostic source architecture - Source files remain clean of TTS-specific instructions - TTS details injected at install-time only when needed - Preserves provider-agnostic design principle Changes made: - src/core/workflows/party-mode/instructions.md - Reverted .bmad to {bmad_folder} variable - Replaced hardcoded hook guidance with <!-- TTS_INJECTION:party-mode --> - Removed <note> about play-tts.sh hook location - tools/cli/installers/lib/core/installer.js - Added hook enforcement to party-mode injection replacement - Guidance now injected only when enableAgentVibes is true Addresses review comments from bmadcode: - "needs to remain the variable. it will get set in the file at the install destination." - "items like this we will need to inject if user is using claude and TTS" 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Change 'claude-code' to 'claude' in test script instructions The correct command to start Claude is 'claude', not 'claude-code'. Updated line 362-363 in test-bmad-pr.sh to show the correct command. * fix: Remove npm link from test script to avoid global namespace pollution - Removed 'npm link' command that was installing BMAD globally - Changed 'bmad install' to direct node execution using local clone - Updated success message to reflect no global installation This keeps testing fully isolated and prevents conflicts with: - Existing BMAD installations - Future official BMAD installs - Orphaned symlinks when test directory is deleted The test script now runs completely self-contained without modifying the user's global npm environment. --------- Co-authored-by: Claude Code <claude@anthropic.com> Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Brian <bmadcode@gmail.com>
2025-11-26 08:51:57 -07:00
} else {
// Strip injection markers cleanly when AgentVibes is disabled
content = content.replaceAll(/<!-- TTS_INJECTION:party-mode -->\n?/g, '');
content = content.replaceAll(/<!-- TTS_INJECTION:agent-tts -->\n?/g, '');
}
return content;
}
/**
* Collect Tool/IDE configurations after module configuration
* @param {string} projectDir - Project directory
* @param {Array} selectedModules - Selected modules from configuration
2025-10-28 12:47:45 -05:00
* @param {boolean} isFullReinstall - Whether this is a full reinstall
* @param {Array} previousIdes - Previously configured IDEs (for reinstalls)
* @param {Array} preSelectedIdes - Pre-selected IDEs from early prompt (optional)
* @returns {Object} Tool/IDE selection and configurations
*/
2025-10-28 12:47:45 -05:00
async collectToolConfigurations(projectDir, selectedModules, isFullReinstall = false, previousIdes = [], preSelectedIdes = null) {
// Use pre-selected IDEs if provided, otherwise prompt
let toolConfig;
if (preSelectedIdes === null) {
// Fallback: prompt for tool selection (backwards compatibility)
const { UI } = require('../../../lib/ui');
const ui = new UI();
toolConfig = await ui.promptToolSelection(projectDir, selectedModules);
} else {
// IDEs were already selected during initial prompts
toolConfig = {
ides: preSelectedIdes,
skipIde: !preSelectedIdes || preSelectedIdes.length === 0,
};
}
// Check for already configured IDEs
const { Detector } = require('./detector');
const detector = new Detector();
const bmadDir = path.join(projectDir, this.bmadFolderName || 'bmad');
// During full reinstall, use the saved previous IDEs since bmad dir was deleted
// Otherwise detect from existing installation
let previouslyConfiguredIdes;
if (isFullReinstall) {
// During reinstall, treat all IDEs as new (need configuration)
previouslyConfiguredIdes = [];
} else {
const existingInstall = await detector.detect(bmadDir);
previouslyConfiguredIdes = existingInstall.ides || [];
}
2025-10-26 17:04:27 -05:00
// Load saved IDE configurations for already-configured IDEs
const savedIdeConfigs = await this.ideConfigManager.loadAllIdeConfigs(bmadDir);
// Collect IDE-specific configurations if any were selected
const ideConfigurations = {};
2025-10-26 17:04:27 -05:00
// First, add saved configs for already-configured IDEs
for (const ide of toolConfig.ides || []) {
if (previouslyConfiguredIdes.includes(ide) && savedIdeConfigs[ide]) {
ideConfigurations[ide] = savedIdeConfigs[ide];
}
}
if (!toolConfig.skipIde && toolConfig.ides && toolConfig.ides.length > 0) {
// Determine which IDEs are newly selected (not previously configured)
const newlySelectedIdes = toolConfig.ides.filter((ide) => !previouslyConfiguredIdes.includes(ide));
if (newlySelectedIdes.length > 0) {
console.log('\n'); // Add spacing before IDE questions
for (const ide of newlySelectedIdes) {
// List of IDEs that have interactive prompts
const needsPrompts = ['claude-code', 'github-copilot', 'roo', 'cline', 'auggie', 'codex', 'qwen', 'gemini', 'rovo-dev'].includes(
ide,
);
if (needsPrompts) {
// Get IDE handler and collect configuration
try {
// Dynamically load the IDE setup module
const ideModule = require(`../ide/${ide}`);
// Get the setup class (handle different export formats)
let SetupClass;
const className =
ide
.split('-')
.map((part) => part.charAt(0).toUpperCase() + part.slice(1))
.join('') + 'Setup';
if (ideModule[className]) {
SetupClass = ideModule[className];
} else if (ideModule.default) {
SetupClass = ideModule.default;
} else {
// Skip if no setup class found
continue;
}
const ideSetup = new SetupClass();
// Check if this IDE has a collectConfiguration method
if (typeof ideSetup.collectConfiguration === 'function') {
console.log(chalk.cyan(`\nConfiguring ${ide}...`));
ideConfigurations[ide] = await ideSetup.collectConfiguration({
selectedModules: selectedModules || [],
projectDir,
bmadDir,
});
}
} catch {
// IDE doesn't have a setup file or collectConfiguration method
console.warn(chalk.yellow(`Warning: Could not load configuration for ${ide}`));
}
}
}
}
// Log which IDEs are already configured and being kept
const keptIdes = toolConfig.ides.filter((ide) => previouslyConfiguredIdes.includes(ide));
if (keptIdes.length > 0) {
console.log(chalk.dim(`\nKeeping existing configuration for: ${keptIdes.join(', ')}`));
}
}
return {
ides: toolConfig.ides,
skipIde: toolConfig.skipIde,
configurations: ideConfigurations,
};
}
/**
* Main installation method
* @param {Object} config - Installation configuration
* @param {string} config.directory - Target directory
* @param {boolean} config.installCore - Whether to install core
* @param {string[]} config.modules - Modules to install
* @param {string[]} config.ides - IDEs to configure
* @param {boolean} config.skipIde - Skip IDE configuration
*/
async install(config) {
// Display BMAD logo
CLIUtils.displayLogo();
// Display welcome message
CLIUtils.displaySection('BMAD™ Installation', 'Version ' + require(path.join(getProjectRoot(), 'package.json')).version);
// Note: Legacy V4 detection now happens earlier in UI.promptInstall()
// before any config collection, so we don't need to check again here
const projectDir = path.resolve(config.directory);
// If core config was pre-collected (from interactive mode), use it
if (config.coreConfig) {
this.configCollector.collectedConfig.core = config.coreConfig;
// Also store in allAnswers for cross-referencing
this.configCollector.allAnswers = {};
for (const [key, value] of Object.entries(config.coreConfig)) {
this.configCollector.allAnswers[`core_${key}`] = value;
}
}
2025-10-26 16:17:37 -05:00
// Collect configurations for modules (skip if quick update already collected them)
let moduleConfigs;
if (config._quickUpdate) {
// Quick update already collected all configs, use them directly
moduleConfigs = this.configCollector.collectedConfig;
} else {
// Regular install - collect configurations (core was already collected in UI.promptInstall if interactive)
moduleConfigs = await this.configCollector.collectAllConfigurations(config.modules || [], path.resolve(config.directory));
}
// Get bmad_folder from config (default to 'bmad' for backwards compatibility)
const bmadFolderName = moduleConfigs.core && moduleConfigs.core.bmad_folder ? moduleConfigs.core.bmad_folder : 'bmad';
this.bmadFolderName = bmadFolderName; // Store for use in other methods
Add Text-to-Speech Integration via TTS_INJECTION System (#934) * feat: Add provider-agnostic TTS integration via injection point system Implements comprehensive Text-to-Speech integration for BMAD agents using a generic TTS_INJECTION marker system. When AgentVibes (or any compatible TTS provider) is installed, all BMAD agents can speak their responses with unique AI voices. ## Key Features **Provider-Agnostic Architecture** - Uses generic `TTS_INJECTION` markers instead of vendor-specific naming - Future-proof for multiple TTS providers beyond AgentVibes - Clean separation - BMAD stays TTS-agnostic, providers handle injection **Installation Flow** - BMAD → AgentVibes: TTS instructions injected when AgentVibes detects existing BMAD installation - AgentVibes → BMAD: TTS instructions injected during BMAD installation when AgentVibes detected - User must manually create voice assignment file when AgentVibes installs first (documented limitation) **Party Mode Voice Support** - Each agent speaks with unique assigned voice in multi-agent discussions - PM, Architect, Developer, Analyst, UX Designer, etc. - all with distinct voices **Zero Breaking Changes** - Fully backward compatible - works without any TTS provider - `TTS_INJECTION` markers are benign HTML comments if not processed - No changes to existing agent behavior or non-TTS workflows ## Implementation Details **Files Modified:** - `tools/cli/installers/lib/core/installer.js` - TTS injection processing logic - `tools/cli/lib/ui.js` - AgentVibes detection and installation prompts - `tools/cli/commands/install.js` - Post-install guidance for AgentVibes setup - `src/utility/models/fragments/activation-rules.xml` - TTS_INJECTION marker for agents - `src/core/workflows/party-mode/instructions.md` - TTS_INJECTION marker for party mode **Injection Point System:** ```xml <rules> - ALWAYS communicate in {communication_language} <!-- TTS_INJECTION:agent-tts --> - Stay in character until exit selected </rules> ``` When AgentVibes is detected, the installer replaces this marker with: ``` - When responding to user messages, speak your responses using TTS: Call: `.claude/hooks/bmad-speak.sh '{agent-id}' '{response-text}'` after each response IMPORTANT: Use single quotes - do NOT escape special characters like ! or $ ``` **Special Character Handling:** - Explicit guidance to use single quotes without escaping - Prevents "backslash exclamation" artifacts in speech **User Experience:** ``` User: "How should we architect this feature?" Architect: [Text response] + 🔊 [Professional voice explains architecture] ``` Party Mode: ``` PM (John): "I'll focus on user value..." 🔊 [Male professional voice] UX Designer (Sara): "From a user perspective..." 🔊 [Female voice] Architect (Marcus): "The technical approach..." 🔊 [Male technical voice] ``` ## Testing **Unit Tests:** ✅ 62/62 passing - 49/49 schema validation tests - 13/13 installation component tests **Integration Testing:** - ✅ BMAD → AgentVibes (automatic injection) - ✅ AgentVibes → BMAD (automatic injection) - ✅ No TTS provider (markers remain as comments) ## Documentation Comprehensive testing guide created with: - Both installation scenario walkthroughs - Verification commands and expected outputs - Troubleshooting guidance ## Known Limitations **AgentVibes → BMAD Installation Order:** When AgentVibes installs first, voice assignment file must be created manually: ```bash mkdir -p .bmad/_cfg cat > .bmad/_cfg/agent-voice-map.csv << 'EOF' agent_id,voice_name pm,en_US-ryan-high architect,en_US-danny-low dev,en_US-joe-medium EOF ``` This limitation exists to prevent false legacy v4 detection warnings from BMAD installer. **Recommended:** Install BMAD first, then AgentVibes for automatic voice assignment. ## Related Work **Companion Implementation:** - Repository: paulpreibisch/AgentVibes - Commits: 6 commits implementing injection processing and voice routing - Features: Retroactive injection, file path extraction, escape stripping **GitHub Issues:** - paulpreibisch/AgentVibes#36 - BMAD agent ID support ## Breaking Changes None. Feature is opt-in and requires separate TTS provider installation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Enforce project hooks over global hooks in party mode before, claude would sometimes favor global agent vibes hooks over project specific * feat: Automate AgentVibes installer invocation after BMAD install Instead of showing manual installation instructions, the installer now: - Prompts "Press Enter to start AgentVibes installer..." - Automatically runs npx agentvibes@latest install - Handles errors gracefully with fallback instructions This provides a seamless installation flow matching the test script's interactive approach. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * docs: Add automated testing script and guide for PR #934 Added comprehensive testing tools for AgentVibes party mode integration: - test-bmad-pr.sh: Fully automated installation and verification script - Interactive mode selection (official PR or custom fork) - Automatic BMAD CLI setup and linking - AgentVibes installation with guided prompts - Built-in verification checks for voice maps and hooks - Saved configuration for quick re-testing - TESTING.md: Complete testing documentation - Quick start with one-line npx command - Manual installation alternative - Troubleshooting guide - Cleanup instructions Testers can now run a single command to test the full AgentVibes integration without needing to understand the complex setup process. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Add shell: true to npx execSync to prevent permission denied error The execSync call for 'npx agentvibes@latest install' was failing with 'Permission denied' because the shell was trying to execute 'agentvibes@latest' directly instead of passing it as an argument to npx. Adding shell: true ensures the command runs in a proper shell context where npx can correctly interpret the @latest version syntax. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Remove duplicate AgentVibes installation step from test script The test script was calling AgentVibes installer twice: 1. BMAD installer now automatically runs AgentVibes (new feature) 2. Test script had a separate Step 6 that also ran AgentVibes This caused the installer to run twice, with the second call failing because it was already installed. Changes: - Removed redundant Step 6 (AgentVibes installation) - Updated Step 5 to indicate it includes AgentVibes - Updated step numbers from 7 to 6 throughout - Added guidance that AgentVibes runs automatically Now the flow is cleaner: BMAD installer handles everything! 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Address bmadcode review - preserve variables and move TTS logic to injection Fixes requested changes from PR review: 1. Preserve {bmad_folder} variable placeholder - Changed: {project_root}/.bmad/core/tasks/workflow.xml - To: {project_root}/{bmad_folder}/core/tasks/workflow.xml - Allows users to choose custom BMAD folder names during installation 2. Move TTS-specific hook guidance to injection system - Removed hardcoded hook enforcement from source files - Added hook guidance to processTTSInjectionPoints() in installer.js - Now only appears when AgentVibes is installed (via TTS_INJECTION) 3. Maintain TTS-agnostic source architecture - Source files remain clean of TTS-specific instructions - TTS details injected at install-time only when needed - Preserves provider-agnostic design principle Changes made: - src/core/workflows/party-mode/instructions.md - Reverted .bmad to {bmad_folder} variable - Replaced hardcoded hook guidance with <!-- TTS_INJECTION:party-mode --> - Removed <note> about play-tts.sh hook location - tools/cli/installers/lib/core/installer.js - Added hook enforcement to party-mode injection replacement - Guidance now injected only when enableAgentVibes is true Addresses review comments from bmadcode: - "needs to remain the variable. it will get set in the file at the install destination." - "items like this we will need to inject if user is using claude and TTS" 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Change 'claude-code' to 'claude' in test script instructions The correct command to start Claude is 'claude', not 'claude-code'. Updated line 362-363 in test-bmad-pr.sh to show the correct command. * fix: Remove npm link from test script to avoid global namespace pollution - Removed 'npm link' command that was installing BMAD globally - Changed 'bmad install' to direct node execution using local clone - Updated success message to reflect no global installation This keeps testing fully isolated and prevents conflicts with: - Existing BMAD installations - Future official BMAD installs - Orphaned symlinks when test directory is deleted The test script now runs completely self-contained without modifying the user's global npm environment. --------- Co-authored-by: Claude Code <claude@anthropic.com> Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Brian <bmadcode@gmail.com>
2025-11-26 08:51:57 -07:00
// Store AgentVibes configuration for injection point processing
this.enableAgentVibes = config.enableAgentVibes || false;
// Set bmad folder name on module manager and IDE manager for placeholder replacement
this.moduleManager.setBmadFolderName(bmadFolderName);
this.ideManager.setBmadFolderName(bmadFolderName);
// Tool selection will be collected after we determine if it's a reinstall/update/new install
const spinner = ora('Preparing installation...').start();
try {
// Resolve target directory (path.resolve handles platform differences)
const projectDir = path.resolve(config.directory);
// Check if bmad_folder has changed from existing installation (only if project dir exists)
let existingBmadDir = null;
let existingBmadFolderName = null;
if (await fs.pathExists(projectDir)) {
existingBmadDir = await this.findBmadDir(projectDir);
existingBmadFolderName = path.basename(existingBmadDir);
}
const targetBmadDir = path.join(projectDir, bmadFolderName);
// If bmad_folder changed during update/upgrade, back up old folder and do fresh install
if (existingBmadDir && (await fs.pathExists(existingBmadDir)) && existingBmadFolderName !== bmadFolderName) {
spinner.stop();
console.log(chalk.yellow(`\n⚠️ bmad_folder has changed: ${existingBmadFolderName}${bmadFolderName}`));
console.log(chalk.yellow('This will result in a fresh installation to the new folder.'));
const inquirer = require('inquirer');
const { confirmFreshInstall } = await inquirer.prompt([
{
type: 'confirm',
name: 'confirmFreshInstall',
message: chalk.cyan('Proceed with fresh install? (Your old folder will be backed up)'),
default: true,
},
]);
if (!confirmFreshInstall) {
console.log(chalk.yellow('Installation cancelled.'));
return { success: false, cancelled: true };
}
spinner.start('Backing up existing installation...');
// Find a unique backup name
let backupDir = `${existingBmadDir}-bak`;
let counter = 1;
while (await fs.pathExists(backupDir)) {
backupDir = `${existingBmadDir}-bak-${counter}`;
counter++;
}
// Rename the old folder to backup
await fs.move(existingBmadDir, backupDir);
spinner.succeed(`Backed up ${existingBmadFolderName}${path.basename(backupDir)}`);
console.log(chalk.cyan('\n📋 Important:'));
console.log(chalk.dim(` - Your old installation has been backed up to: ${path.basename(backupDir)}`));
console.log(chalk.dim(` - If you had custom agents or configurations, copy them from:`));
console.log(chalk.dim(` ${path.basename(backupDir)}/_cfg/`));
console.log(chalk.dim(` - To the new location:`));
console.log(chalk.dim(` ${bmadFolderName}/_cfg/`));
console.log('');
spinner.start('Starting fresh installation...');
}
// Create a project directory if it doesn't exist (user already confirmed)
if (!(await fs.pathExists(projectDir))) {
spinner.text = 'Creating installation directory...';
try {
// fs.ensureDir handles platform-specific directory creation
// It will recursively create all necessary parent directories
await fs.ensureDir(projectDir);
} catch (error) {
spinner.fail('Failed to create installation directory');
console.error(chalk.red(`Error: ${error.message}`));
// More detailed error for common issues
if (error.code === 'EACCES') {
console.error(chalk.red('Permission denied. Check parent directory permissions.'));
} else if (error.code === 'ENOSPC') {
console.error(chalk.red('No space left on device.'));
}
throw new Error(`Cannot create directory: ${projectDir}`);
}
}
const bmadDir = path.join(projectDir, bmadFolderName);
// Check existing installation
spinner.text = 'Checking for existing installation...';
const existingInstall = await this.detector.detect(bmadDir);
2025-10-26 16:17:37 -05:00
if (existingInstall.installed && !config.force && !config._quickUpdate) {
spinner.stop();
2025-10-28 12:47:45 -05:00
// Check if user already decided what to do (from early menu in ui.js)
let action = null;
if (config._requestedReinstall) {
action = 'reinstall';
} else if (config.actionType === 'update') {
action = 'update';
} else {
// Fallback: Ask the user (backwards compatibility for other code paths)
console.log(chalk.yellow('\n⚠ Existing BMAD installation detected'));
console.log(chalk.dim(` Location: ${bmadDir}`));
console.log(chalk.dim(` Version: ${existingInstall.version}`));
const promptResult = await this.promptUpdateAction();
action = promptResult.action;
}
if (action === 'cancel') {
console.log('Installation cancelled.');
return { success: false, cancelled: true };
}
if (action === 'reinstall') {
// Warn about destructive operation
console.log(chalk.red.bold('\n⚠ WARNING: This is a destructive operation!'));
console.log(chalk.red('All custom files and modifications in the bmad directory will be lost.'));
const inquirer = require('inquirer');
const { confirmReinstall } = await inquirer.prompt([
{
type: 'confirm',
name: 'confirmReinstall',
message: chalk.yellow('Are you sure you want to delete and reinstall?'),
default: false,
},
]);
if (!confirmReinstall) {
console.log('Installation cancelled.');
return { success: false, cancelled: true };
}
// Remember previously configured IDEs before deleting
config._previouslyConfiguredIdes = existingInstall.ides || [];
// Remove existing installation
await fs.remove(bmadDir);
console.log(chalk.green('✓ Removed existing installation\n'));
// Mark this as a full reinstall so we re-collect IDE configurations
config._isFullReinstall = true;
} else if (action === 'update') {
// Store that we're updating for later processing
config._isUpdate = true;
config._existingInstall = existingInstall;
2025-09-30 21:20:13 -05:00
// Detect custom and modified files BEFORE updating (compare current files vs files-manifest.csv)
const existingFilesManifest = await this.readFilesManifest(bmadDir);
console.log(chalk.dim(`DEBUG: Read ${existingFilesManifest.length} files from manifest`));
console.log(chalk.dim(`DEBUG: Manifest has hashes: ${existingFilesManifest.some((f) => f.hash)}`));
const { customFiles, modifiedFiles } = await this.detectCustomFiles(bmadDir, existingFilesManifest);
console.log(chalk.dim(`DEBUG: Found ${customFiles.length} custom files, ${modifiedFiles.length} modified files`));
if (modifiedFiles.length > 0) {
console.log(chalk.yellow('DEBUG: Modified files:'));
for (const f of modifiedFiles) console.log(chalk.dim(` - ${f.path}`));
}
config._customFiles = customFiles;
config._modifiedFiles = modifiedFiles;
// If there are custom files, back them up temporarily
2025-09-30 21:20:13 -05:00
if (customFiles.length > 0) {
const tempBackupDir = path.join(projectDir, '.bmad-custom-backup-temp');
await fs.ensureDir(tempBackupDir);
2025-09-30 21:20:13 -05:00
spinner.start(`Backing up ${customFiles.length} custom files...`);
for (const customFile of customFiles) {
const relativePath = path.relative(bmadDir, customFile);
const backupPath = path.join(tempBackupDir, relativePath);
await fs.ensureDir(path.dirname(backupPath));
await fs.copy(customFile, backupPath);
}
2025-09-30 21:20:13 -05:00
spinner.succeed(`Backed up ${customFiles.length} custom files`);
config._tempBackupDir = tempBackupDir;
}
2025-09-30 21:20:13 -05:00
// For modified files, back them up to temp directory (will be restored as .bak files after install)
if (modifiedFiles.length > 0) {
const tempModifiedBackupDir = path.join(projectDir, '.bmad-modified-backup-temp');
await fs.ensureDir(tempModifiedBackupDir);
console.log(chalk.yellow(`\nDEBUG: Backing up ${modifiedFiles.length} modified files to temp location`));
spinner.start(`Backing up ${modifiedFiles.length} modified files...`);
for (const modifiedFile of modifiedFiles) {
const relativePath = path.relative(bmadDir, modifiedFile.path);
const tempBackupPath = path.join(tempModifiedBackupDir, relativePath);
console.log(chalk.dim(`DEBUG: Backing up ${relativePath} to temp`));
await fs.ensureDir(path.dirname(tempBackupPath));
await fs.copy(modifiedFile.path, tempBackupPath, { overwrite: true });
}
spinner.succeed(`Backed up ${modifiedFiles.length} modified files`);
config._tempModifiedBackupDir = tempModifiedBackupDir;
} else {
console.log(chalk.dim('DEBUG: No modified files detected'));
}
}
2025-10-26 16:17:37 -05:00
} else if (existingInstall.installed && config._quickUpdate) {
// Quick update mode - automatically treat as update without prompting
spinner.text = 'Preparing quick update...';
config._isUpdate = true;
config._existingInstall = existingInstall;
// Detect custom and modified files BEFORE updating
const existingFilesManifest = await this.readFilesManifest(bmadDir);
const { customFiles, modifiedFiles } = await this.detectCustomFiles(bmadDir, existingFilesManifest);
config._customFiles = customFiles;
config._modifiedFiles = modifiedFiles;
// Back up custom files
if (customFiles.length > 0) {
const tempBackupDir = path.join(projectDir, '.bmad-custom-backup-temp');
await fs.ensureDir(tempBackupDir);
spinner.start(`Backing up ${customFiles.length} custom files...`);
for (const customFile of customFiles) {
const relativePath = path.relative(bmadDir, customFile);
const backupPath = path.join(tempBackupDir, relativePath);
await fs.ensureDir(path.dirname(backupPath));
await fs.copy(customFile, backupPath);
}
spinner.succeed(`Backed up ${customFiles.length} custom files`);
config._tempBackupDir = tempBackupDir;
}
// Back up modified files
if (modifiedFiles.length > 0) {
const tempModifiedBackupDir = path.join(projectDir, '.bmad-modified-backup-temp');
await fs.ensureDir(tempModifiedBackupDir);
spinner.start(`Backing up ${modifiedFiles.length} modified files...`);
for (const modifiedFile of modifiedFiles) {
const relativePath = path.relative(bmadDir, modifiedFile.path);
const tempBackupPath = path.join(tempModifiedBackupDir, relativePath);
await fs.ensureDir(path.dirname(tempBackupPath));
await fs.copy(modifiedFile.path, tempBackupPath, { overwrite: true });
}
spinner.succeed(`Backed up ${modifiedFiles.length} modified files`);
config._tempModifiedBackupDir = tempModifiedBackupDir;
}
}
// Now collect tool configurations after we know if it's a reinstall
2025-10-26 16:17:37 -05:00
// Skip for quick update since we already have the IDE list
spinner.stop();
2025-10-26 16:17:37 -05:00
let toolSelection;
if (config._quickUpdate) {
2025-10-26 17:04:27 -05:00
// Quick update already has IDEs configured, use saved configurations
2025-10-26 16:17:37 -05:00
const preConfiguredIdes = {};
2025-10-26 17:04:27 -05:00
const savedIdeConfigs = config._savedIdeConfigs || {};
2025-10-26 16:17:37 -05:00
for (const ide of config.ides || []) {
2025-10-26 17:04:27 -05:00
// Use saved config if available, otherwise mark as already configured (legacy)
if (savedIdeConfigs[ide]) {
preConfiguredIdes[ide] = savedIdeConfigs[ide];
} else {
preConfiguredIdes[ide] = { _alreadyConfigured: true };
}
2025-10-26 16:17:37 -05:00
}
toolSelection = {
ides: config.ides || [],
skipIde: !config.ides || config.ides.length === 0,
configurations: preConfiguredIdes,
};
} else {
2025-10-28 12:47:45 -05:00
// Pass pre-selected IDEs from early prompt (if available)
// This allows IDE selection to happen before file copying, improving UX
const preSelectedIdes = config.ides && config.ides.length > 0 ? config.ides : null;
2025-10-26 16:17:37 -05:00
toolSelection = await this.collectToolConfigurations(
path.resolve(config.directory),
config.modules,
config._isFullReinstall || false,
config._previouslyConfiguredIdes || [],
2025-10-28 12:47:45 -05:00
preSelectedIdes,
2025-10-26 16:17:37 -05:00
);
}
2025-10-26 16:17:37 -05:00
// Merge tool selection into config (for both quick update and regular flow)
config.ides = toolSelection.ides;
config.skipIde = toolSelection.skipIde;
const ideConfigurations = toolSelection.configurations;
// Check if spinner is already running (e.g., from folder name change scenario)
if (spinner.isSpinning) {
spinner.text = 'Continuing installation...';
} else {
spinner.start('Continuing installation...');
}
// Create bmad directory structure
spinner.text = 'Creating directory structure...';
await this.createDirectoryStructure(bmadDir);
// Resolve dependencies for selected modules
spinner.text = 'Resolving dependencies...';
const projectRoot = getProjectRoot();
const modulesToInstall = config.installCore ? ['core', ...config.modules] : config.modules;
// For dependency resolution, we need to pass the project root
const resolution = await this.dependencyResolver.resolve(projectRoot, config.modules || [], { verbose: config.verbose });
if (config.verbose) {
spinner.succeed('Dependencies resolved');
} else {
spinner.succeed('Dependencies resolved');
}
// Install core if requested or if dependencies require it
if (config.installCore || resolution.byModule.core) {
spinner.start('Installing BMAD core...');
await this.installCoreWithDependencies(bmadDir, resolution.byModule.core);
spinner.succeed('Core installed');
}
// Install modules with their dependencies
if (config.modules && config.modules.length > 0) {
for (const moduleName of config.modules) {
spinner.start(`Installing module: ${moduleName}...`);
await this.installModuleWithDependencies(moduleName, bmadDir, resolution.byModule[moduleName]);
spinner.succeed(`Module installed: ${moduleName}`);
}
// Install partial modules (only dependencies)
for (const [module, files] of Object.entries(resolution.byModule)) {
if (!config.modules.includes(module) && module !== 'core') {
2025-10-26 19:38:38 -05:00
const totalFiles =
files.agents.length +
files.tasks.length +
files.tools.length +
files.templates.length +
files.data.length +
files.other.length;
if (totalFiles > 0) {
spinner.start(`Installing ${module} dependencies...`);
await this.installPartialModule(module, bmadDir, files);
spinner.succeed(`${module} dependencies installed`);
}
}
}
}
// Generate clean config.yaml files for each installed module
spinner.start('Generating module configurations...');
await this.generateModuleConfigs(bmadDir, moduleConfigs);
spinner.succeed('Module configurations generated');
// Create agent configuration files
2025-10-02 21:45:59 -05:00
// Note: Legacy createAgentConfigs removed - using YAML customize system instead
// Customize templates are now created in processAgentFiles when building YAML agents
2025-09-30 21:20:13 -05:00
// Pre-register manifest files that will be created (except files-manifest.csv to avoid recursion)
const cfgDir = path.join(bmadDir, '_cfg');
this.installedFiles.push(
path.join(cfgDir, 'manifest.yaml'),
path.join(cfgDir, 'workflow-manifest.csv'),
path.join(cfgDir, 'agent-manifest.csv'),
path.join(cfgDir, 'task-manifest.csv'),
);
// Generate CSV manifests for workflows, agents, tasks AND ALL FILES with hashes BEFORE IDE setup
spinner.start('Generating workflow and agent manifests...');
const manifestGen = new ManifestGenerator();
2025-10-26 16:17:37 -05:00
// Include preserved modules (from quick update) in the manifest
const allModulesToList = config._preserveModules ? [...(config.modules || []), ...config._preserveModules] : config.modules || [];
const manifestStats = await manifestGen.generateManifests(bmadDir, config.modules || [], this.installedFiles, {
ides: config.ides || [],
2025-10-26 16:17:37 -05:00
preservedModules: config._preserveModules || [], // Scan these from installed bmad/ dir
});
2025-09-30 21:20:13 -05:00
spinner.succeed(
2025-10-26 19:38:38 -05:00
`Manifests generated: ${manifestStats.workflows} workflows, ${manifestStats.agents} agents, ${manifestStats.tasks} tasks, ${manifestStats.tools} tools, ${manifestStats.files} files`,
);
// Configure IDEs and copy documentation
if (!config.skipIde && config.ides && config.ides.length > 0) {
2025-10-26 19:38:38 -05:00
// Filter out any undefined/null values from the IDE list
const validIdes = config.ides.filter((ide) => ide && typeof ide === 'string');
2025-10-26 17:04:27 -05:00
2025-10-26 19:38:38 -05:00
if (validIdes.length === 0) {
console.log(chalk.yellow('⚠️ No valid IDEs selected. Skipping IDE configuration.'));
} else {
// Check if any IDE might need prompting (no pre-collected config)
const needsPrompting = validIdes.some((ide) => !ideConfigurations[ide]);
2025-10-26 19:38:38 -05:00
if (!needsPrompting) {
spinner.start('Configuring IDEs...');
}
2025-10-26 19:38:38 -05:00
// Temporarily suppress console output if not verbose
const originalLog = console.log;
if (!config.verbose) {
console.log = () => {};
2025-10-26 17:04:27 -05:00
}
2025-10-26 19:38:38 -05:00
for (const ide of validIdes) {
// Only show spinner if we have pre-collected config (no prompts expected)
if (ideConfigurations[ide] && !needsPrompting) {
spinner.text = `Configuring ${ide}...`;
} else if (!ideConfigurations[ide]) {
// Stop spinner before prompting
if (spinner.isSpinning) {
spinner.stop();
}
console.log(chalk.cyan(`\nConfiguring ${ide}...`));
}
2025-10-26 17:04:27 -05:00
2025-10-26 19:38:38 -05:00
// Pass pre-collected configuration to avoid re-prompting
await this.ideManager.setup(ide, projectDir, bmadDir, {
selectedModules: config.modules || [],
preCollectedConfig: ideConfigurations[ide] || null,
verbose: config.verbose,
});
2025-10-26 17:04:27 -05:00
2025-10-26 19:38:38 -05:00
// Save IDE configuration for future updates
if (ideConfigurations[ide] && !ideConfigurations[ide]._alreadyConfigured) {
await this.ideConfigManager.saveIdeConfig(bmadDir, ide, ideConfigurations[ide]);
}
// Restart spinner if we stopped it
if (!ideConfigurations[ide] && !spinner.isSpinning) {
spinner.start('Configuring IDEs...');
}
2025-10-26 17:04:27 -05:00
}
2025-10-26 19:38:38 -05:00
// Restore console.log
console.log = originalLog;
2025-10-26 19:38:38 -05:00
if (spinner.isSpinning) {
spinner.succeed(`Configured ${validIdes.length} IDE${validIdes.length > 1 ? 's' : ''}`);
} else {
console.log(chalk.green(`✓ Configured ${validIdes.length} IDE${validIdes.length > 1 ? 's' : ''}`));
}
2025-10-26 17:04:27 -05:00
}
2025-10-26 19:38:38 -05:00
// Copy IDE-specific documentation (only for valid IDEs)
const validIdesForDocs = (config.ides || []).filter((ide) => ide && typeof ide === 'string');
if (validIdesForDocs.length > 0) {
spinner.start('Copying IDE documentation...');
await this.copyIdeDocumentation(validIdesForDocs, bmadDir);
spinner.succeed('IDE documentation copied');
}
}
// Run module-specific installers after IDE setup
spinner.start('Running module-specific installers...');
// Run core module installer if core was installed
if (config.installCore || resolution.byModule.core) {
spinner.text = 'Running core module installer...';
await this.moduleManager.runModuleInstaller('core', bmadDir, {
installedIDEs: config.ides || [],
moduleConfig: moduleConfigs.core || {},
logger: {
log: (msg) => console.log(msg),
error: (msg) => console.error(msg),
warn: (msg) => console.warn(msg),
},
});
}
// Run installers for user-selected modules
if (config.modules && config.modules.length > 0) {
for (const moduleName of config.modules) {
spinner.text = `Running ${moduleName} module installer...`;
// Pass installed IDEs and module config to module installer
await this.moduleManager.runModuleInstaller(moduleName, bmadDir, {
installedIDEs: config.ides || [],
moduleConfig: moduleConfigs[moduleName] || {},
logger: {
log: (msg) => console.log(msg),
error: (msg) => console.error(msg),
warn: (msg) => console.warn(msg),
},
});
}
}
spinner.succeed('Module-specific installers completed');
2025-10-03 11:54:32 -05:00
// Note: Manifest files are already created by ManifestGenerator above
// No need to create legacy manifest.csv anymore
// If this was an update, restore custom files
let customFiles = [];
2025-09-30 21:20:13 -05:00
let modifiedFiles = [];
if (config._isUpdate) {
if (config._customFiles && config._customFiles.length > 0) {
spinner.start(`Restoring ${config._customFiles.length} custom files...`);
for (const originalPath of config._customFiles) {
const relativePath = path.relative(bmadDir, originalPath);
const backupPath = path.join(config._tempBackupDir, relativePath);
if (await fs.pathExists(backupPath)) {
await fs.ensureDir(path.dirname(originalPath));
await fs.copy(backupPath, originalPath, { overwrite: true });
}
}
2025-09-30 21:20:13 -05:00
// Clean up temp backup
if (config._tempBackupDir && (await fs.pathExists(config._tempBackupDir))) {
await fs.remove(config._tempBackupDir);
}
2025-09-30 21:20:13 -05:00
spinner.succeed(`Restored ${config._customFiles.length} custom files`);
customFiles = config._customFiles;
}
2025-09-30 21:20:13 -05:00
if (config._modifiedFiles && config._modifiedFiles.length > 0) {
modifiedFiles = config._modifiedFiles;
// Restore modified files as .bak files
if (config._tempModifiedBackupDir && (await fs.pathExists(config._tempModifiedBackupDir))) {
spinner.start(`Restoring ${modifiedFiles.length} modified files as .bak...`);
for (const modifiedFile of modifiedFiles) {
const relativePath = path.relative(bmadDir, modifiedFile.path);
const tempBackupPath = path.join(config._tempModifiedBackupDir, relativePath);
const bakPath = modifiedFile.path + '.bak';
if (await fs.pathExists(tempBackupPath)) {
await fs.ensureDir(path.dirname(bakPath));
await fs.copy(tempBackupPath, bakPath, { overwrite: true });
}
}
// Clean up temp backup
await fs.remove(config._tempModifiedBackupDir);
spinner.succeed(`Restored ${modifiedFiles.length} modified files as .bak`);
}
}
}
spinner.stop();
2025-09-30 21:20:13 -05:00
// Report custom and modified files if any were found
if (customFiles.length > 0) {
console.log(chalk.cyan(`\n📁 Custom files preserved: ${customFiles.length}`));
console.log(chalk.dim('The following custom files were found and restored:\n'));
for (const file of customFiles) {
console.log(chalk.dim(` - ${path.relative(bmadDir, file)}`));
}
console.log('');
}
2025-09-30 21:20:13 -05:00
if (modifiedFiles.length > 0) {
console.log(chalk.yellow(`\n⚠️ Modified files detected: ${modifiedFiles.length}`));
console.log(chalk.dim('The following files were modified and backed up with .bak extension:\n'));
for (const file of modifiedFiles) {
console.log(chalk.dim(` - ${file.relativePath}${file.relativePath}.bak`));
}
console.log(chalk.dim('\nThese files have been updated with the new version.'));
console.log(chalk.dim('Review the .bak files to see your changes and merge if needed.\n'));
}
feat: Complete BMAD agent creation system with install tooling, references, and field guidance ## Overview This commit represents a complete overhaul of the BMAD agent creation system, establishing clear standards for agent development, installation workflows, and persona design. The changes span documentation, tooling, reference implementations, and field-specific guidance. ## Key Components ### 1. Agent Installation Infrastructure **New CLI Command: `agent-install`** - Interactive agent installation with persona customization - Supports Simple (single YAML), Expert (sidecar files), and Module agents - Template variable processing with Handlebars-style syntax - Automatic compilation from YAML to XML (.md) format - Manifest tracking and IDE integration (Claude Code, Cursor, Windsurf, etc.) - Source preservation in `_cfg/custom/agents/` for reinstallation **Files Created:** - `tools/cli/commands/agent-install.js` - Main CLI command - `tools/cli/lib/agent/compiler.js` - YAML to XML compilation engine - `tools/cli/lib/agent/installer.js` - Installation orchestration - `tools/cli/lib/agent/template-engine.js` - Handlebars template processing **Compiler Features:** - Auto-injects frontmatter, activation, handlers, help/exit menu items - Smart handler inclusion (only includes action/workflow/exec/tmpl handlers actually used) - Proper XML escaping and formatting - Persona name customization (e.g., "Fred the Commit Poet") ### 2. Documentation Overhaul **Deleted Bloated/Outdated Docs (2,651 lines removed):** - Old verbose architecture docs - Redundant pattern files - Outdated workflow guides **Created Focused, Type-Specific Docs:** - `src/modules/bmb/docs/understanding-agent-types.md` - Architecture vs capability distinction - `src/modules/bmb/docs/simple-agent-architecture.md` - Self-contained agents - `src/modules/bmb/docs/expert-agent-architecture.md` - Agents with sidecar files - `src/modules/bmb/docs/module-agent-architecture.md` - Workflow-integrated agents - `src/modules/bmb/docs/agent-compilation.md` - YAML → XML process - `src/modules/bmb/docs/agent-menu-patterns.md` - Menu design patterns - `src/modules/bmb/docs/index.md` - Documentation hub **Net Result:** ~1,930 line reduction while adding MORE value through focused content ### 3. Create-Agent Workflow Enhancements **Critical Persona Field Guidance Added to Step 4:** Explains how the LLM interprets each persona field when the agent activates: - **role** → "What knowledge, skills, and capabilities do I possess?" - **identity** → "What background, experience, and context shape my responses?" - **communication_style** → "What verbal patterns, word choice, quirks, and phrasing do I use?" - **principles** → "What beliefs and operating philosophy drive my choices?" **Key Insight:** `communication_style` should ONLY describe HOW the agent talks, not restate role/identity/principles. The `communication-presets.csv` provides 60 pure communication styles with NO role/identity/principles mixed in. **Files Updated:** - `src/modules/bmb/workflows/create-agent/instructions.md` - Added persona field interpretation guide - `src/modules/bmb/workflows/create-agent/brainstorm-context.md` - Refined to 137 lines - `src/modules/bmb/workflows/create-agent/communication-presets.csv` - 60 styles across 13 categories ### 4. Reference Agent Cleanup **Removed install_config Personality Bloat:** Understanding: Future installer will handle personality customization, so stripped all personality toggles from reference agents. **commit-poet.agent.yaml** (Simple Agent): - BEFORE: 36 personality combinations (3 enthusiasm × 3 depths × 4 styles) = decision fatigue - AFTER: Single concise persona with pure communication style - Changed from verbose conditionals to: "Poetic drama and flair with every turn of a phrase. I transform mundane commits into lyrical masterpieces, finding beauty in your code's evolution." - Reduction: 248 lines → 153 lines (38% reduction) **journal-keeper.agent.yaml** (Expert Agent): - Stripped install_config, simplified communication_style - Shows proper Expert agent structure with sidecar files **security-engineer.agent.yaml & trend-analyst.agent.yaml** (Module Agents): - Added header comments explaining WHY Module Agent (design intent, not just location) - Clarified: Module agents are designed FOR ecosystem integration, not capability-limited **Files Updated:** - `src/modules/bmb/reference/agents/simple-examples/commit-poet.agent.yaml` - `src/modules/bmb/reference/agents/expert-examples/journal-keeper/journal-keeper.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/security-engineer.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/trend-analyst.agent.yaml` ### 5. BMM Agent Voice Enhancement **Gave all 9 BMM agents distinct, memorable communication voices:** **Mary (analyst)** - The favorite! Changed from generic "systematic and probing" to: "Treats analysis like a treasure hunt - excited by every clue, thrilled when patterns emerge. Asks questions that spark 'aha!' moments while structuring insights with precision." **Other Notable Voices:** - **John (pm):** "Asks 'WHY?' relentlessly like a detective on a case. Direct and data-sharp, cuts through fluff to what actually matters." - **Winston (architect):** "Speaks in calm, pragmatic tones, balancing 'what could be' with 'what should be.' Champions boring technology that actually works." - **Amelia (dev):** "Ultra-succinct. Speaks in file paths and AC IDs - every statement citable. No fluff, all precision." - **Bob (sm):** "Crisp and checklist-driven. Every word has a purpose, every requirement crystal clear. Zero tolerance for ambiguity." - **Sally (ux-designer):** "Paints pictures with words, telling user stories that make you FEEL the problem. Empathetic advocate with creative storytelling flair." **Pattern Applied:** Moved behaviors from communication_style to principles, keeping communication_style as PURE verbal patterns. **Files Updated:** - `src/modules/bmm/agents/analyst.agent.yaml` - `src/modules/bmm/agents/pm.agent.yaml` - `src/modules/bmm/agents/architect.agent.yaml` - `src/modules/bmm/agents/dev.agent.yaml` - `src/modules/bmm/agents/sm.agent.yaml` - `src/modules/bmm/agents/tea.agent.yaml` - `src/modules/bmm/agents/tech-writer.agent.yaml` - `src/modules/bmm/agents/ux-designer.agent.yaml` - `src/modules/bmm/agents/frame-expert.agent.yaml` ### 6. Linting Fixes **ESLint Compliance:** - Replaced all `'utf-8'` with `'utf8'` (unicorn/text-encoding-identifier-case) - Changed `variables.hasOwnProperty(varName)` to `Object.hasOwn(variables, varName)` (unicorn/prefer-object-has-own) - Replaced `JSON.parse(JSON.stringify(...))` with `structuredClone(...)` (unicorn/prefer-structured-clone) - Fixed empty YAML mapping values in sample files **Files Fixed:** - 7 JavaScript files across agent tooling (compiler, installer, commands, IDE integration) - 1 YAML sample file ## Architecture Decisions ### Agent Types Are About Architecture, Not Capability - **Simple:** Self-contained in single YAML (NOT limited in capability) - **Expert:** Includes sidecar files (templates, docs, etc.) - **Module:** Designed for BMAD ecosystem integration (workflows, cross-agent coordination) ### Persona Field Separation Critical for LLM Interpretation The LLM needs distinct fields to understand its role: - Mixing role/identity/principles into communication_style confuses the persona - Pure communication styles (from communication-presets.csv) have ZERO role/identity/principles content - Example DON'T: "Experienced analyst who uses systematic approaches..." (mixing identity + style) - Example DO: "Systematic and probing. Structures findings hierarchically." (pure style) ### Install-Time vs Runtime Configuration - Template variables ({{var}}) resolve at compile-time - Runtime variables ({user_name}, {bmad_folder}) resolve when agent activates - Future installer will handle personality customization, so agents should ship with single default persona ## Testing - All linting passes (ESLint with max-warnings=0) - Agent compilation tested with commit-poet, journal-keeper examples - Install workflow validated with Simple and Expert agent types - Manifest tracking and IDE integration verified ## Impact This establishes BMAD as having a complete, production-ready agent creation and installation system with: - Clear documentation for all agent types - Automated compilation and installation - Strong persona design guidance - Reference implementations showing best practices - Distinct, memorable agent voices throughout BMM module Co-Authored-By: BMad Builder <builder@bmad.dev> Co-Authored-By: Mary the Analyst <analyst@bmad.dev> Co-Authored-By: Paige the Tech Writer <tech-writer@bmad.dev>
2025-11-17 22:25:15 -06:00
// Reinstall custom agents from _cfg/custom/agents/ sources
const customAgentResults = await this.reinstallCustomAgents(projectDir, bmadDir);
if (customAgentResults.count > 0) {
console.log(chalk.green(`\n✓ Reinstalled ${customAgentResults.count} custom agent${customAgentResults.count > 1 ? 's' : ''}`));
for (const agent of customAgentResults.agents) {
console.log(chalk.dim(` - ${agent}`));
}
}
// Display completion message
const { UI } = require('../../../lib/ui');
const ui = new UI();
ui.showInstallSummary({
path: bmadDir,
modules: config.modules,
ides: config.ides,
customFiles: customFiles.length > 0 ? customFiles : undefined,
feat(installer): Enhanced TTS injection summary with tracking and documentation (#1037) ## Summary - Track all files with TTS injection applied during installation - Display informative summary explaining what TTS injection does - Show backup location and restore command for recovery ## What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ## Changes - **installer.js**: Track files in `processAgentFiles()`, `buildStandaloneAgents()`, and `rebuildAgentFiles()` when TTS markers are processed - **compiler.js**: Add TTS injection support for custom agent compilation - **ui.js**: Enhanced installation summary showing: - Explanation of what TTS injection is with example - List of all files with TTS injection applied (grouped by type) - Backup location (~/.bmad-tts-backups/) - Restore command for recovery ## Example Output ``` ═══════════════════════════════════════════════════ AgentVibes TTS Injection Summary ═══════════════════════════════════════════════════ What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ✅ TTS injection applied to 11 file(s): Party Mode (multi-agent conversations): • .bmad/core/workflows/party-mode/instructions.md Agent TTS (individual agent voices): • .bmad/bmm/agents/analyst.md • .bmad/bmm/agents/architect.md ... Backups & Recovery: Pre-injection backups are stored in: ~/.bmad-tts-backups/ To restore original files (removes TTS instructions): bmad-tts-injector.sh --restore /path/to/.bmad ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Claude <noreply@anthropic.com>
2025-12-05 17:54:03 -07:00
ttsInjectedFiles: this.enableAgentVibes && this.ttsInjectedFiles.length > 0 ? this.ttsInjectedFiles : undefined,
agentVibesEnabled: this.enableAgentVibes || false,
});
// Offer cleanup for legacy files (only for updates, not fresh installs, and only if not skipped)
if (!config.skipCleanup && config._isUpdate) {
try {
const cleanupResult = await this.performCleanup(bmadDir, false);
if (cleanupResult.deleted > 0) {
console.log(chalk.green(`\n✓ Cleaned up ${cleanupResult.deleted} legacy file${cleanupResult.deleted > 1 ? 's' : ''}`));
}
if (cleanupResult.retained > 0) {
console.log(chalk.dim(`Run 'bmad cleanup' anytime to manage retained files`));
}
} catch (cleanupError) {
// Don't fail the installation for cleanup errors
console.log(chalk.yellow(`\n⚠️ Cleanup warning: ${cleanupError.message}`));
console.log(chalk.dim('Run "bmad cleanup" to manually clean up legacy files'));
}
}
Add Text-to-Speech Integration via TTS_INJECTION System (#934) * feat: Add provider-agnostic TTS integration via injection point system Implements comprehensive Text-to-Speech integration for BMAD agents using a generic TTS_INJECTION marker system. When AgentVibes (or any compatible TTS provider) is installed, all BMAD agents can speak their responses with unique AI voices. ## Key Features **Provider-Agnostic Architecture** - Uses generic `TTS_INJECTION` markers instead of vendor-specific naming - Future-proof for multiple TTS providers beyond AgentVibes - Clean separation - BMAD stays TTS-agnostic, providers handle injection **Installation Flow** - BMAD → AgentVibes: TTS instructions injected when AgentVibes detects existing BMAD installation - AgentVibes → BMAD: TTS instructions injected during BMAD installation when AgentVibes detected - User must manually create voice assignment file when AgentVibes installs first (documented limitation) **Party Mode Voice Support** - Each agent speaks with unique assigned voice in multi-agent discussions - PM, Architect, Developer, Analyst, UX Designer, etc. - all with distinct voices **Zero Breaking Changes** - Fully backward compatible - works without any TTS provider - `TTS_INJECTION` markers are benign HTML comments if not processed - No changes to existing agent behavior or non-TTS workflows ## Implementation Details **Files Modified:** - `tools/cli/installers/lib/core/installer.js` - TTS injection processing logic - `tools/cli/lib/ui.js` - AgentVibes detection and installation prompts - `tools/cli/commands/install.js` - Post-install guidance for AgentVibes setup - `src/utility/models/fragments/activation-rules.xml` - TTS_INJECTION marker for agents - `src/core/workflows/party-mode/instructions.md` - TTS_INJECTION marker for party mode **Injection Point System:** ```xml <rules> - ALWAYS communicate in {communication_language} <!-- TTS_INJECTION:agent-tts --> - Stay in character until exit selected </rules> ``` When AgentVibes is detected, the installer replaces this marker with: ``` - When responding to user messages, speak your responses using TTS: Call: `.claude/hooks/bmad-speak.sh '{agent-id}' '{response-text}'` after each response IMPORTANT: Use single quotes - do NOT escape special characters like ! or $ ``` **Special Character Handling:** - Explicit guidance to use single quotes without escaping - Prevents "backslash exclamation" artifacts in speech **User Experience:** ``` User: "How should we architect this feature?" Architect: [Text response] + 🔊 [Professional voice explains architecture] ``` Party Mode: ``` PM (John): "I'll focus on user value..." 🔊 [Male professional voice] UX Designer (Sara): "From a user perspective..." 🔊 [Female voice] Architect (Marcus): "The technical approach..." 🔊 [Male technical voice] ``` ## Testing **Unit Tests:** ✅ 62/62 passing - 49/49 schema validation tests - 13/13 installation component tests **Integration Testing:** - ✅ BMAD → AgentVibes (automatic injection) - ✅ AgentVibes → BMAD (automatic injection) - ✅ No TTS provider (markers remain as comments) ## Documentation Comprehensive testing guide created with: - Both installation scenario walkthroughs - Verification commands and expected outputs - Troubleshooting guidance ## Known Limitations **AgentVibes → BMAD Installation Order:** When AgentVibes installs first, voice assignment file must be created manually: ```bash mkdir -p .bmad/_cfg cat > .bmad/_cfg/agent-voice-map.csv << 'EOF' agent_id,voice_name pm,en_US-ryan-high architect,en_US-danny-low dev,en_US-joe-medium EOF ``` This limitation exists to prevent false legacy v4 detection warnings from BMAD installer. **Recommended:** Install BMAD first, then AgentVibes for automatic voice assignment. ## Related Work **Companion Implementation:** - Repository: paulpreibisch/AgentVibes - Commits: 6 commits implementing injection processing and voice routing - Features: Retroactive injection, file path extraction, escape stripping **GitHub Issues:** - paulpreibisch/AgentVibes#36 - BMAD agent ID support ## Breaking Changes None. Feature is opt-in and requires separate TTS provider installation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Enforce project hooks over global hooks in party mode before, claude would sometimes favor global agent vibes hooks over project specific * feat: Automate AgentVibes installer invocation after BMAD install Instead of showing manual installation instructions, the installer now: - Prompts "Press Enter to start AgentVibes installer..." - Automatically runs npx agentvibes@latest install - Handles errors gracefully with fallback instructions This provides a seamless installation flow matching the test script's interactive approach. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * docs: Add automated testing script and guide for PR #934 Added comprehensive testing tools for AgentVibes party mode integration: - test-bmad-pr.sh: Fully automated installation and verification script - Interactive mode selection (official PR or custom fork) - Automatic BMAD CLI setup and linking - AgentVibes installation with guided prompts - Built-in verification checks for voice maps and hooks - Saved configuration for quick re-testing - TESTING.md: Complete testing documentation - Quick start with one-line npx command - Manual installation alternative - Troubleshooting guide - Cleanup instructions Testers can now run a single command to test the full AgentVibes integration without needing to understand the complex setup process. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Add shell: true to npx execSync to prevent permission denied error The execSync call for 'npx agentvibes@latest install' was failing with 'Permission denied' because the shell was trying to execute 'agentvibes@latest' directly instead of passing it as an argument to npx. Adding shell: true ensures the command runs in a proper shell context where npx can correctly interpret the @latest version syntax. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Remove duplicate AgentVibes installation step from test script The test script was calling AgentVibes installer twice: 1. BMAD installer now automatically runs AgentVibes (new feature) 2. Test script had a separate Step 6 that also ran AgentVibes This caused the installer to run twice, with the second call failing because it was already installed. Changes: - Removed redundant Step 6 (AgentVibes installation) - Updated Step 5 to indicate it includes AgentVibes - Updated step numbers from 7 to 6 throughout - Added guidance that AgentVibes runs automatically Now the flow is cleaner: BMAD installer handles everything! 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Address bmadcode review - preserve variables and move TTS logic to injection Fixes requested changes from PR review: 1. Preserve {bmad_folder} variable placeholder - Changed: {project_root}/.bmad/core/tasks/workflow.xml - To: {project_root}/{bmad_folder}/core/tasks/workflow.xml - Allows users to choose custom BMAD folder names during installation 2. Move TTS-specific hook guidance to injection system - Removed hardcoded hook enforcement from source files - Added hook guidance to processTTSInjectionPoints() in installer.js - Now only appears when AgentVibes is installed (via TTS_INJECTION) 3. Maintain TTS-agnostic source architecture - Source files remain clean of TTS-specific instructions - TTS details injected at install-time only when needed - Preserves provider-agnostic design principle Changes made: - src/core/workflows/party-mode/instructions.md - Reverted .bmad to {bmad_folder} variable - Replaced hardcoded hook guidance with <!-- TTS_INJECTION:party-mode --> - Removed <note> about play-tts.sh hook location - tools/cli/installers/lib/core/installer.js - Added hook enforcement to party-mode injection replacement - Guidance now injected only when enableAgentVibes is true Addresses review comments from bmadcode: - "needs to remain the variable. it will get set in the file at the install destination." - "items like this we will need to inject if user is using claude and TTS" 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com> * fix: Change 'claude-code' to 'claude' in test script instructions The correct command to start Claude is 'claude', not 'claude-code'. Updated line 362-363 in test-bmad-pr.sh to show the correct command. * fix: Remove npm link from test script to avoid global namespace pollution - Removed 'npm link' command that was installing BMAD globally - Changed 'bmad install' to direct node execution using local clone - Updated success message to reflect no global installation This keeps testing fully isolated and prevents conflicts with: - Existing BMAD installations - Future official BMAD installs - Orphaned symlinks when test directory is deleted The test script now runs completely self-contained without modifying the user's global npm environment. --------- Co-authored-by: Claude Code <claude@anthropic.com> Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Brian <bmadcode@gmail.com>
2025-11-26 08:51:57 -07:00
return {
success: true,
path: bmadDir,
modules: config.modules,
ides: config.ides,
needsAgentVibes: this.enableAgentVibes && !config.agentVibesInstalled,
projectDir: projectDir,
};
} catch (error) {
spinner.fail('Installation failed');
throw error;
}
}
/**
* Update existing installation
*/
async update(config) {
const spinner = ora('Checking installation...').start();
try {
const projectDir = path.resolve(config.directory);
const bmadDir = await this.findBmadDir(projectDir);
const existingInstall = await this.detector.detect(bmadDir);
if (!existingInstall.installed) {
spinner.fail('No BMAD installation found');
throw new Error(`No BMAD installation found at ${bmadDir}`);
}
spinner.text = 'Analyzing update requirements...';
// Compare versions and determine what needs updating
const currentVersion = existingInstall.version;
const newVersion = require(path.join(getProjectRoot(), 'package.json')).version;
if (config.dryRun) {
spinner.stop();
console.log(chalk.cyan('\n🔍 Update Preview (Dry Run)\n'));
console.log(chalk.bold('Current version:'), currentVersion);
console.log(chalk.bold('New version:'), newVersion);
console.log(chalk.bold('Core:'), existingInstall.hasCore ? 'Will be updated' : 'Not installed');
if (existingInstall.modules.length > 0) {
console.log(chalk.bold('\nModules to update:'));
for (const mod of existingInstall.modules) {
console.log(` - ${mod.id}`);
}
}
return;
}
// Perform actual update
if (existingInstall.hasCore) {
spinner.text = 'Updating core...';
await this.updateCore(bmadDir, config.force);
}
for (const module of existingInstall.modules) {
spinner.text = `Updating module: ${module.id}...`;
await this.moduleManager.update(module.id, bmadDir, config.force);
}
// Update manifest
spinner.text = 'Updating manifest...';
await this.manifest.update(bmadDir, {
version: newVersion,
updateDate: new Date().toISOString(),
});
spinner.succeed('Update complete');
return { success: true };
} catch (error) {
spinner.fail('Update failed');
throw error;
}
}
/**
* Get installation status
*/
async getStatus(directory) {
const projectDir = path.resolve(directory);
const bmadDir = await this.findBmadDir(projectDir);
return await this.detector.detect(bmadDir);
}
/**
* Get available modules
*/
async getAvailableModules() {
return await this.moduleManager.listAvailable();
}
/**
* Uninstall BMAD
*/
async uninstall(directory) {
const projectDir = path.resolve(directory);
const bmadDir = await this.findBmadDir(projectDir);
if (await fs.pathExists(bmadDir)) {
await fs.remove(bmadDir);
}
// Clean up IDE configurations
await this.ideManager.cleanup(projectDir);
return { success: true };
}
/**
* Private: Create directory structure
*/
async createDirectoryStructure(bmadDir) {
await fs.ensureDir(bmadDir);
await fs.ensureDir(path.join(bmadDir, '_cfg'));
await fs.ensureDir(path.join(bmadDir, '_cfg', 'agents'));
}
/**
* Generate clean config.yaml files for each installed module
* @param {string} bmadDir - BMAD installation directory
* @param {Object} moduleConfigs - Collected configuration values
*/
async generateModuleConfigs(bmadDir, moduleConfigs) {
const yaml = require('js-yaml');
// Extract core config values to share with other modules
const coreConfig = moduleConfigs.core || {};
// Get all installed module directories
const entries = await fs.readdir(bmadDir, { withFileTypes: true });
const installedModules = entries
.filter((entry) => entry.isDirectory() && entry.name !== '_cfg' && entry.name !== 'docs')
.map((entry) => entry.name);
// Generate config.yaml for each installed module
for (const moduleName of installedModules) {
const modulePath = path.join(bmadDir, moduleName);
// Get module-specific config or use empty object if none
const config = moduleConfigs[moduleName] || {};
if (await fs.pathExists(modulePath)) {
const configPath = path.join(modulePath, 'config.yaml');
// Create header
const packageJson = require(path.join(getProjectRoot(), 'package.json'));
const header = `# ${moduleName.toUpperCase()} Module Configuration
# Generated by BMAD installer
# Version: ${packageJson.version}
# Date: ${new Date().toISOString()}
`;
// For non-core modules, add core config values directly
let finalConfig = { ...config };
let coreSection = '';
if (moduleName !== 'core' && coreConfig && Object.keys(coreConfig).length > 0) {
// Add core values directly to the module config
// These will be available for reference in the module
finalConfig = {
...config,
...coreConfig, // Spread core config values directly into the module config
};
// Create a comment section to identify core values
coreSection = '\n# Core Configuration Values\n';
}
// Convert config to YAML
let yamlContent = yaml.dump(finalConfig, {
indent: 2,
lineWidth: -1,
noRefs: true,
sortKeys: false,
});
// If we have core values, reorganize the YAML to group them with their comment
if (coreSection && moduleName !== 'core') {
// Split the YAML into lines
const lines = yamlContent.split('\n');
const moduleConfigLines = [];
const coreConfigLines = [];
// Separate module-specific and core config lines
for (const line of lines) {
const key = line.split(':')[0].trim();
if (Object.prototype.hasOwnProperty.call(coreConfig, key)) {
coreConfigLines.push(line);
} else {
moduleConfigLines.push(line);
}
}
// Rebuild YAML with module config first, then core config with comment
yamlContent = moduleConfigLines.join('\n');
if (coreConfigLines.length > 0) {
yamlContent += coreSection + coreConfigLines.join('\n');
}
}
// Write the clean config file with POSIX-compliant final newline
const content = header + yamlContent;
await fs.writeFile(configPath, content.endsWith('\n') ? content : content + '\n', 'utf8');
2025-09-30 21:20:13 -05:00
// Track the config file in installedFiles
this.installedFiles.push(configPath);
}
}
}
/**
* Install core with resolved dependencies
* @param {string} bmadDir - BMAD installation directory
* @param {Object} coreFiles - Core files to install
*/
async installCoreWithDependencies(bmadDir, coreFiles) {
const sourcePath = getModulePath('core');
const targetPath = path.join(bmadDir, 'core');
// Install full core
await this.installCore(bmadDir);
// If there are specific dependency files, ensure they're included
if (coreFiles) {
// Already handled by installCore for core module
}
}
/**
* Install module with resolved dependencies
* @param {string} moduleName - Module name
* @param {string} bmadDir - BMAD installation directory
* @param {Object} moduleFiles - Module files to install
*/
async installModuleWithDependencies(moduleName, bmadDir, moduleFiles) {
// Get module configuration for conditional installation
const moduleConfig = this.configCollector.collectedConfig[moduleName] || {};
// Use existing module manager for full installation with file tracking
// Note: Module-specific installers are called separately after IDE setup
await this.moduleManager.install(
moduleName,
bmadDir,
(filePath) => {
this.installedFiles.push(filePath);
},
{
skipModuleInstaller: true, // We'll run it later after IDE setup
moduleConfig: moduleConfig, // Pass module config for conditional filtering
},
);
2025-10-02 21:45:59 -05:00
// Process agent files to build YAML agents and create customize templates
const modulePath = path.join(bmadDir, moduleName);
await this.processAgentFiles(modulePath, moduleName);
// Dependencies are already included in full module install
}
/**
* Install partial module (only dependencies needed by other modules)
*/
async installPartialModule(moduleName, bmadDir, files) {
const sourceBase = getModulePath(moduleName);
const targetBase = path.join(bmadDir, moduleName);
// Create module directory
await fs.ensureDir(targetBase);
// Copy only the required dependency files
if (files.agents && files.agents.length > 0) {
const agentsDir = path.join(targetBase, 'agents');
await fs.ensureDir(agentsDir);
for (const agentPath of files.agents) {
const fileName = path.basename(agentPath);
const sourcePath = path.join(sourceBase, 'agents', fileName);
const targetPath = path.join(agentsDir, fileName);
if (await fs.pathExists(sourcePath)) {
await this.copyFileWithPlaceholderReplacement(sourcePath, targetPath, this.bmadFolderName || 'bmad');
this.installedFiles.push(targetPath);
}
}
}
if (files.tasks && files.tasks.length > 0) {
const tasksDir = path.join(targetBase, 'tasks');
await fs.ensureDir(tasksDir);
for (const taskPath of files.tasks) {
const fileName = path.basename(taskPath);
const sourcePath = path.join(sourceBase, 'tasks', fileName);
const targetPath = path.join(tasksDir, fileName);
if (await fs.pathExists(sourcePath)) {
await this.copyFileWithPlaceholderReplacement(sourcePath, targetPath, this.bmadFolderName || 'bmad');
this.installedFiles.push(targetPath);
}
}
}
2025-10-26 19:38:38 -05:00
if (files.tools && files.tools.length > 0) {
const toolsDir = path.join(targetBase, 'tools');
await fs.ensureDir(toolsDir);
for (const toolPath of files.tools) {
const fileName = path.basename(toolPath);
const sourcePath = path.join(sourceBase, 'tools', fileName);
const targetPath = path.join(toolsDir, fileName);
if (await fs.pathExists(sourcePath)) {
await this.copyFileWithPlaceholderReplacement(sourcePath, targetPath, this.bmadFolderName || 'bmad');
2025-10-26 19:38:38 -05:00
this.installedFiles.push(targetPath);
}
}
}
if (files.templates && files.templates.length > 0) {
const templatesDir = path.join(targetBase, 'templates');
await fs.ensureDir(templatesDir);
for (const templatePath of files.templates) {
const fileName = path.basename(templatePath);
const sourcePath = path.join(sourceBase, 'templates', fileName);
const targetPath = path.join(templatesDir, fileName);
if (await fs.pathExists(sourcePath)) {
await this.copyFileWithPlaceholderReplacement(sourcePath, targetPath, this.bmadFolderName || 'bmad');
this.installedFiles.push(targetPath);
}
}
}
if (files.data && files.data.length > 0) {
for (const dataPath of files.data) {
// Preserve directory structure for data files
const relative = path.relative(sourceBase, dataPath);
const targetPath = path.join(targetBase, relative);
await fs.ensureDir(path.dirname(targetPath));
if (await fs.pathExists(dataPath)) {
await this.copyFileWithPlaceholderReplacement(dataPath, targetPath, this.bmadFolderName || 'bmad');
this.installedFiles.push(targetPath);
}
}
}
// Create a marker file to indicate this is a partial installation
const markerPath = path.join(targetBase, '.partial');
await fs.writeFile(
markerPath,
`This module contains only dependencies required by other modules.\nInstalled: ${new Date().toISOString()}\n`,
);
}
/**
* Private: Install core
* @param {string} bmadDir - BMAD installation directory
*/
async installCore(bmadDir) {
const sourcePath = getModulePath('core');
const targetPath = path.join(bmadDir, 'core');
// Copy core files with filtering for localskip agents
await this.copyDirectoryWithFiltering(sourcePath, targetPath);
// Process agent files to inject activation block
await this.processAgentFiles(targetPath, 'core');
}
/**
* Copy directory with filtering for localskip agents
* @param {string} sourcePath - Source directory path
* @param {string} targetPath - Target directory path
*/
async copyDirectoryWithFiltering(sourcePath, targetPath) {
// Get all files in source directory
const files = await this.getFileList(sourcePath);
for (const file of files) {
// Skip config.yaml templates - we'll generate clean ones with actual values
if (file === 'config.yaml' || file.endsWith('/config.yaml')) {
continue;
}
const sourceFile = path.join(sourcePath, file);
const targetFile = path.join(targetPath, file);
// Check if this is an agent file
if (file.includes('agents/') && file.endsWith('.md')) {
// Read the file to check for localskip
const content = await fs.readFile(sourceFile, 'utf8');
// Check for localskip="true" in the agent tag
const agentMatch = content.match(/<agent[^>]*\slocalskip="true"[^>]*>/);
if (agentMatch) {
console.log(chalk.dim(` Skipping web-only agent: ${path.basename(file)}`));
continue; // Skip this agent
}
}
// Copy the file with placeholder replacement
await this.copyFileWithPlaceholderReplacement(sourceFile, targetFile, this.bmadFolderName || 'bmad');
// Track the installed file
this.installedFiles.push(targetFile);
}
}
/**
* Get list of all files in a directory recursively
* @param {string} dir - Directory path
* @param {string} baseDir - Base directory for relative paths
* @returns {Array} List of relative file paths
*/
async getFileList(dir, baseDir = dir) {
const files = [];
const entries = await fs.readdir(dir, { withFileTypes: true });
for (const entry of entries) {
const fullPath = path.join(dir, entry.name);
if (entry.isDirectory()) {
// Skip _module-installer directories
if (entry.name === '_module-installer') {
continue;
}
const subFiles = await this.getFileList(fullPath, baseDir);
files.push(...subFiles);
} else {
files.push(path.relative(baseDir, fullPath));
}
}
return files;
}
/**
2025-10-02 21:45:59 -05:00
* Process agent files to build YAML agents and inject activation blocks
* @param {string} modulePath - Path to module in bmad/ installation
* @param {string} moduleName - Module name
*/
async processAgentFiles(modulePath, moduleName) {
const agentsPath = path.join(modulePath, 'agents');
// Check if agents directory exists
if (!(await fs.pathExists(agentsPath))) {
return; // No agents to process
}
2025-10-02 21:45:59 -05:00
// Determine project directory (parent of bmad/ directory)
const bmadDir = path.dirname(modulePath);
const projectDir = path.dirname(bmadDir);
const cfgAgentsDir = path.join(bmadDir, '_cfg', 'agents');
// Ensure _cfg/agents directory exists
await fs.ensureDir(cfgAgentsDir);
// Get all agent files
const agentFiles = await fs.readdir(agentsPath);
for (const agentFile of agentFiles) {
2025-10-02 21:45:59 -05:00
// Handle YAML agents - build them to .md
if (agentFile.endsWith('.agent.yaml')) {
const agentName = agentFile.replace('.agent.yaml', '');
const yamlPath = path.join(agentsPath, agentFile);
const mdPath = path.join(agentsPath, `${agentName}.md`);
const customizePath = path.join(cfgAgentsDir, `${moduleName}-${agentName}.customize.yaml`);
// Create customize template if it doesn't exist
if (!(await fs.pathExists(customizePath))) {
const genericTemplatePath = getSourcePath('utility', 'templates', 'agent.customize.template.yaml');
if (await fs.pathExists(genericTemplatePath)) {
await this.copyFileWithPlaceholderReplacement(genericTemplatePath, customizePath, this.bmadFolderName || 'bmad');
2025-10-02 21:45:59 -05:00
console.log(chalk.dim(` Created customize: ${moduleName}-${agentName}.customize.yaml`));
}
}
// Build YAML + customize to .md
const customizeExists = await fs.pathExists(customizePath);
feat(installer): Enhanced TTS injection summary with tracking and documentation (#1037) ## Summary - Track all files with TTS injection applied during installation - Display informative summary explaining what TTS injection does - Show backup location and restore command for recovery ## What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ## Changes - **installer.js**: Track files in `processAgentFiles()`, `buildStandaloneAgents()`, and `rebuildAgentFiles()` when TTS markers are processed - **compiler.js**: Add TTS injection support for custom agent compilation - **ui.js**: Enhanced installation summary showing: - Explanation of what TTS injection is with example - List of all files with TTS injection applied (grouped by type) - Backup location (~/.bmad-tts-backups/) - Restore command for recovery ## Example Output ``` ═══════════════════════════════════════════════════ AgentVibes TTS Injection Summary ═══════════════════════════════════════════════════ What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ✅ TTS injection applied to 11 file(s): Party Mode (multi-agent conversations): • .bmad/core/workflows/party-mode/instructions.md Agent TTS (individual agent voices): • .bmad/bmm/agents/analyst.md • .bmad/bmm/agents/architect.md ... Backups & Recovery: Pre-injection backups are stored in: ~/.bmad-tts-backups/ To restore original files (removes TTS instructions): bmad-tts-injector.sh --restore /path/to/.bmad ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Claude <noreply@anthropic.com>
2025-12-05 17:54:03 -07:00
let xmlContent = await this.xmlHandler.buildFromYaml(yamlPath, customizeExists ? customizePath : null, {
2025-10-02 21:45:59 -05:00
includeMetadata: true,
});
2025-10-04 21:33:19 -05:00
// DO NOT replace {project-root} - LLMs understand this placeholder at runtime
// const processedContent = xmlContent.replaceAll('{project-root}', projectDir);
2025-10-02 21:45:59 -05:00
feat(installer): Enhanced TTS injection summary with tracking and documentation (#1037) ## Summary - Track all files with TTS injection applied during installation - Display informative summary explaining what TTS injection does - Show backup location and restore command for recovery ## What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ## Changes - **installer.js**: Track files in `processAgentFiles()`, `buildStandaloneAgents()`, and `rebuildAgentFiles()` when TTS markers are processed - **compiler.js**: Add TTS injection support for custom agent compilation - **ui.js**: Enhanced installation summary showing: - Explanation of what TTS injection is with example - List of all files with TTS injection applied (grouped by type) - Backup location (~/.bmad-tts-backups/) - Restore command for recovery ## Example Output ``` ═══════════════════════════════════════════════════ AgentVibes TTS Injection Summary ═══════════════════════════════════════════════════ What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ✅ TTS injection applied to 11 file(s): Party Mode (multi-agent conversations): • .bmad/core/workflows/party-mode/instructions.md Agent TTS (individual agent voices): • .bmad/bmm/agents/analyst.md • .bmad/bmm/agents/architect.md ... Backups & Recovery: Pre-injection backups are stored in: ~/.bmad-tts-backups/ To restore original files (removes TTS instructions): bmad-tts-injector.sh --restore /path/to/.bmad ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Claude <noreply@anthropic.com>
2025-12-05 17:54:03 -07:00
// Process TTS injection points (pass targetPath for tracking)
xmlContent = this.processTTSInjectionPoints(xmlContent, mdPath);
// Write the built .md file to bmad/{module}/agents/ with POSIX-compliant final newline
const content = xmlContent.endsWith('\n') ? xmlContent : xmlContent + '\n';
await fs.writeFile(mdPath, content, 'utf8');
2025-10-02 21:45:59 -05:00
this.installedFiles.push(mdPath);
2025-10-03 11:54:32 -05:00
// Remove the source YAML file - we can regenerate from installer source if needed
await fs.remove(yamlPath);
2025-10-02 21:45:59 -05:00
console.log(chalk.dim(` Built agent: ${agentName}.md`));
}
// Handle legacy .md agents - inject activation if needed
else if (agentFile.endsWith('.md')) {
const agentPath = path.join(agentsPath, agentFile);
let content = await fs.readFile(agentPath, 'utf8');
// Check if content has agent XML and no activation block
if (content.includes('<agent') && !content.includes('<activation')) {
// Inject the activation block using XML handler
content = this.xmlHandler.injectActivationSimple(content);
// Ensure POSIX-compliant final newline
const finalContent = content.endsWith('\n') ? content : content + '\n';
await fs.writeFile(agentPath, finalContent, 'utf8');
2025-10-02 21:45:59 -05:00
}
}
}
}
/**
* Build standalone agents in bmad/agents/ directory
* @param {string} bmadDir - Path to bmad directory
* @param {string} projectDir - Path to project directory
*/
async buildStandaloneAgents(bmadDir, projectDir) {
const standaloneAgentsPath = path.join(bmadDir, 'agents');
const cfgAgentsDir = path.join(bmadDir, '_cfg', 'agents');
// Check if standalone agents directory exists
if (!(await fs.pathExists(standaloneAgentsPath))) {
return;
}
// Get all subdirectories in agents/
const agentDirs = await fs.readdir(standaloneAgentsPath, { withFileTypes: true });
for (const agentDir of agentDirs) {
if (!agentDir.isDirectory()) continue;
const agentDirPath = path.join(standaloneAgentsPath, agentDir.name);
// Find any .agent.yaml file in the directory
const files = await fs.readdir(agentDirPath);
const yamlFile = files.find((f) => f.endsWith('.agent.yaml'));
if (!yamlFile) continue;
const agentName = path.basename(yamlFile, '.agent.yaml');
const sourceYamlPath = path.join(agentDirPath, yamlFile);
const targetMdPath = path.join(agentDirPath, `${agentName}.md`);
const customizePath = path.join(cfgAgentsDir, `${agentName}.customize.yaml`);
// Check for customizations
const customizeExists = await fs.pathExists(customizePath);
let customizedFields = [];
if (customizeExists) {
const customizeContent = await fs.readFile(customizePath, 'utf8');
const yaml = require('js-yaml');
const customizeYaml = yaml.load(customizeContent);
// Detect what fields are customized (similar to rebuildAgentFiles)
if (customizeYaml) {
if (customizeYaml.persona) {
for (const [key, value] of Object.entries(customizeYaml.persona)) {
if (value !== '' && value !== null && !(Array.isArray(value) && value.length === 0)) {
customizedFields.push(`persona.${key}`);
}
}
}
if (customizeYaml.agent?.metadata) {
for (const [key, value] of Object.entries(customizeYaml.agent.metadata)) {
if (value !== '' && value !== null) {
customizedFields.push(`metadata.${key}`);
}
}
}
if (customizeYaml.critical_actions && customizeYaml.critical_actions.length > 0) {
customizedFields.push('critical_actions');
}
if (customizeYaml.menu && customizeYaml.menu.length > 0) {
customizedFields.push('menu');
}
}
}
// Build YAML to XML .md
feat(installer): Enhanced TTS injection summary with tracking and documentation (#1037) ## Summary - Track all files with TTS injection applied during installation - Display informative summary explaining what TTS injection does - Show backup location and restore command for recovery ## What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ## Changes - **installer.js**: Track files in `processAgentFiles()`, `buildStandaloneAgents()`, and `rebuildAgentFiles()` when TTS markers are processed - **compiler.js**: Add TTS injection support for custom agent compilation - **ui.js**: Enhanced installation summary showing: - Explanation of what TTS injection is with example - List of all files with TTS injection applied (grouped by type) - Backup location (~/.bmad-tts-backups/) - Restore command for recovery ## Example Output ``` ═══════════════════════════════════════════════════ AgentVibes TTS Injection Summary ═══════════════════════════════════════════════════ What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ✅ TTS injection applied to 11 file(s): Party Mode (multi-agent conversations): • .bmad/core/workflows/party-mode/instructions.md Agent TTS (individual agent voices): • .bmad/bmm/agents/analyst.md • .bmad/bmm/agents/architect.md ... Backups & Recovery: Pre-injection backups are stored in: ~/.bmad-tts-backups/ To restore original files (removes TTS instructions): bmad-tts-injector.sh --restore /path/to/.bmad ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Claude <noreply@anthropic.com>
2025-12-05 17:54:03 -07:00
let xmlContent = await this.xmlHandler.buildFromYaml(sourceYamlPath, customizeExists ? customizePath : null, {
includeMetadata: true,
});
2025-10-04 21:33:19 -05:00
// DO NOT replace {project-root} - LLMs understand this placeholder at runtime
// const processedContent = xmlContent.replaceAll('{project-root}', projectDir);
feat(installer): Enhanced TTS injection summary with tracking and documentation (#1037) ## Summary - Track all files with TTS injection applied during installation - Display informative summary explaining what TTS injection does - Show backup location and restore command for recovery ## What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ## Changes - **installer.js**: Track files in `processAgentFiles()`, `buildStandaloneAgents()`, and `rebuildAgentFiles()` when TTS markers are processed - **compiler.js**: Add TTS injection support for custom agent compilation - **ui.js**: Enhanced installation summary showing: - Explanation of what TTS injection is with example - List of all files with TTS injection applied (grouped by type) - Backup location (~/.bmad-tts-backups/) - Restore command for recovery ## Example Output ``` ═══════════════════════════════════════════════════ AgentVibes TTS Injection Summary ═══════════════════════════════════════════════════ What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ✅ TTS injection applied to 11 file(s): Party Mode (multi-agent conversations): • .bmad/core/workflows/party-mode/instructions.md Agent TTS (individual agent voices): • .bmad/bmm/agents/analyst.md • .bmad/bmm/agents/architect.md ... Backups & Recovery: Pre-injection backups are stored in: ~/.bmad-tts-backups/ To restore original files (removes TTS instructions): bmad-tts-injector.sh --restore /path/to/.bmad ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Claude <noreply@anthropic.com>
2025-12-05 17:54:03 -07:00
// Process TTS injection points (pass targetPath for tracking)
xmlContent = this.processTTSInjectionPoints(xmlContent, targetMdPath);
// Write the built .md file with POSIX-compliant final newline
const content = xmlContent.endsWith('\n') ? xmlContent : xmlContent + '\n';
await fs.writeFile(targetMdPath, content, 'utf8');
// Display result
if (customizedFields.length > 0) {
console.log(chalk.dim(` Built standalone agent: ${agentName}.md `) + chalk.yellow(`(customized: ${customizedFields.join(', ')})`));
} else {
console.log(chalk.dim(` Built standalone agent: ${agentName}.md`));
}
}
}
2025-10-03 11:54:32 -05:00
/**
* Rebuild agent files from installer source (for compile command)
* @param {string} modulePath - Path to module in bmad/ installation
* @param {string} moduleName - Module name
*/
async rebuildAgentFiles(modulePath, moduleName) {
// Get source agents directory from installer
const sourceAgentsPath =
moduleName === 'core' ? path.join(getModulePath('core'), 'agents') : path.join(getSourcePath(`modules/${moduleName}`), 'agents');
if (!(await fs.pathExists(sourceAgentsPath))) {
return; // No source agents to rebuild
}
// Determine project directory (parent of bmad/ directory)
const bmadDir = path.dirname(modulePath);
const projectDir = path.dirname(bmadDir);
const cfgAgentsDir = path.join(bmadDir, '_cfg', 'agents');
const targetAgentsPath = path.join(modulePath, 'agents');
// Ensure target directory exists
await fs.ensureDir(targetAgentsPath);
// Get all YAML agent files from source
const sourceFiles = await fs.readdir(sourceAgentsPath);
for (const file of sourceFiles) {
if (file.endsWith('.agent.yaml')) {
const agentName = file.replace('.agent.yaml', '');
const sourceYamlPath = path.join(sourceAgentsPath, file);
const targetMdPath = path.join(targetAgentsPath, `${agentName}.md`);
const customizePath = path.join(cfgAgentsDir, `${moduleName}-${agentName}.customize.yaml`);
// Check for customizations
const customizeExists = await fs.pathExists(customizePath);
let customizedFields = [];
if (customizeExists) {
const customizeContent = await fs.readFile(customizePath, 'utf8');
const yaml = require('js-yaml');
const customizeYaml = yaml.load(customizeContent);
// Detect what fields are customized
if (customizeYaml) {
if (customizeYaml.persona) {
for (const [key, value] of Object.entries(customizeYaml.persona)) {
if (value !== '' && value !== null && !(Array.isArray(value) && value.length === 0)) {
customizedFields.push(`persona.${key}`);
}
}
}
if (customizeYaml.agent?.metadata) {
for (const [key, value] of Object.entries(customizeYaml.agent.metadata)) {
if (value !== '' && value !== null) {
customizedFields.push(`metadata.${key}`);
}
}
}
if (customizeYaml.critical_actions && customizeYaml.critical_actions.length > 0) {
customizedFields.push('critical_actions');
}
if (customizeYaml.memories && customizeYaml.memories.length > 0) {
customizedFields.push('memories');
}
if (customizeYaml.menu && customizeYaml.menu.length > 0) {
customizedFields.push('menu');
}
if (customizeYaml.prompts && customizeYaml.prompts.length > 0) {
customizedFields.push('prompts');
}
}
}
// Build YAML + customize to .md
feat(installer): Enhanced TTS injection summary with tracking and documentation (#1037) ## Summary - Track all files with TTS injection applied during installation - Display informative summary explaining what TTS injection does - Show backup location and restore command for recovery ## What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ## Changes - **installer.js**: Track files in `processAgentFiles()`, `buildStandaloneAgents()`, and `rebuildAgentFiles()` when TTS markers are processed - **compiler.js**: Add TTS injection support for custom agent compilation - **ui.js**: Enhanced installation summary showing: - Explanation of what TTS injection is with example - List of all files with TTS injection applied (grouped by type) - Backup location (~/.bmad-tts-backups/) - Restore command for recovery ## Example Output ``` ═══════════════════════════════════════════════════ AgentVibes TTS Injection Summary ═══════════════════════════════════════════════════ What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ✅ TTS injection applied to 11 file(s): Party Mode (multi-agent conversations): • .bmad/core/workflows/party-mode/instructions.md Agent TTS (individual agent voices): • .bmad/bmm/agents/analyst.md • .bmad/bmm/agents/architect.md ... Backups & Recovery: Pre-injection backups are stored in: ~/.bmad-tts-backups/ To restore original files (removes TTS instructions): bmad-tts-injector.sh --restore /path/to/.bmad ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Claude <noreply@anthropic.com>
2025-12-05 17:54:03 -07:00
let xmlContent = await this.xmlHandler.buildFromYaml(sourceYamlPath, customizeExists ? customizePath : null, {
2025-10-03 11:54:32 -05:00
includeMetadata: true,
});
2025-10-04 21:33:19 -05:00
// DO NOT replace {project-root} - LLMs understand this placeholder at runtime
// const processedContent = xmlContent.replaceAll('{project-root}', projectDir);
2025-10-03 11:54:32 -05:00
feat(installer): Enhanced TTS injection summary with tracking and documentation (#1037) ## Summary - Track all files with TTS injection applied during installation - Display informative summary explaining what TTS injection does - Show backup location and restore command for recovery ## What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ## Changes - **installer.js**: Track files in `processAgentFiles()`, `buildStandaloneAgents()`, and `rebuildAgentFiles()` when TTS markers are processed - **compiler.js**: Add TTS injection support for custom agent compilation - **ui.js**: Enhanced installation summary showing: - Explanation of what TTS injection is with example - List of all files with TTS injection applied (grouped by type) - Backup location (~/.bmad-tts-backups/) - Restore command for recovery ## Example Output ``` ═══════════════════════════════════════════════════ AgentVibes TTS Injection Summary ═══════════════════════════════════════════════════ What is TTS Injection? TTS (Text-to-Speech) injection adds voice instructions to BMAD agents, enabling them to speak their responses aloud using AgentVibes. Example: When you activate the PM agent, it will greet you with spoken audio like "Hey! I'm your Project Manager. How can I help?" ✅ TTS injection applied to 11 file(s): Party Mode (multi-agent conversations): • .bmad/core/workflows/party-mode/instructions.md Agent TTS (individual agent voices): • .bmad/bmm/agents/analyst.md • .bmad/bmm/agents/architect.md ... Backups & Recovery: Pre-injection backups are stored in: ~/.bmad-tts-backups/ To restore original files (removes TTS instructions): bmad-tts-injector.sh --restore /path/to/.bmad ``` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Paul Preibisch <paul@paulpreibisch.com> Co-authored-by: Claude <noreply@anthropic.com>
2025-12-05 17:54:03 -07:00
// Process TTS injection points (pass targetPath for tracking)
xmlContent = this.processTTSInjectionPoints(xmlContent, targetMdPath);
// Write the rebuilt .md file with POSIX-compliant final newline
const content = xmlContent.endsWith('\n') ? xmlContent : xmlContent + '\n';
await fs.writeFile(targetMdPath, content, 'utf8');
2025-10-03 11:54:32 -05:00
// Display result with customizations if any
if (customizedFields.length > 0) {
console.log(chalk.dim(` Rebuilt agent: ${agentName}.md `) + chalk.yellow(`(customized: ${customizedFields.join(', ')})`));
} else {
console.log(chalk.dim(` Rebuilt agent: ${agentName}.md`));
}
}
}
}
2025-10-02 21:45:59 -05:00
/**
* Compile/rebuild all agents and tasks for quick updates
* @param {Object} config - Compilation configuration
* @returns {Object} Compilation results
*/
async compileAgents(config) {
const ora = require('ora');
const spinner = ora('Starting agent compilation...').start();
try {
const projectDir = path.resolve(config.directory);
const bmadDir = await this.findBmadDir(projectDir);
2025-10-02 21:45:59 -05:00
// Check if bmad directory exists
if (!(await fs.pathExists(bmadDir))) {
spinner.fail('No BMAD installation found');
throw new Error(`BMAD not installed at ${bmadDir}`);
}
let agentCount = 0;
let taskCount = 0;
// Process all modules in bmad directory
spinner.text = 'Rebuilding agent files...';
const entries = await fs.readdir(bmadDir, { withFileTypes: true });
for (const entry of entries) {
2025-10-03 11:54:32 -05:00
if (entry.isDirectory() && entry.name !== '_cfg' && entry.name !== 'docs') {
2025-10-02 21:45:59 -05:00
const modulePath = path.join(bmadDir, entry.name);
// Special handling for standalone agents in bmad/agents/ directory
if (entry.name === 'agents') {
spinner.text = 'Building standalone agents...';
await this.buildStandaloneAgents(bmadDir, projectDir);
// Count standalone agents
const standaloneAgentsPath = path.join(bmadDir, 'agents');
const standaloneAgentDirs = await fs.readdir(standaloneAgentsPath, { withFileTypes: true });
for (const agentDir of standaloneAgentDirs) {
if (agentDir.isDirectory()) {
const agentDirPath = path.join(standaloneAgentsPath, agentDir.name);
const agentFiles = await fs.readdir(agentDirPath);
agentCount += agentFiles.filter((f) => f.endsWith('.md') && !f.endsWith('.agent.yaml')).length;
}
}
} else {
// Rebuild module agents from installer source
const agentsPath = path.join(modulePath, 'agents');
if (await fs.pathExists(agentsPath)) {
await this.rebuildAgentFiles(modulePath, entry.name);
const agentFiles = await fs.readdir(agentsPath);
agentCount += agentFiles.filter((f) => f.endsWith('.md')).length;
}
2025-10-02 21:45:59 -05:00
// Count tasks (already built)
const tasksPath = path.join(modulePath, 'tasks');
if (await fs.pathExists(tasksPath)) {
const taskFiles = await fs.readdir(tasksPath);
taskCount += taskFiles.filter((f) => f.endsWith('.md')).length;
}
2025-10-02 21:45:59 -05:00
}
}
}
// Reinstall custom agents from _cfg/custom/agents/ sources
spinner.start('Rebuilding custom agents...');
const customAgentResults = await this.reinstallCustomAgents(projectDir, bmadDir);
if (customAgentResults.count > 0) {
spinner.succeed(`Rebuilt ${customAgentResults.count} custom agent${customAgentResults.count > 1 ? 's' : ''}`);
agentCount += customAgentResults.count;
} else {
spinner.succeed('No custom agents found to rebuild');
}
feat: add custom agents and quick-flow workflows, remove tech-spec track Major Changes: - Add sample custom agents demonstrating installable agent system - commit-poet: Generates semantic commit messages (BMAD Method repo sample) - toolsmith: Development tooling expert with knowledge base covering bundlers, deployment, docs, installers, modules, and tests (BMAD Method repo sample) - Both agents demonstrate custom agent architecture and are installable to projects via BMAD installer system - Include comprehensive installation guides and sidecar knowledge bases - Add bmad-quick-flow methodology for rapid development - create-tech-spec: Direct technical specification workflow - quick-dev: Flexible execution workflow supporting both tech-spec-driven and direct instruction development - quick-flow-solo-dev (Barry): 1 man show agent specialized in bmad-quick-flow methodology - Comprehensive documentation for quick-flow approach and solo development - Remove deprecated tech-spec workflow track - Delete entire tech-spec workflow directory and templates - Remove quick-spec-flow.md documentation (replaced by quick-flow docs) - Clean up unused epic and story templates - Fix custom agent installation across IDE installers - Repair antigravity and multiple IDE installers to properly support custom agents - Enable custom agent installation via quick installer, agent installer, regular installer, and special agent installer - All installation methods now accessible via npx with full documentation Infrastructure: - Update BMM module configurations and team setups - Modify workflow status paths to support quick-flow integration - Reorganize documentation with new agent and workflow guides - Add custom/ directory for user customizations - Update platform codes and installer configurations
2025-11-23 08:50:36 -06:00
// Skip full manifest regeneration during compileAgents to preserve custom agents
// Custom agents are already added to manifests during individual installation
// Only regenerate YAML manifest for IDE updates if needed
const existingManifestPath = path.join(bmadDir, '_cfg', 'manifest.yaml');
let existingIdes = [];
if (await fs.pathExists(existingManifestPath)) {
const manifestContent = await fs.readFile(existingManifestPath, 'utf8');
const yaml = require('js-yaml');
const manifest = yaml.load(manifestContent);
existingIdes = manifest.ides || [];
}
// Update IDE configurations using the existing IDE list from manifest
if (existingIdes && existingIdes.length > 0) {
2025-10-02 21:45:59 -05:00
spinner.start('Updating IDE configurations...');
for (const ide of existingIdes) {
2025-10-02 21:45:59 -05:00
spinner.text = `Updating ${ide}...`;
// Stop spinner before IDE setup to prevent blocking any potential prompts
// However, we pass _alreadyConfigured to skip all prompts during compile
spinner.stop();
2025-10-02 21:45:59 -05:00
await this.ideManager.setup(ide, projectDir, bmadDir, {
selectedModules: installedModules,
2025-10-02 21:45:59 -05:00
skipModuleInstall: true, // Skip module installation, just update IDE files
verbose: config.verbose,
preCollectedConfig: { _alreadyConfigured: true }, // Skip all interactive prompts during compile
2025-10-02 21:45:59 -05:00
});
// Restart spinner for next IDE
if (existingIdes.indexOf(ide) < existingIdes.length - 1) {
spinner.start('Updating IDE configurations...');
}
2025-10-02 21:45:59 -05:00
}
console.log(chalk.green('✓ IDE configurations updated'));
} else {
console.log(chalk.yellow('⚠️ No IDEs configured. Skipping IDE update.'));
}
2025-10-02 21:45:59 -05:00
return { agentCount, taskCount };
} catch (error) {
spinner.fail('Compilation failed');
throw error;
}
}
/**
* Private: Update core
*/
async updateCore(bmadDir, force = false) {
const sourcePath = getModulePath('core');
const targetPath = path.join(bmadDir, 'core');
if (force) {
await fs.remove(targetPath);
await this.installCore(bmadDir);
} else {
// Selective update - preserve user modifications
await this.fileOps.syncDirectory(sourcePath, targetPath);
}
}
2025-10-26 16:17:37 -05:00
/**
* Quick update method - preserves all settings and only prompts for new config fields
* @param {Object} config - Configuration with directory
* @returns {Object} Update result
*/
async quickUpdate(config) {
const ora = require('ora');
const spinner = ora('Starting quick update...').start();
try {
const projectDir = path.resolve(config.directory);
const bmadDir = await this.findBmadDir(projectDir);
2025-10-26 16:17:37 -05:00
// Check if bmad directory exists
if (!(await fs.pathExists(bmadDir))) {
spinner.fail('No BMAD installation found');
throw new Error(`BMAD not installed at ${bmadDir}. Use regular install for first-time setup.`);
}
spinner.text = 'Detecting installed modules and configuration...';
// Detect existing installation
const existingInstall = await this.detector.detect(bmadDir);
const installedModules = existingInstall.modules.map((m) => m.id);
const configuredIdes = existingInstall.ides || [];
2025-10-26 17:04:27 -05:00
// Load saved IDE configurations
const savedIdeConfigs = await this.ideConfigManager.loadAllIdeConfigs(bmadDir);
2025-10-26 16:17:37 -05:00
// Get available modules (what we have source for)
const availableModules = await this.moduleManager.listAvailable();
const availableModuleIds = new Set(availableModules.map((m) => m.id));
// Only update modules that are BOTH installed AND available (we have source for)
const modulesToUpdate = installedModules.filter((id) => availableModuleIds.has(id));
const skippedModules = installedModules.filter((id) => !availableModuleIds.has(id));
spinner.succeed(`Found ${modulesToUpdate.length} module(s) to update and ${configuredIdes.length} configured tool(s)`);
if (skippedModules.length > 0) {
console.log(chalk.yellow(`⚠️ Skipping ${skippedModules.length} module(s) - no source available: ${skippedModules.join(', ')}`));
}
// Load existing configs and collect new fields (if any)
console.log(chalk.cyan('\n📋 Checking for new configuration options...'));
await this.configCollector.loadExistingConfig(projectDir);
let promptedForNewFields = false;
// Check core config for new fields
const corePrompted = await this.configCollector.collectModuleConfigQuick('core', projectDir, true);
if (corePrompted) {
promptedForNewFields = true;
}
// Check each module we're updating for new fields (NOT skipped modules)
for (const moduleName of modulesToUpdate) {
const modulePrompted = await this.configCollector.collectModuleConfigQuick(moduleName, projectDir, true);
if (modulePrompted) {
promptedForNewFields = true;
}
}
if (!promptedForNewFields) {
console.log(chalk.green('✓ All configuration is up to date, no new options to configure'));
}
// Add metadata
this.configCollector.collectedConfig._meta = {
version: require(path.join(getProjectRoot(), 'package.json')).version,
installDate: new Date().toISOString(),
lastModified: new Date().toISOString(),
};
// Check if bmad_folder has changed
const existingBmadFolderName = path.basename(bmadDir);
const newBmadFolderName = this.configCollector.collectedConfig.core?.bmad_folder || existingBmadFolderName;
if (existingBmadFolderName === newBmadFolderName) {
// Normal quick update - start the spinner
console.log(chalk.cyan('Updating BMAD installation...'));
} else {
// Folder name has changed - stop spinner and let install() handle it
spinner.stop();
console.log(chalk.yellow(`\n⚠️ Folder name will change: ${existingBmadFolderName}${newBmadFolderName}`));
console.log(chalk.yellow('The installer will handle the folder migration.\n'));
}
2025-10-26 16:17:37 -05:00
// Build the config object for the installer
const installConfig = {
directory: projectDir,
installCore: true,
modules: modulesToUpdate, // Only update modules we have source for
ides: configuredIdes,
skipIde: configuredIdes.length === 0,
coreConfig: this.configCollector.collectedConfig.core,
actionType: 'install', // Use regular install flow
_quickUpdate: true, // Flag to skip certain prompts
_preserveModules: skippedModules, // Preserve these in manifest even though we didn't update them
2025-10-26 17:04:27 -05:00
_savedIdeConfigs: savedIdeConfigs, // Pass saved IDE configs to installer
2025-10-26 16:17:37 -05:00
};
// Call the standard install method
const result = await this.install(installConfig);
// Only succeed the spinner if it's still spinning
// (install method might have stopped it if folder name changed)
if (spinner.isSpinning) {
spinner.succeed('Quick update complete!');
}
2025-10-26 16:17:37 -05:00
return {
success: true,
moduleCount: modulesToUpdate.length + 1, // +1 for core
hadNewFields: promptedForNewFields,
modules: ['core', ...modulesToUpdate],
skippedModules: skippedModules,
ides: configuredIdes,
};
} catch (error) {
spinner.fail('Quick update failed');
throw error;
}
}
/**
* Private: Prompt for update action
*/
async promptUpdateAction() {
const inquirer = require('inquirer');
return await inquirer.prompt([
{
type: 'list',
name: 'action',
message: 'What would you like to do?',
choices: [
{ name: 'Update existing installation', value: 'update' },
{ name: 'Remove and reinstall', value: 'reinstall' },
{ name: 'Cancel', value: 'cancel' },
],
},
]);
}
/**
* Handle legacy BMAD v4 migration with automatic backup
* @param {string} projectDir - Project directory
* @param {Object} legacyV4 - Legacy V4 detection result with offenders array
*/
async handleLegacyV4Migration(projectDir, legacyV4) {
console.log(chalk.yellow.bold('\n⚠ Legacy BMAD v4 detected'));
console.log(chalk.dim('The installer found legacy artefacts in your project.\n'));
// Separate .bmad* folders (auto-backup) from other offending paths (manual cleanup)
const bmadFolders = legacyV4.offenders.filter((p) => {
const name = path.basename(p);
return name.startsWith('.bmad'); // Only dot-prefixed folders get auto-backed up
});
const otherOffenders = legacyV4.offenders.filter((p) => {
const name = path.basename(p);
return !name.startsWith('.bmad'); // Everything else is manual cleanup
});
const inquirer = require('inquirer');
// Show warning for other offending paths FIRST
if (otherOffenders.length > 0) {
console.log(chalk.yellow('⚠️ Recommended cleanup:'));
console.log(chalk.dim('It is recommended to remove the following items before proceeding:\n'));
for (const p of otherOffenders) console.log(chalk.dim(` - ${p}`));
console.log(chalk.cyan('\nCleanup commands you can copy/paste:'));
console.log(chalk.dim('macOS/Linux:'));
for (const p of otherOffenders) console.log(chalk.dim(` rm -rf '${p}'`));
console.log(chalk.dim('Windows:'));
for (const p of otherOffenders) console.log(chalk.dim(` rmdir /S /Q "${p}"`));
const { cleanedUp } = await inquirer.prompt([
{
type: 'confirm',
name: 'cleanedUp',
message: 'Have you completed the recommended cleanup? (You can proceed without it, but it is recommended)',
default: false,
},
]);
if (cleanedUp) {
console.log(chalk.green('✓ Cleanup acknowledged\n'));
} else {
console.log(chalk.yellow('⚠️ Proceeding without recommended cleanup\n'));
}
}
// Handle .bmad* folders with automatic backup
if (bmadFolders.length > 0) {
console.log(chalk.cyan('The following legacy folders will be moved to v4-backup:'));
for (const p of bmadFolders) console.log(chalk.dim(` - ${p}`));
const { proceed } = await inquirer.prompt([
{
type: 'confirm',
name: 'proceed',
message: 'Proceed with backing up legacy v4 folders?',
default: true,
},
]);
if (proceed) {
const backupDir = path.join(projectDir, 'v4-backup');
await fs.ensureDir(backupDir);
for (const folder of bmadFolders) {
const folderName = path.basename(folder);
const backupPath = path.join(backupDir, folderName);
// If backup already exists, add timestamp
let finalBackupPath = backupPath;
if (await fs.pathExists(backupPath)) {
const timestamp = new Date().toISOString().replaceAll(/[:.]/g, '-').split('T')[0];
finalBackupPath = path.join(backupDir, `${folderName}-${timestamp}`);
}
await fs.move(folder, finalBackupPath, { overwrite: false });
console.log(chalk.green(`✓ Moved ${folderName} to ${path.relative(projectDir, finalBackupPath)}`));
}
} else {
throw new Error('Installation cancelled by user');
}
}
}
/**
2025-09-30 21:20:13 -05:00
* Read files-manifest.csv
* @param {string} bmadDir - BMAD installation directory
2025-09-30 21:20:13 -05:00
* @returns {Array} Array of file entries from files-manifest.csv
*/
2025-09-30 21:20:13 -05:00
async readFilesManifest(bmadDir) {
const filesManifestPath = path.join(bmadDir, '_cfg', 'files-manifest.csv');
if (!(await fs.pathExists(filesManifestPath))) {
return [];
}
2025-09-30 21:20:13 -05:00
try {
const content = await fs.readFile(filesManifestPath, 'utf8');
const lines = content.split('\n');
const files = [];
for (let i = 1; i < lines.length; i++) {
// Skip header
const line = lines[i].trim();
if (!line) continue;
// Parse CSV line properly handling quoted values
const parts = [];
let current = '';
let inQuotes = false;
for (const char of line) {
if (char === '"') {
inQuotes = !inQuotes;
} else if (char === ',' && !inQuotes) {
parts.push(current);
current = '';
} else {
current += char;
}
}
parts.push(current); // Add last part
if (parts.length >= 4) {
files.push({
type: parts[0],
name: parts[1],
module: parts[2],
path: parts[3],
hash: parts[4] || null, // Hash may not exist in old manifests
});
}
}
2025-09-30 21:20:13 -05:00
return files;
} catch (error) {
console.warn('Warning: Could not read files-manifest.csv:', error.message);
return [];
}
}
/**
* Detect custom and modified files
* @param {string} bmadDir - BMAD installation directory
* @param {Array} existingFilesManifest - Previous files from files-manifest.csv
* @returns {Object} Object with customFiles and modifiedFiles arrays
*/
async detectCustomFiles(bmadDir, existingFilesManifest) {
const customFiles = [];
const modifiedFiles = [];
// Check if the manifest has hashes - if not, we can't detect modifications
let manifestHasHashes = false;
if (existingFilesManifest && existingFilesManifest.length > 0) {
manifestHasHashes = existingFilesManifest.some((f) => f.hash);
}
// Build map of previously installed files from files-manifest.csv with their hashes
const installedFilesMap = new Map();
for (const fileEntry of existingFilesManifest) {
if (fileEntry.path) {
// Paths are relative to bmadDir. Legacy manifests incorrectly prefixed 'bmad/' -
// strip it if present. This is safe because no real path inside bmadDir would
// start with 'bmad/' (you'd never have .bmad/bmad/... as an actual structure).
2025-09-30 21:20:13 -05:00
const relativePath = fileEntry.path.startsWith('bmad/') ? fileEntry.path.slice(5) : fileEntry.path;
const absolutePath = path.join(bmadDir, relativePath);
installedFilesMap.set(path.normalize(absolutePath), {
hash: fileEntry.hash,
relativePath: relativePath,
});
}
}
// Recursively scan bmadDir for all files
const scanDirectory = async (dir) => {
try {
const entries = await fs.readdir(dir, { withFileTypes: true });
for (const entry of entries) {
const fullPath = path.join(dir, entry.name);
if (entry.isDirectory()) {
// Skip certain directories
if (entry.name === 'node_modules' || entry.name === '.git') {
continue;
}
await scanDirectory(fullPath);
} else if (entry.isFile()) {
const normalizedPath = path.normalize(fullPath);
2025-09-30 21:20:13 -05:00
const fileInfo = installedFilesMap.get(normalizedPath);
// Skip certain system files that are auto-generated
const relativePath = path.relative(bmadDir, fullPath);
const fileName = path.basename(fullPath);
// Skip _cfg directory - system files
if (relativePath.startsWith('_cfg/') || relativePath.startsWith('_cfg\\')) {
continue;
}
// Skip config.yaml files - these are regenerated on each install/update
// Users should use _cfg/agents/ override files instead
if (fileName === 'config.yaml') {
continue;
}
if (!fileInfo) {
// File not in manifest = custom file
customFiles.push(fullPath);
2025-09-30 21:20:13 -05:00
} else if (manifestHasHashes && fileInfo.hash) {
// File in manifest with hash - check if it was modified
const currentHash = await this.manifest.calculateFileHash(fullPath);
if (currentHash && currentHash !== fileInfo.hash) {
// Hash changed = file was modified
modifiedFiles.push({
path: fullPath,
relativePath: fileInfo.relativePath,
});
}
}
2025-09-30 21:20:13 -05:00
// If manifest doesn't have hashes, we can't detect modifications
// so we just skip files that are in the manifest
}
}
} catch {
// Ignore errors scanning directories
}
};
await scanDirectory(bmadDir);
2025-09-30 21:20:13 -05:00
return { customFiles, modifiedFiles };
}
/**
* Private: Create agent configuration files
* @param {string} bmadDir - BMAD installation directory
* @param {Object} userInfo - User information including name and language
*/
async createAgentConfigs(bmadDir, userInfo = null) {
const agentConfigDir = path.join(bmadDir, '_cfg', 'agents');
await fs.ensureDir(agentConfigDir);
// Get all agents from all modules
const agents = [];
const agentDetails = []; // For manifest generation
// Check modules for agents (including core)
const entries = await fs.readdir(bmadDir, { withFileTypes: true });
for (const entry of entries) {
if (entry.isDirectory() && entry.name !== '_cfg') {
const moduleAgentsPath = path.join(bmadDir, entry.name, 'agents');
if (await fs.pathExists(moduleAgentsPath)) {
const agentFiles = await fs.readdir(moduleAgentsPath);
for (const agentFile of agentFiles) {
if (agentFile.endsWith('.md')) {
const agentPath = path.join(moduleAgentsPath, agentFile);
const agentContent = await fs.readFile(agentPath, 'utf8');
// Skip agents with localskip="true"
const hasLocalSkip = agentContent.match(/<agent[^>]*\slocalskip="true"[^>]*>/);
if (hasLocalSkip) {
continue; // Skip this agent - it should not have been installed
}
const agentName = path.basename(agentFile, '.md');
// Extract any nodes with agentConfig="true"
const agentConfigNodes = this.extractAgentConfigNodes(agentContent);
agents.push({
name: agentName,
module: entry.name,
agentConfigNodes: agentConfigNodes,
});
// Use shared AgentPartyGenerator to extract details
let details = AgentPartyGenerator.extractAgentDetails(agentContent, entry.name, agentName);
// Apply config overrides if they exist
if (details) {
const configPath = path.join(agentConfigDir, `${entry.name}-${agentName}.md`);
if (await fs.pathExists(configPath)) {
const configContent = await fs.readFile(configPath, 'utf8');
details = AgentPartyGenerator.applyConfigOverrides(details, configContent);
}
agentDetails.push(details);
}
}
}
}
}
}
// Create config file for each agent
let createdCount = 0;
let skippedCount = 0;
// Load agent config template
const templatePath = getSourcePath('utility', 'models', 'agent-config-template.md');
const templateContent = await fs.readFile(templatePath, 'utf8');
for (const agent of agents) {
const configPath = path.join(agentConfigDir, `${agent.module}-${agent.name}.md`);
// Skip if config file already exists (preserve custom configurations)
if (await fs.pathExists(configPath)) {
skippedCount++;
continue;
}
// Build config content header
let configContent = `# Agent Config: ${agent.name}\n\n`;
// Process template and add agent-specific config nodes
let processedTemplate = templateContent;
// Replace {core:user_name} placeholder with actual user name if available
if (userInfo && userInfo.userName) {
processedTemplate = processedTemplate.replaceAll('{core:user_name}', userInfo.userName);
}
// Replace {core:communication_language} placeholder with actual language if available
if (userInfo && userInfo.responseLanguage) {
processedTemplate = processedTemplate.replaceAll('{core:communication_language}', userInfo.responseLanguage);
}
// If this agent has agentConfig nodes, add them after the existing comment
if (agent.agentConfigNodes && agent.agentConfigNodes.length > 0) {
// Find the agent-specific configuration nodes comment
const commentPattern = /(\s*<!-- Agent-specific configuration nodes -->)/;
const commentMatch = processedTemplate.match(commentPattern);
if (commentMatch) {
// Add nodes right after the comment
let agentSpecificNodes = '';
for (const node of agent.agentConfigNodes) {
agentSpecificNodes += `\n ${node}`;
}
processedTemplate = processedTemplate.replace(commentPattern, `$1${agentSpecificNodes}`);
}
}
configContent += processedTemplate;
// Ensure POSIX-compliant final newline
if (!configContent.endsWith('\n')) {
configContent += '\n';
}
await fs.writeFile(configPath, configContent, 'utf8');
2025-09-30 21:20:13 -05:00
this.installedFiles.push(configPath); // Track agent config files
createdCount++;
}
// Generate agent manifest with overrides applied
await this.generateAgentManifest(bmadDir, agentDetails);
return { total: agents.length, created: createdCount, skipped: skippedCount };
}
/**
* Generate agent manifest XML file
* @param {string} bmadDir - BMAD installation directory
* @param {Array} agentDetails - Array of agent details
*/
async generateAgentManifest(bmadDir, agentDetails) {
const manifestPath = path.join(bmadDir, '_cfg', 'agent-manifest.csv');
await AgentPartyGenerator.writeAgentParty(manifestPath, agentDetails, { forWeb: false });
}
/**
* Extract nodes with agentConfig="true" from agent content
* @param {string} content - Agent file content
* @returns {Array} Array of XML nodes that should be added to agent config
*/
extractAgentConfigNodes(content) {
const nodes = [];
try {
// Find all XML nodes with agentConfig="true"
// Match self-closing tags and tags with content
const selfClosingPattern = /<([a-zA-Z][a-zA-Z0-9_-]*)\s+[^>]*agentConfig="true"[^>]*\/>/g;
const withContentPattern = /<([a-zA-Z][a-zA-Z0-9_-]*)\s+[^>]*agentConfig="true"[^>]*>([\s\S]*?)<\/\1>/g;
// Extract self-closing tags
let match;
while ((match = selfClosingPattern.exec(content)) !== null) {
// Extract just the tag without children (structure only)
const tagMatch = match[0].match(/<([a-zA-Z][a-zA-Z0-9_-]*)([^>]*)\/>/);
if (tagMatch) {
const tagName = tagMatch[1];
const attributes = tagMatch[2].replace(/\s*agentConfig="true"/, ''); // Remove agentConfig attribute
nodes.push(`<${tagName}${attributes}></${tagName}>`);
}
}
// Extract tags with content
while ((match = withContentPattern.exec(content)) !== null) {
const fullMatch = match[0];
const tagName = match[1];
// Extract opening tag with attributes (removing agentConfig="true")
const openingTagMatch = fullMatch.match(new RegExp(`<${tagName}([^>]*)>`));
if (openingTagMatch) {
const attributes = openingTagMatch[1].replace(/\s*agentConfig="true"/, '');
// Add empty node structure (no children)
nodes.push(`<${tagName}${attributes}></${tagName}>`);
}
}
} catch (error) {
console.error('Error extracting agentConfig nodes:', error);
}
return nodes;
}
feat: Complete BMAD agent creation system with install tooling, references, and field guidance ## Overview This commit represents a complete overhaul of the BMAD agent creation system, establishing clear standards for agent development, installation workflows, and persona design. The changes span documentation, tooling, reference implementations, and field-specific guidance. ## Key Components ### 1. Agent Installation Infrastructure **New CLI Command: `agent-install`** - Interactive agent installation with persona customization - Supports Simple (single YAML), Expert (sidecar files), and Module agents - Template variable processing with Handlebars-style syntax - Automatic compilation from YAML to XML (.md) format - Manifest tracking and IDE integration (Claude Code, Cursor, Windsurf, etc.) - Source preservation in `_cfg/custom/agents/` for reinstallation **Files Created:** - `tools/cli/commands/agent-install.js` - Main CLI command - `tools/cli/lib/agent/compiler.js` - YAML to XML compilation engine - `tools/cli/lib/agent/installer.js` - Installation orchestration - `tools/cli/lib/agent/template-engine.js` - Handlebars template processing **Compiler Features:** - Auto-injects frontmatter, activation, handlers, help/exit menu items - Smart handler inclusion (only includes action/workflow/exec/tmpl handlers actually used) - Proper XML escaping and formatting - Persona name customization (e.g., "Fred the Commit Poet") ### 2. Documentation Overhaul **Deleted Bloated/Outdated Docs (2,651 lines removed):** - Old verbose architecture docs - Redundant pattern files - Outdated workflow guides **Created Focused, Type-Specific Docs:** - `src/modules/bmb/docs/understanding-agent-types.md` - Architecture vs capability distinction - `src/modules/bmb/docs/simple-agent-architecture.md` - Self-contained agents - `src/modules/bmb/docs/expert-agent-architecture.md` - Agents with sidecar files - `src/modules/bmb/docs/module-agent-architecture.md` - Workflow-integrated agents - `src/modules/bmb/docs/agent-compilation.md` - YAML → XML process - `src/modules/bmb/docs/agent-menu-patterns.md` - Menu design patterns - `src/modules/bmb/docs/index.md` - Documentation hub **Net Result:** ~1,930 line reduction while adding MORE value through focused content ### 3. Create-Agent Workflow Enhancements **Critical Persona Field Guidance Added to Step 4:** Explains how the LLM interprets each persona field when the agent activates: - **role** → "What knowledge, skills, and capabilities do I possess?" - **identity** → "What background, experience, and context shape my responses?" - **communication_style** → "What verbal patterns, word choice, quirks, and phrasing do I use?" - **principles** → "What beliefs and operating philosophy drive my choices?" **Key Insight:** `communication_style` should ONLY describe HOW the agent talks, not restate role/identity/principles. The `communication-presets.csv` provides 60 pure communication styles with NO role/identity/principles mixed in. **Files Updated:** - `src/modules/bmb/workflows/create-agent/instructions.md` - Added persona field interpretation guide - `src/modules/bmb/workflows/create-agent/brainstorm-context.md` - Refined to 137 lines - `src/modules/bmb/workflows/create-agent/communication-presets.csv` - 60 styles across 13 categories ### 4. Reference Agent Cleanup **Removed install_config Personality Bloat:** Understanding: Future installer will handle personality customization, so stripped all personality toggles from reference agents. **commit-poet.agent.yaml** (Simple Agent): - BEFORE: 36 personality combinations (3 enthusiasm × 3 depths × 4 styles) = decision fatigue - AFTER: Single concise persona with pure communication style - Changed from verbose conditionals to: "Poetic drama and flair with every turn of a phrase. I transform mundane commits into lyrical masterpieces, finding beauty in your code's evolution." - Reduction: 248 lines → 153 lines (38% reduction) **journal-keeper.agent.yaml** (Expert Agent): - Stripped install_config, simplified communication_style - Shows proper Expert agent structure with sidecar files **security-engineer.agent.yaml & trend-analyst.agent.yaml** (Module Agents): - Added header comments explaining WHY Module Agent (design intent, not just location) - Clarified: Module agents are designed FOR ecosystem integration, not capability-limited **Files Updated:** - `src/modules/bmb/reference/agents/simple-examples/commit-poet.agent.yaml` - `src/modules/bmb/reference/agents/expert-examples/journal-keeper/journal-keeper.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/security-engineer.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/trend-analyst.agent.yaml` ### 5. BMM Agent Voice Enhancement **Gave all 9 BMM agents distinct, memorable communication voices:** **Mary (analyst)** - The favorite! Changed from generic "systematic and probing" to: "Treats analysis like a treasure hunt - excited by every clue, thrilled when patterns emerge. Asks questions that spark 'aha!' moments while structuring insights with precision." **Other Notable Voices:** - **John (pm):** "Asks 'WHY?' relentlessly like a detective on a case. Direct and data-sharp, cuts through fluff to what actually matters." - **Winston (architect):** "Speaks in calm, pragmatic tones, balancing 'what could be' with 'what should be.' Champions boring technology that actually works." - **Amelia (dev):** "Ultra-succinct. Speaks in file paths and AC IDs - every statement citable. No fluff, all precision." - **Bob (sm):** "Crisp and checklist-driven. Every word has a purpose, every requirement crystal clear. Zero tolerance for ambiguity." - **Sally (ux-designer):** "Paints pictures with words, telling user stories that make you FEEL the problem. Empathetic advocate with creative storytelling flair." **Pattern Applied:** Moved behaviors from communication_style to principles, keeping communication_style as PURE verbal patterns. **Files Updated:** - `src/modules/bmm/agents/analyst.agent.yaml` - `src/modules/bmm/agents/pm.agent.yaml` - `src/modules/bmm/agents/architect.agent.yaml` - `src/modules/bmm/agents/dev.agent.yaml` - `src/modules/bmm/agents/sm.agent.yaml` - `src/modules/bmm/agents/tea.agent.yaml` - `src/modules/bmm/agents/tech-writer.agent.yaml` - `src/modules/bmm/agents/ux-designer.agent.yaml` - `src/modules/bmm/agents/frame-expert.agent.yaml` ### 6. Linting Fixes **ESLint Compliance:** - Replaced all `'utf-8'` with `'utf8'` (unicorn/text-encoding-identifier-case) - Changed `variables.hasOwnProperty(varName)` to `Object.hasOwn(variables, varName)` (unicorn/prefer-object-has-own) - Replaced `JSON.parse(JSON.stringify(...))` with `structuredClone(...)` (unicorn/prefer-structured-clone) - Fixed empty YAML mapping values in sample files **Files Fixed:** - 7 JavaScript files across agent tooling (compiler, installer, commands, IDE integration) - 1 YAML sample file ## Architecture Decisions ### Agent Types Are About Architecture, Not Capability - **Simple:** Self-contained in single YAML (NOT limited in capability) - **Expert:** Includes sidecar files (templates, docs, etc.) - **Module:** Designed for BMAD ecosystem integration (workflows, cross-agent coordination) ### Persona Field Separation Critical for LLM Interpretation The LLM needs distinct fields to understand its role: - Mixing role/identity/principles into communication_style confuses the persona - Pure communication styles (from communication-presets.csv) have ZERO role/identity/principles content - Example DON'T: "Experienced analyst who uses systematic approaches..." (mixing identity + style) - Example DO: "Systematic and probing. Structures findings hierarchically." (pure style) ### Install-Time vs Runtime Configuration - Template variables ({{var}}) resolve at compile-time - Runtime variables ({user_name}, {bmad_folder}) resolve when agent activates - Future installer will handle personality customization, so agents should ship with single default persona ## Testing - All linting passes (ESLint with max-warnings=0) - Agent compilation tested with commit-poet, journal-keeper examples - Install workflow validated with Simple and Expert agent types - Manifest tracking and IDE integration verified ## Impact This establishes BMAD as having a complete, production-ready agent creation and installation system with: - Clear documentation for all agent types - Automated compilation and installation - Strong persona design guidance - Reference implementations showing best practices - Distinct, memorable agent voices throughout BMM module Co-Authored-By: BMad Builder <builder@bmad.dev> Co-Authored-By: Mary the Analyst <analyst@bmad.dev> Co-Authored-By: Paige the Tech Writer <tech-writer@bmad.dev>
2025-11-17 22:25:15 -06:00
/**
* Reinstall custom agents from backup and source locations
feat: Complete BMAD agent creation system with install tooling, references, and field guidance ## Overview This commit represents a complete overhaul of the BMAD agent creation system, establishing clear standards for agent development, installation workflows, and persona design. The changes span documentation, tooling, reference implementations, and field-specific guidance. ## Key Components ### 1. Agent Installation Infrastructure **New CLI Command: `agent-install`** - Interactive agent installation with persona customization - Supports Simple (single YAML), Expert (sidecar files), and Module agents - Template variable processing with Handlebars-style syntax - Automatic compilation from YAML to XML (.md) format - Manifest tracking and IDE integration (Claude Code, Cursor, Windsurf, etc.) - Source preservation in `_cfg/custom/agents/` for reinstallation **Files Created:** - `tools/cli/commands/agent-install.js` - Main CLI command - `tools/cli/lib/agent/compiler.js` - YAML to XML compilation engine - `tools/cli/lib/agent/installer.js` - Installation orchestration - `tools/cli/lib/agent/template-engine.js` - Handlebars template processing **Compiler Features:** - Auto-injects frontmatter, activation, handlers, help/exit menu items - Smart handler inclusion (only includes action/workflow/exec/tmpl handlers actually used) - Proper XML escaping and formatting - Persona name customization (e.g., "Fred the Commit Poet") ### 2. Documentation Overhaul **Deleted Bloated/Outdated Docs (2,651 lines removed):** - Old verbose architecture docs - Redundant pattern files - Outdated workflow guides **Created Focused, Type-Specific Docs:** - `src/modules/bmb/docs/understanding-agent-types.md` - Architecture vs capability distinction - `src/modules/bmb/docs/simple-agent-architecture.md` - Self-contained agents - `src/modules/bmb/docs/expert-agent-architecture.md` - Agents with sidecar files - `src/modules/bmb/docs/module-agent-architecture.md` - Workflow-integrated agents - `src/modules/bmb/docs/agent-compilation.md` - YAML → XML process - `src/modules/bmb/docs/agent-menu-patterns.md` - Menu design patterns - `src/modules/bmb/docs/index.md` - Documentation hub **Net Result:** ~1,930 line reduction while adding MORE value through focused content ### 3. Create-Agent Workflow Enhancements **Critical Persona Field Guidance Added to Step 4:** Explains how the LLM interprets each persona field when the agent activates: - **role** → "What knowledge, skills, and capabilities do I possess?" - **identity** → "What background, experience, and context shape my responses?" - **communication_style** → "What verbal patterns, word choice, quirks, and phrasing do I use?" - **principles** → "What beliefs and operating philosophy drive my choices?" **Key Insight:** `communication_style` should ONLY describe HOW the agent talks, not restate role/identity/principles. The `communication-presets.csv` provides 60 pure communication styles with NO role/identity/principles mixed in. **Files Updated:** - `src/modules/bmb/workflows/create-agent/instructions.md` - Added persona field interpretation guide - `src/modules/bmb/workflows/create-agent/brainstorm-context.md` - Refined to 137 lines - `src/modules/bmb/workflows/create-agent/communication-presets.csv` - 60 styles across 13 categories ### 4. Reference Agent Cleanup **Removed install_config Personality Bloat:** Understanding: Future installer will handle personality customization, so stripped all personality toggles from reference agents. **commit-poet.agent.yaml** (Simple Agent): - BEFORE: 36 personality combinations (3 enthusiasm × 3 depths × 4 styles) = decision fatigue - AFTER: Single concise persona with pure communication style - Changed from verbose conditionals to: "Poetic drama and flair with every turn of a phrase. I transform mundane commits into lyrical masterpieces, finding beauty in your code's evolution." - Reduction: 248 lines → 153 lines (38% reduction) **journal-keeper.agent.yaml** (Expert Agent): - Stripped install_config, simplified communication_style - Shows proper Expert agent structure with sidecar files **security-engineer.agent.yaml & trend-analyst.agent.yaml** (Module Agents): - Added header comments explaining WHY Module Agent (design intent, not just location) - Clarified: Module agents are designed FOR ecosystem integration, not capability-limited **Files Updated:** - `src/modules/bmb/reference/agents/simple-examples/commit-poet.agent.yaml` - `src/modules/bmb/reference/agents/expert-examples/journal-keeper/journal-keeper.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/security-engineer.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/trend-analyst.agent.yaml` ### 5. BMM Agent Voice Enhancement **Gave all 9 BMM agents distinct, memorable communication voices:** **Mary (analyst)** - The favorite! Changed from generic "systematic and probing" to: "Treats analysis like a treasure hunt - excited by every clue, thrilled when patterns emerge. Asks questions that spark 'aha!' moments while structuring insights with precision." **Other Notable Voices:** - **John (pm):** "Asks 'WHY?' relentlessly like a detective on a case. Direct and data-sharp, cuts through fluff to what actually matters." - **Winston (architect):** "Speaks in calm, pragmatic tones, balancing 'what could be' with 'what should be.' Champions boring technology that actually works." - **Amelia (dev):** "Ultra-succinct. Speaks in file paths and AC IDs - every statement citable. No fluff, all precision." - **Bob (sm):** "Crisp and checklist-driven. Every word has a purpose, every requirement crystal clear. Zero tolerance for ambiguity." - **Sally (ux-designer):** "Paints pictures with words, telling user stories that make you FEEL the problem. Empathetic advocate with creative storytelling flair." **Pattern Applied:** Moved behaviors from communication_style to principles, keeping communication_style as PURE verbal patterns. **Files Updated:** - `src/modules/bmm/agents/analyst.agent.yaml` - `src/modules/bmm/agents/pm.agent.yaml` - `src/modules/bmm/agents/architect.agent.yaml` - `src/modules/bmm/agents/dev.agent.yaml` - `src/modules/bmm/agents/sm.agent.yaml` - `src/modules/bmm/agents/tea.agent.yaml` - `src/modules/bmm/agents/tech-writer.agent.yaml` - `src/modules/bmm/agents/ux-designer.agent.yaml` - `src/modules/bmm/agents/frame-expert.agent.yaml` ### 6. Linting Fixes **ESLint Compliance:** - Replaced all `'utf-8'` with `'utf8'` (unicorn/text-encoding-identifier-case) - Changed `variables.hasOwnProperty(varName)` to `Object.hasOwn(variables, varName)` (unicorn/prefer-object-has-own) - Replaced `JSON.parse(JSON.stringify(...))` with `structuredClone(...)` (unicorn/prefer-structured-clone) - Fixed empty YAML mapping values in sample files **Files Fixed:** - 7 JavaScript files across agent tooling (compiler, installer, commands, IDE integration) - 1 YAML sample file ## Architecture Decisions ### Agent Types Are About Architecture, Not Capability - **Simple:** Self-contained in single YAML (NOT limited in capability) - **Expert:** Includes sidecar files (templates, docs, etc.) - **Module:** Designed for BMAD ecosystem integration (workflows, cross-agent coordination) ### Persona Field Separation Critical for LLM Interpretation The LLM needs distinct fields to understand its role: - Mixing role/identity/principles into communication_style confuses the persona - Pure communication styles (from communication-presets.csv) have ZERO role/identity/principles content - Example DON'T: "Experienced analyst who uses systematic approaches..." (mixing identity + style) - Example DO: "Systematic and probing. Structures findings hierarchically." (pure style) ### Install-Time vs Runtime Configuration - Template variables ({{var}}) resolve at compile-time - Runtime variables ({user_name}, {bmad_folder}) resolve when agent activates - Future installer will handle personality customization, so agents should ship with single default persona ## Testing - All linting passes (ESLint with max-warnings=0) - Agent compilation tested with commit-poet, journal-keeper examples - Install workflow validated with Simple and Expert agent types - Manifest tracking and IDE integration verified ## Impact This establishes BMAD as having a complete, production-ready agent creation and installation system with: - Clear documentation for all agent types - Automated compilation and installation - Strong persona design guidance - Reference implementations showing best practices - Distinct, memorable agent voices throughout BMM module Co-Authored-By: BMad Builder <builder@bmad.dev> Co-Authored-By: Mary the Analyst <analyst@bmad.dev> Co-Authored-By: Paige the Tech Writer <tech-writer@bmad.dev>
2025-11-17 22:25:15 -06:00
* This preserves custom agents across quick updates/reinstalls
* @param {string} projectDir - Project directory
* @param {string} bmadDir - BMAD installation directory
* @returns {Object} Result with count and agent names
*/
async reinstallCustomAgents(projectDir, bmadDir) {
const {
discoverAgents,
loadAgentConfig,
extractManifestData,
addToManifest,
createIdeSlashCommands,
updateManifestYaml,
} = require('../../../lib/agent/installer');
const { compileAgent } = require('../../../lib/agent/compiler');
feat: Complete BMAD agent creation system with install tooling, references, and field guidance ## Overview This commit represents a complete overhaul of the BMAD agent creation system, establishing clear standards for agent development, installation workflows, and persona design. The changes span documentation, tooling, reference implementations, and field-specific guidance. ## Key Components ### 1. Agent Installation Infrastructure **New CLI Command: `agent-install`** - Interactive agent installation with persona customization - Supports Simple (single YAML), Expert (sidecar files), and Module agents - Template variable processing with Handlebars-style syntax - Automatic compilation from YAML to XML (.md) format - Manifest tracking and IDE integration (Claude Code, Cursor, Windsurf, etc.) - Source preservation in `_cfg/custom/agents/` for reinstallation **Files Created:** - `tools/cli/commands/agent-install.js` - Main CLI command - `tools/cli/lib/agent/compiler.js` - YAML to XML compilation engine - `tools/cli/lib/agent/installer.js` - Installation orchestration - `tools/cli/lib/agent/template-engine.js` - Handlebars template processing **Compiler Features:** - Auto-injects frontmatter, activation, handlers, help/exit menu items - Smart handler inclusion (only includes action/workflow/exec/tmpl handlers actually used) - Proper XML escaping and formatting - Persona name customization (e.g., "Fred the Commit Poet") ### 2. Documentation Overhaul **Deleted Bloated/Outdated Docs (2,651 lines removed):** - Old verbose architecture docs - Redundant pattern files - Outdated workflow guides **Created Focused, Type-Specific Docs:** - `src/modules/bmb/docs/understanding-agent-types.md` - Architecture vs capability distinction - `src/modules/bmb/docs/simple-agent-architecture.md` - Self-contained agents - `src/modules/bmb/docs/expert-agent-architecture.md` - Agents with sidecar files - `src/modules/bmb/docs/module-agent-architecture.md` - Workflow-integrated agents - `src/modules/bmb/docs/agent-compilation.md` - YAML → XML process - `src/modules/bmb/docs/agent-menu-patterns.md` - Menu design patterns - `src/modules/bmb/docs/index.md` - Documentation hub **Net Result:** ~1,930 line reduction while adding MORE value through focused content ### 3. Create-Agent Workflow Enhancements **Critical Persona Field Guidance Added to Step 4:** Explains how the LLM interprets each persona field when the agent activates: - **role** → "What knowledge, skills, and capabilities do I possess?" - **identity** → "What background, experience, and context shape my responses?" - **communication_style** → "What verbal patterns, word choice, quirks, and phrasing do I use?" - **principles** → "What beliefs and operating philosophy drive my choices?" **Key Insight:** `communication_style` should ONLY describe HOW the agent talks, not restate role/identity/principles. The `communication-presets.csv` provides 60 pure communication styles with NO role/identity/principles mixed in. **Files Updated:** - `src/modules/bmb/workflows/create-agent/instructions.md` - Added persona field interpretation guide - `src/modules/bmb/workflows/create-agent/brainstorm-context.md` - Refined to 137 lines - `src/modules/bmb/workflows/create-agent/communication-presets.csv` - 60 styles across 13 categories ### 4. Reference Agent Cleanup **Removed install_config Personality Bloat:** Understanding: Future installer will handle personality customization, so stripped all personality toggles from reference agents. **commit-poet.agent.yaml** (Simple Agent): - BEFORE: 36 personality combinations (3 enthusiasm × 3 depths × 4 styles) = decision fatigue - AFTER: Single concise persona with pure communication style - Changed from verbose conditionals to: "Poetic drama and flair with every turn of a phrase. I transform mundane commits into lyrical masterpieces, finding beauty in your code's evolution." - Reduction: 248 lines → 153 lines (38% reduction) **journal-keeper.agent.yaml** (Expert Agent): - Stripped install_config, simplified communication_style - Shows proper Expert agent structure with sidecar files **security-engineer.agent.yaml & trend-analyst.agent.yaml** (Module Agents): - Added header comments explaining WHY Module Agent (design intent, not just location) - Clarified: Module agents are designed FOR ecosystem integration, not capability-limited **Files Updated:** - `src/modules/bmb/reference/agents/simple-examples/commit-poet.agent.yaml` - `src/modules/bmb/reference/agents/expert-examples/journal-keeper/journal-keeper.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/security-engineer.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/trend-analyst.agent.yaml` ### 5. BMM Agent Voice Enhancement **Gave all 9 BMM agents distinct, memorable communication voices:** **Mary (analyst)** - The favorite! Changed from generic "systematic and probing" to: "Treats analysis like a treasure hunt - excited by every clue, thrilled when patterns emerge. Asks questions that spark 'aha!' moments while structuring insights with precision." **Other Notable Voices:** - **John (pm):** "Asks 'WHY?' relentlessly like a detective on a case. Direct and data-sharp, cuts through fluff to what actually matters." - **Winston (architect):** "Speaks in calm, pragmatic tones, balancing 'what could be' with 'what should be.' Champions boring technology that actually works." - **Amelia (dev):** "Ultra-succinct. Speaks in file paths and AC IDs - every statement citable. No fluff, all precision." - **Bob (sm):** "Crisp and checklist-driven. Every word has a purpose, every requirement crystal clear. Zero tolerance for ambiguity." - **Sally (ux-designer):** "Paints pictures with words, telling user stories that make you FEEL the problem. Empathetic advocate with creative storytelling flair." **Pattern Applied:** Moved behaviors from communication_style to principles, keeping communication_style as PURE verbal patterns. **Files Updated:** - `src/modules/bmm/agents/analyst.agent.yaml` - `src/modules/bmm/agents/pm.agent.yaml` - `src/modules/bmm/agents/architect.agent.yaml` - `src/modules/bmm/agents/dev.agent.yaml` - `src/modules/bmm/agents/sm.agent.yaml` - `src/modules/bmm/agents/tea.agent.yaml` - `src/modules/bmm/agents/tech-writer.agent.yaml` - `src/modules/bmm/agents/ux-designer.agent.yaml` - `src/modules/bmm/agents/frame-expert.agent.yaml` ### 6. Linting Fixes **ESLint Compliance:** - Replaced all `'utf-8'` with `'utf8'` (unicorn/text-encoding-identifier-case) - Changed `variables.hasOwnProperty(varName)` to `Object.hasOwn(variables, varName)` (unicorn/prefer-object-has-own) - Replaced `JSON.parse(JSON.stringify(...))` with `structuredClone(...)` (unicorn/prefer-structured-clone) - Fixed empty YAML mapping values in sample files **Files Fixed:** - 7 JavaScript files across agent tooling (compiler, installer, commands, IDE integration) - 1 YAML sample file ## Architecture Decisions ### Agent Types Are About Architecture, Not Capability - **Simple:** Self-contained in single YAML (NOT limited in capability) - **Expert:** Includes sidecar files (templates, docs, etc.) - **Module:** Designed for BMAD ecosystem integration (workflows, cross-agent coordination) ### Persona Field Separation Critical for LLM Interpretation The LLM needs distinct fields to understand its role: - Mixing role/identity/principles into communication_style confuses the persona - Pure communication styles (from communication-presets.csv) have ZERO role/identity/principles content - Example DON'T: "Experienced analyst who uses systematic approaches..." (mixing identity + style) - Example DO: "Systematic and probing. Structures findings hierarchically." (pure style) ### Install-Time vs Runtime Configuration - Template variables ({{var}}) resolve at compile-time - Runtime variables ({user_name}, {bmad_folder}) resolve when agent activates - Future installer will handle personality customization, so agents should ship with single default persona ## Testing - All linting passes (ESLint with max-warnings=0) - Agent compilation tested with commit-poet, journal-keeper examples - Install workflow validated with Simple and Expert agent types - Manifest tracking and IDE integration verified ## Impact This establishes BMAD as having a complete, production-ready agent creation and installation system with: - Clear documentation for all agent types - Automated compilation and installation - Strong persona design guidance - Reference implementations showing best practices - Distinct, memorable agent voices throughout BMM module Co-Authored-By: BMad Builder <builder@bmad.dev> Co-Authored-By: Mary the Analyst <analyst@bmad.dev> Co-Authored-By: Paige the Tech Writer <tech-writer@bmad.dev>
2025-11-17 22:25:15 -06:00
const results = { count: 0, agents: [] };
// Check multiple locations for custom agents
const sourceLocations = [
path.join(bmadDir, '_cfg', 'custom', 'agents'), // Backup location
path.join(bmadDir, 'custom', 'src', 'agents'), // BMAD folder source location
path.join(projectDir, 'custom', 'src', 'agents'), // Project root source location
];
let foundAgents = [];
let processedAgents = new Set(); // Track to avoid duplicates
// Discover agents from all locations
for (const location of sourceLocations) {
if (await fs.pathExists(location)) {
const agents = discoverAgents(location);
// Only add agents we haven't processed yet
const newAgents = agents.filter((agent) => !processedAgents.has(agent.name));
foundAgents.push(...newAgents);
for (const agent of newAgents) processedAgents.add(agent.name);
}
}
if (foundAgents.length === 0) {
feat: Complete BMAD agent creation system with install tooling, references, and field guidance ## Overview This commit represents a complete overhaul of the BMAD agent creation system, establishing clear standards for agent development, installation workflows, and persona design. The changes span documentation, tooling, reference implementations, and field-specific guidance. ## Key Components ### 1. Agent Installation Infrastructure **New CLI Command: `agent-install`** - Interactive agent installation with persona customization - Supports Simple (single YAML), Expert (sidecar files), and Module agents - Template variable processing with Handlebars-style syntax - Automatic compilation from YAML to XML (.md) format - Manifest tracking and IDE integration (Claude Code, Cursor, Windsurf, etc.) - Source preservation in `_cfg/custom/agents/` for reinstallation **Files Created:** - `tools/cli/commands/agent-install.js` - Main CLI command - `tools/cli/lib/agent/compiler.js` - YAML to XML compilation engine - `tools/cli/lib/agent/installer.js` - Installation orchestration - `tools/cli/lib/agent/template-engine.js` - Handlebars template processing **Compiler Features:** - Auto-injects frontmatter, activation, handlers, help/exit menu items - Smart handler inclusion (only includes action/workflow/exec/tmpl handlers actually used) - Proper XML escaping and formatting - Persona name customization (e.g., "Fred the Commit Poet") ### 2. Documentation Overhaul **Deleted Bloated/Outdated Docs (2,651 lines removed):** - Old verbose architecture docs - Redundant pattern files - Outdated workflow guides **Created Focused, Type-Specific Docs:** - `src/modules/bmb/docs/understanding-agent-types.md` - Architecture vs capability distinction - `src/modules/bmb/docs/simple-agent-architecture.md` - Self-contained agents - `src/modules/bmb/docs/expert-agent-architecture.md` - Agents with sidecar files - `src/modules/bmb/docs/module-agent-architecture.md` - Workflow-integrated agents - `src/modules/bmb/docs/agent-compilation.md` - YAML → XML process - `src/modules/bmb/docs/agent-menu-patterns.md` - Menu design patterns - `src/modules/bmb/docs/index.md` - Documentation hub **Net Result:** ~1,930 line reduction while adding MORE value through focused content ### 3. Create-Agent Workflow Enhancements **Critical Persona Field Guidance Added to Step 4:** Explains how the LLM interprets each persona field when the agent activates: - **role** → "What knowledge, skills, and capabilities do I possess?" - **identity** → "What background, experience, and context shape my responses?" - **communication_style** → "What verbal patterns, word choice, quirks, and phrasing do I use?" - **principles** → "What beliefs and operating philosophy drive my choices?" **Key Insight:** `communication_style` should ONLY describe HOW the agent talks, not restate role/identity/principles. The `communication-presets.csv` provides 60 pure communication styles with NO role/identity/principles mixed in. **Files Updated:** - `src/modules/bmb/workflows/create-agent/instructions.md` - Added persona field interpretation guide - `src/modules/bmb/workflows/create-agent/brainstorm-context.md` - Refined to 137 lines - `src/modules/bmb/workflows/create-agent/communication-presets.csv` - 60 styles across 13 categories ### 4. Reference Agent Cleanup **Removed install_config Personality Bloat:** Understanding: Future installer will handle personality customization, so stripped all personality toggles from reference agents. **commit-poet.agent.yaml** (Simple Agent): - BEFORE: 36 personality combinations (3 enthusiasm × 3 depths × 4 styles) = decision fatigue - AFTER: Single concise persona with pure communication style - Changed from verbose conditionals to: "Poetic drama and flair with every turn of a phrase. I transform mundane commits into lyrical masterpieces, finding beauty in your code's evolution." - Reduction: 248 lines → 153 lines (38% reduction) **journal-keeper.agent.yaml** (Expert Agent): - Stripped install_config, simplified communication_style - Shows proper Expert agent structure with sidecar files **security-engineer.agent.yaml & trend-analyst.agent.yaml** (Module Agents): - Added header comments explaining WHY Module Agent (design intent, not just location) - Clarified: Module agents are designed FOR ecosystem integration, not capability-limited **Files Updated:** - `src/modules/bmb/reference/agents/simple-examples/commit-poet.agent.yaml` - `src/modules/bmb/reference/agents/expert-examples/journal-keeper/journal-keeper.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/security-engineer.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/trend-analyst.agent.yaml` ### 5. BMM Agent Voice Enhancement **Gave all 9 BMM agents distinct, memorable communication voices:** **Mary (analyst)** - The favorite! Changed from generic "systematic and probing" to: "Treats analysis like a treasure hunt - excited by every clue, thrilled when patterns emerge. Asks questions that spark 'aha!' moments while structuring insights with precision." **Other Notable Voices:** - **John (pm):** "Asks 'WHY?' relentlessly like a detective on a case. Direct and data-sharp, cuts through fluff to what actually matters." - **Winston (architect):** "Speaks in calm, pragmatic tones, balancing 'what could be' with 'what should be.' Champions boring technology that actually works." - **Amelia (dev):** "Ultra-succinct. Speaks in file paths and AC IDs - every statement citable. No fluff, all precision." - **Bob (sm):** "Crisp and checklist-driven. Every word has a purpose, every requirement crystal clear. Zero tolerance for ambiguity." - **Sally (ux-designer):** "Paints pictures with words, telling user stories that make you FEEL the problem. Empathetic advocate with creative storytelling flair." **Pattern Applied:** Moved behaviors from communication_style to principles, keeping communication_style as PURE verbal patterns. **Files Updated:** - `src/modules/bmm/agents/analyst.agent.yaml` - `src/modules/bmm/agents/pm.agent.yaml` - `src/modules/bmm/agents/architect.agent.yaml` - `src/modules/bmm/agents/dev.agent.yaml` - `src/modules/bmm/agents/sm.agent.yaml` - `src/modules/bmm/agents/tea.agent.yaml` - `src/modules/bmm/agents/tech-writer.agent.yaml` - `src/modules/bmm/agents/ux-designer.agent.yaml` - `src/modules/bmm/agents/frame-expert.agent.yaml` ### 6. Linting Fixes **ESLint Compliance:** - Replaced all `'utf-8'` with `'utf8'` (unicorn/text-encoding-identifier-case) - Changed `variables.hasOwnProperty(varName)` to `Object.hasOwn(variables, varName)` (unicorn/prefer-object-has-own) - Replaced `JSON.parse(JSON.stringify(...))` with `structuredClone(...)` (unicorn/prefer-structured-clone) - Fixed empty YAML mapping values in sample files **Files Fixed:** - 7 JavaScript files across agent tooling (compiler, installer, commands, IDE integration) - 1 YAML sample file ## Architecture Decisions ### Agent Types Are About Architecture, Not Capability - **Simple:** Self-contained in single YAML (NOT limited in capability) - **Expert:** Includes sidecar files (templates, docs, etc.) - **Module:** Designed for BMAD ecosystem integration (workflows, cross-agent coordination) ### Persona Field Separation Critical for LLM Interpretation The LLM needs distinct fields to understand its role: - Mixing role/identity/principles into communication_style confuses the persona - Pure communication styles (from communication-presets.csv) have ZERO role/identity/principles content - Example DON'T: "Experienced analyst who uses systematic approaches..." (mixing identity + style) - Example DO: "Systematic and probing. Structures findings hierarchically." (pure style) ### Install-Time vs Runtime Configuration - Template variables ({{var}}) resolve at compile-time - Runtime variables ({user_name}, {bmad_folder}) resolve when agent activates - Future installer will handle personality customization, so agents should ship with single default persona ## Testing - All linting passes (ESLint with max-warnings=0) - Agent compilation tested with commit-poet, journal-keeper examples - Install workflow validated with Simple and Expert agent types - Manifest tracking and IDE integration verified ## Impact This establishes BMAD as having a complete, production-ready agent creation and installation system with: - Clear documentation for all agent types - Automated compilation and installation - Strong persona design guidance - Reference implementations showing best practices - Distinct, memorable agent voices throughout BMM module Co-Authored-By: BMad Builder <builder@bmad.dev> Co-Authored-By: Mary the Analyst <analyst@bmad.dev> Co-Authored-By: Paige the Tech Writer <tech-writer@bmad.dev>
2025-11-17 22:25:15 -06:00
return results;
}
try {
const customAgentsDir = path.join(bmadDir, 'custom', 'agents');
await fs.ensureDir(customAgentsDir);
const manifestFile = path.join(bmadDir, '_cfg', 'agent-manifest.csv');
const manifestYamlFile = path.join(bmadDir, '_cfg', 'manifest.yaml');
for (const agent of foundAgents) {
feat: Complete BMAD agent creation system with install tooling, references, and field guidance ## Overview This commit represents a complete overhaul of the BMAD agent creation system, establishing clear standards for agent development, installation workflows, and persona design. The changes span documentation, tooling, reference implementations, and field-specific guidance. ## Key Components ### 1. Agent Installation Infrastructure **New CLI Command: `agent-install`** - Interactive agent installation with persona customization - Supports Simple (single YAML), Expert (sidecar files), and Module agents - Template variable processing with Handlebars-style syntax - Automatic compilation from YAML to XML (.md) format - Manifest tracking and IDE integration (Claude Code, Cursor, Windsurf, etc.) - Source preservation in `_cfg/custom/agents/` for reinstallation **Files Created:** - `tools/cli/commands/agent-install.js` - Main CLI command - `tools/cli/lib/agent/compiler.js` - YAML to XML compilation engine - `tools/cli/lib/agent/installer.js` - Installation orchestration - `tools/cli/lib/agent/template-engine.js` - Handlebars template processing **Compiler Features:** - Auto-injects frontmatter, activation, handlers, help/exit menu items - Smart handler inclusion (only includes action/workflow/exec/tmpl handlers actually used) - Proper XML escaping and formatting - Persona name customization (e.g., "Fred the Commit Poet") ### 2. Documentation Overhaul **Deleted Bloated/Outdated Docs (2,651 lines removed):** - Old verbose architecture docs - Redundant pattern files - Outdated workflow guides **Created Focused, Type-Specific Docs:** - `src/modules/bmb/docs/understanding-agent-types.md` - Architecture vs capability distinction - `src/modules/bmb/docs/simple-agent-architecture.md` - Self-contained agents - `src/modules/bmb/docs/expert-agent-architecture.md` - Agents with sidecar files - `src/modules/bmb/docs/module-agent-architecture.md` - Workflow-integrated agents - `src/modules/bmb/docs/agent-compilation.md` - YAML → XML process - `src/modules/bmb/docs/agent-menu-patterns.md` - Menu design patterns - `src/modules/bmb/docs/index.md` - Documentation hub **Net Result:** ~1,930 line reduction while adding MORE value through focused content ### 3. Create-Agent Workflow Enhancements **Critical Persona Field Guidance Added to Step 4:** Explains how the LLM interprets each persona field when the agent activates: - **role** → "What knowledge, skills, and capabilities do I possess?" - **identity** → "What background, experience, and context shape my responses?" - **communication_style** → "What verbal patterns, word choice, quirks, and phrasing do I use?" - **principles** → "What beliefs and operating philosophy drive my choices?" **Key Insight:** `communication_style` should ONLY describe HOW the agent talks, not restate role/identity/principles. The `communication-presets.csv` provides 60 pure communication styles with NO role/identity/principles mixed in. **Files Updated:** - `src/modules/bmb/workflows/create-agent/instructions.md` - Added persona field interpretation guide - `src/modules/bmb/workflows/create-agent/brainstorm-context.md` - Refined to 137 lines - `src/modules/bmb/workflows/create-agent/communication-presets.csv` - 60 styles across 13 categories ### 4. Reference Agent Cleanup **Removed install_config Personality Bloat:** Understanding: Future installer will handle personality customization, so stripped all personality toggles from reference agents. **commit-poet.agent.yaml** (Simple Agent): - BEFORE: 36 personality combinations (3 enthusiasm × 3 depths × 4 styles) = decision fatigue - AFTER: Single concise persona with pure communication style - Changed from verbose conditionals to: "Poetic drama and flair with every turn of a phrase. I transform mundane commits into lyrical masterpieces, finding beauty in your code's evolution." - Reduction: 248 lines → 153 lines (38% reduction) **journal-keeper.agent.yaml** (Expert Agent): - Stripped install_config, simplified communication_style - Shows proper Expert agent structure with sidecar files **security-engineer.agent.yaml & trend-analyst.agent.yaml** (Module Agents): - Added header comments explaining WHY Module Agent (design intent, not just location) - Clarified: Module agents are designed FOR ecosystem integration, not capability-limited **Files Updated:** - `src/modules/bmb/reference/agents/simple-examples/commit-poet.agent.yaml` - `src/modules/bmb/reference/agents/expert-examples/journal-keeper/journal-keeper.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/security-engineer.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/trend-analyst.agent.yaml` ### 5. BMM Agent Voice Enhancement **Gave all 9 BMM agents distinct, memorable communication voices:** **Mary (analyst)** - The favorite! Changed from generic "systematic and probing" to: "Treats analysis like a treasure hunt - excited by every clue, thrilled when patterns emerge. Asks questions that spark 'aha!' moments while structuring insights with precision." **Other Notable Voices:** - **John (pm):** "Asks 'WHY?' relentlessly like a detective on a case. Direct and data-sharp, cuts through fluff to what actually matters." - **Winston (architect):** "Speaks in calm, pragmatic tones, balancing 'what could be' with 'what should be.' Champions boring technology that actually works." - **Amelia (dev):** "Ultra-succinct. Speaks in file paths and AC IDs - every statement citable. No fluff, all precision." - **Bob (sm):** "Crisp and checklist-driven. Every word has a purpose, every requirement crystal clear. Zero tolerance for ambiguity." - **Sally (ux-designer):** "Paints pictures with words, telling user stories that make you FEEL the problem. Empathetic advocate with creative storytelling flair." **Pattern Applied:** Moved behaviors from communication_style to principles, keeping communication_style as PURE verbal patterns. **Files Updated:** - `src/modules/bmm/agents/analyst.agent.yaml` - `src/modules/bmm/agents/pm.agent.yaml` - `src/modules/bmm/agents/architect.agent.yaml` - `src/modules/bmm/agents/dev.agent.yaml` - `src/modules/bmm/agents/sm.agent.yaml` - `src/modules/bmm/agents/tea.agent.yaml` - `src/modules/bmm/agents/tech-writer.agent.yaml` - `src/modules/bmm/agents/ux-designer.agent.yaml` - `src/modules/bmm/agents/frame-expert.agent.yaml` ### 6. Linting Fixes **ESLint Compliance:** - Replaced all `'utf-8'` with `'utf8'` (unicorn/text-encoding-identifier-case) - Changed `variables.hasOwnProperty(varName)` to `Object.hasOwn(variables, varName)` (unicorn/prefer-object-has-own) - Replaced `JSON.parse(JSON.stringify(...))` with `structuredClone(...)` (unicorn/prefer-structured-clone) - Fixed empty YAML mapping values in sample files **Files Fixed:** - 7 JavaScript files across agent tooling (compiler, installer, commands, IDE integration) - 1 YAML sample file ## Architecture Decisions ### Agent Types Are About Architecture, Not Capability - **Simple:** Self-contained in single YAML (NOT limited in capability) - **Expert:** Includes sidecar files (templates, docs, etc.) - **Module:** Designed for BMAD ecosystem integration (workflows, cross-agent coordination) ### Persona Field Separation Critical for LLM Interpretation The LLM needs distinct fields to understand its role: - Mixing role/identity/principles into communication_style confuses the persona - Pure communication styles (from communication-presets.csv) have ZERO role/identity/principles content - Example DON'T: "Experienced analyst who uses systematic approaches..." (mixing identity + style) - Example DO: "Systematic and probing. Structures findings hierarchically." (pure style) ### Install-Time vs Runtime Configuration - Template variables ({{var}}) resolve at compile-time - Runtime variables ({user_name}, {bmad_folder}) resolve when agent activates - Future installer will handle personality customization, so agents should ship with single default persona ## Testing - All linting passes (ESLint with max-warnings=0) - Agent compilation tested with commit-poet, journal-keeper examples - Install workflow validated with Simple and Expert agent types - Manifest tracking and IDE integration verified ## Impact This establishes BMAD as having a complete, production-ready agent creation and installation system with: - Clear documentation for all agent types - Automated compilation and installation - Strong persona design guidance - Reference implementations showing best practices - Distinct, memorable agent voices throughout BMM module Co-Authored-By: BMad Builder <builder@bmad.dev> Co-Authored-By: Mary the Analyst <analyst@bmad.dev> Co-Authored-By: Paige the Tech Writer <tech-writer@bmad.dev>
2025-11-17 22:25:15 -06:00
try {
const agentConfig = loadAgentConfig(agent.yamlFile);
const finalAgentName = agent.name; // Already named correctly from save
// Determine agent type from the name (e.g., "fred-commit-poet" → "commit-poet")
let agentType = finalAgentName;
const parts = finalAgentName.split('-');
if (parts.length >= 2) {
// Try to extract type (last part or last two parts)
// For "fred-commit-poet", we want "commit-poet"
// This is heuristic - could be improved with metadata storage
agentType = parts.slice(-2).join('-'); // Take last 2 parts as type
}
// Create target directory
const agentTargetDir = path.join(customAgentsDir, finalAgentName);
await fs.ensureDir(agentTargetDir);
// Calculate paths
const compiledFileName = `${finalAgentName}.md`;
const compiledPath = path.join(agentTargetDir, compiledFileName);
const relativePath = path.relative(projectDir, compiledPath);
// Compile with embedded defaults (answers are already in defaults section)
const { xml, metadata } = compileAgent(
await fs.readFile(agent.yamlFile, 'utf8'),
agentConfig.defaults || {},
finalAgentName,
relativePath,
);
// Write compiled agent
await fs.writeFile(compiledPath, xml, 'utf8');
// Backup source YAML to _cfg/custom/agents if not already there
const cfgAgentsBackupDir = path.join(bmadDir, '_cfg', 'custom', 'agents');
await fs.ensureDir(cfgAgentsBackupDir);
const backupYamlPath = path.join(cfgAgentsBackupDir, `${finalAgentName}.agent.yaml`);
// Only backup if source is not already in backup location
if (agent.yamlFile !== backupYamlPath) {
await fs.copy(agent.yamlFile, backupYamlPath);
}
feat: Complete BMAD agent creation system with install tooling, references, and field guidance ## Overview This commit represents a complete overhaul of the BMAD agent creation system, establishing clear standards for agent development, installation workflows, and persona design. The changes span documentation, tooling, reference implementations, and field-specific guidance. ## Key Components ### 1. Agent Installation Infrastructure **New CLI Command: `agent-install`** - Interactive agent installation with persona customization - Supports Simple (single YAML), Expert (sidecar files), and Module agents - Template variable processing with Handlebars-style syntax - Automatic compilation from YAML to XML (.md) format - Manifest tracking and IDE integration (Claude Code, Cursor, Windsurf, etc.) - Source preservation in `_cfg/custom/agents/` for reinstallation **Files Created:** - `tools/cli/commands/agent-install.js` - Main CLI command - `tools/cli/lib/agent/compiler.js` - YAML to XML compilation engine - `tools/cli/lib/agent/installer.js` - Installation orchestration - `tools/cli/lib/agent/template-engine.js` - Handlebars template processing **Compiler Features:** - Auto-injects frontmatter, activation, handlers, help/exit menu items - Smart handler inclusion (only includes action/workflow/exec/tmpl handlers actually used) - Proper XML escaping and formatting - Persona name customization (e.g., "Fred the Commit Poet") ### 2. Documentation Overhaul **Deleted Bloated/Outdated Docs (2,651 lines removed):** - Old verbose architecture docs - Redundant pattern files - Outdated workflow guides **Created Focused, Type-Specific Docs:** - `src/modules/bmb/docs/understanding-agent-types.md` - Architecture vs capability distinction - `src/modules/bmb/docs/simple-agent-architecture.md` - Self-contained agents - `src/modules/bmb/docs/expert-agent-architecture.md` - Agents with sidecar files - `src/modules/bmb/docs/module-agent-architecture.md` - Workflow-integrated agents - `src/modules/bmb/docs/agent-compilation.md` - YAML → XML process - `src/modules/bmb/docs/agent-menu-patterns.md` - Menu design patterns - `src/modules/bmb/docs/index.md` - Documentation hub **Net Result:** ~1,930 line reduction while adding MORE value through focused content ### 3. Create-Agent Workflow Enhancements **Critical Persona Field Guidance Added to Step 4:** Explains how the LLM interprets each persona field when the agent activates: - **role** → "What knowledge, skills, and capabilities do I possess?" - **identity** → "What background, experience, and context shape my responses?" - **communication_style** → "What verbal patterns, word choice, quirks, and phrasing do I use?" - **principles** → "What beliefs and operating philosophy drive my choices?" **Key Insight:** `communication_style` should ONLY describe HOW the agent talks, not restate role/identity/principles. The `communication-presets.csv` provides 60 pure communication styles with NO role/identity/principles mixed in. **Files Updated:** - `src/modules/bmb/workflows/create-agent/instructions.md` - Added persona field interpretation guide - `src/modules/bmb/workflows/create-agent/brainstorm-context.md` - Refined to 137 lines - `src/modules/bmb/workflows/create-agent/communication-presets.csv` - 60 styles across 13 categories ### 4. Reference Agent Cleanup **Removed install_config Personality Bloat:** Understanding: Future installer will handle personality customization, so stripped all personality toggles from reference agents. **commit-poet.agent.yaml** (Simple Agent): - BEFORE: 36 personality combinations (3 enthusiasm × 3 depths × 4 styles) = decision fatigue - AFTER: Single concise persona with pure communication style - Changed from verbose conditionals to: "Poetic drama and flair with every turn of a phrase. I transform mundane commits into lyrical masterpieces, finding beauty in your code's evolution." - Reduction: 248 lines → 153 lines (38% reduction) **journal-keeper.agent.yaml** (Expert Agent): - Stripped install_config, simplified communication_style - Shows proper Expert agent structure with sidecar files **security-engineer.agent.yaml & trend-analyst.agent.yaml** (Module Agents): - Added header comments explaining WHY Module Agent (design intent, not just location) - Clarified: Module agents are designed FOR ecosystem integration, not capability-limited **Files Updated:** - `src/modules/bmb/reference/agents/simple-examples/commit-poet.agent.yaml` - `src/modules/bmb/reference/agents/expert-examples/journal-keeper/journal-keeper.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/security-engineer.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/trend-analyst.agent.yaml` ### 5. BMM Agent Voice Enhancement **Gave all 9 BMM agents distinct, memorable communication voices:** **Mary (analyst)** - The favorite! Changed from generic "systematic and probing" to: "Treats analysis like a treasure hunt - excited by every clue, thrilled when patterns emerge. Asks questions that spark 'aha!' moments while structuring insights with precision." **Other Notable Voices:** - **John (pm):** "Asks 'WHY?' relentlessly like a detective on a case. Direct and data-sharp, cuts through fluff to what actually matters." - **Winston (architect):** "Speaks in calm, pragmatic tones, balancing 'what could be' with 'what should be.' Champions boring technology that actually works." - **Amelia (dev):** "Ultra-succinct. Speaks in file paths and AC IDs - every statement citable. No fluff, all precision." - **Bob (sm):** "Crisp and checklist-driven. Every word has a purpose, every requirement crystal clear. Zero tolerance for ambiguity." - **Sally (ux-designer):** "Paints pictures with words, telling user stories that make you FEEL the problem. Empathetic advocate with creative storytelling flair." **Pattern Applied:** Moved behaviors from communication_style to principles, keeping communication_style as PURE verbal patterns. **Files Updated:** - `src/modules/bmm/agents/analyst.agent.yaml` - `src/modules/bmm/agents/pm.agent.yaml` - `src/modules/bmm/agents/architect.agent.yaml` - `src/modules/bmm/agents/dev.agent.yaml` - `src/modules/bmm/agents/sm.agent.yaml` - `src/modules/bmm/agents/tea.agent.yaml` - `src/modules/bmm/agents/tech-writer.agent.yaml` - `src/modules/bmm/agents/ux-designer.agent.yaml` - `src/modules/bmm/agents/frame-expert.agent.yaml` ### 6. Linting Fixes **ESLint Compliance:** - Replaced all `'utf-8'` with `'utf8'` (unicorn/text-encoding-identifier-case) - Changed `variables.hasOwnProperty(varName)` to `Object.hasOwn(variables, varName)` (unicorn/prefer-object-has-own) - Replaced `JSON.parse(JSON.stringify(...))` with `structuredClone(...)` (unicorn/prefer-structured-clone) - Fixed empty YAML mapping values in sample files **Files Fixed:** - 7 JavaScript files across agent tooling (compiler, installer, commands, IDE integration) - 1 YAML sample file ## Architecture Decisions ### Agent Types Are About Architecture, Not Capability - **Simple:** Self-contained in single YAML (NOT limited in capability) - **Expert:** Includes sidecar files (templates, docs, etc.) - **Module:** Designed for BMAD ecosystem integration (workflows, cross-agent coordination) ### Persona Field Separation Critical for LLM Interpretation The LLM needs distinct fields to understand its role: - Mixing role/identity/principles into communication_style confuses the persona - Pure communication styles (from communication-presets.csv) have ZERO role/identity/principles content - Example DON'T: "Experienced analyst who uses systematic approaches..." (mixing identity + style) - Example DO: "Systematic and probing. Structures findings hierarchically." (pure style) ### Install-Time vs Runtime Configuration - Template variables ({{var}}) resolve at compile-time - Runtime variables ({user_name}, {bmad_folder}) resolve when agent activates - Future installer will handle personality customization, so agents should ship with single default persona ## Testing - All linting passes (ESLint with max-warnings=0) - Agent compilation tested with commit-poet, journal-keeper examples - Install workflow validated with Simple and Expert agent types - Manifest tracking and IDE integration verified ## Impact This establishes BMAD as having a complete, production-ready agent creation and installation system with: - Clear documentation for all agent types - Automated compilation and installation - Strong persona design guidance - Reference implementations showing best practices - Distinct, memorable agent voices throughout BMM module Co-Authored-By: BMad Builder <builder@bmad.dev> Co-Authored-By: Mary the Analyst <analyst@bmad.dev> Co-Authored-By: Paige the Tech Writer <tech-writer@bmad.dev>
2025-11-17 22:25:15 -06:00
// Copy sidecar files if expert agent
if (agent.hasSidecar && agent.type === 'expert') {
const { copySidecarFiles } = require('../../../lib/agent/installer');
copySidecarFiles(agent.path, agentTargetDir, agent.yamlFile);
}
// Update manifest CSV
if (await fs.pathExists(manifestFile)) {
feat: add custom agents and quick-flow workflows, remove tech-spec track Major Changes: - Add sample custom agents demonstrating installable agent system - commit-poet: Generates semantic commit messages (BMAD Method repo sample) - toolsmith: Development tooling expert with knowledge base covering bundlers, deployment, docs, installers, modules, and tests (BMAD Method repo sample) - Both agents demonstrate custom agent architecture and are installable to projects via BMAD installer system - Include comprehensive installation guides and sidecar knowledge bases - Add bmad-quick-flow methodology for rapid development - create-tech-spec: Direct technical specification workflow - quick-dev: Flexible execution workflow supporting both tech-spec-driven and direct instruction development - quick-flow-solo-dev (Barry): 1 man show agent specialized in bmad-quick-flow methodology - Comprehensive documentation for quick-flow approach and solo development - Remove deprecated tech-spec workflow track - Delete entire tech-spec workflow directory and templates - Remove quick-spec-flow.md documentation (replaced by quick-flow docs) - Clean up unused epic and story templates - Fix custom agent installation across IDE installers - Repair antigravity and multiple IDE installers to properly support custom agents - Enable custom agent installation via quick installer, agent installer, regular installer, and special agent installer - All installation methods now accessible via npx with full documentation Infrastructure: - Update BMM module configurations and team setups - Modify workflow status paths to support quick-flow integration - Reorganize documentation with new agent and workflow guides - Add custom/ directory for user customizations - Update platform codes and installer configurations
2025-11-23 08:50:36 -06:00
// Preserve YAML metadata for persona name, but override id for filename
const manifestMetadata = {
...metadata,
id: relativePath, // Use the compiled agent path for id
name: metadata.name || finalAgentName, // Use YAML metadata.name (persona name) or fallback
title: metadata.title, // Use YAML title
icon: metadata.icon, // Use YAML icon
};
const manifestData = extractManifestData(xml, manifestMetadata, relativePath, 'custom');
manifestData.name = finalAgentName; // Use filename for the name field
feat: Complete BMAD agent creation system with install tooling, references, and field guidance ## Overview This commit represents a complete overhaul of the BMAD agent creation system, establishing clear standards for agent development, installation workflows, and persona design. The changes span documentation, tooling, reference implementations, and field-specific guidance. ## Key Components ### 1. Agent Installation Infrastructure **New CLI Command: `agent-install`** - Interactive agent installation with persona customization - Supports Simple (single YAML), Expert (sidecar files), and Module agents - Template variable processing with Handlebars-style syntax - Automatic compilation from YAML to XML (.md) format - Manifest tracking and IDE integration (Claude Code, Cursor, Windsurf, etc.) - Source preservation in `_cfg/custom/agents/` for reinstallation **Files Created:** - `tools/cli/commands/agent-install.js` - Main CLI command - `tools/cli/lib/agent/compiler.js` - YAML to XML compilation engine - `tools/cli/lib/agent/installer.js` - Installation orchestration - `tools/cli/lib/agent/template-engine.js` - Handlebars template processing **Compiler Features:** - Auto-injects frontmatter, activation, handlers, help/exit menu items - Smart handler inclusion (only includes action/workflow/exec/tmpl handlers actually used) - Proper XML escaping and formatting - Persona name customization (e.g., "Fred the Commit Poet") ### 2. Documentation Overhaul **Deleted Bloated/Outdated Docs (2,651 lines removed):** - Old verbose architecture docs - Redundant pattern files - Outdated workflow guides **Created Focused, Type-Specific Docs:** - `src/modules/bmb/docs/understanding-agent-types.md` - Architecture vs capability distinction - `src/modules/bmb/docs/simple-agent-architecture.md` - Self-contained agents - `src/modules/bmb/docs/expert-agent-architecture.md` - Agents with sidecar files - `src/modules/bmb/docs/module-agent-architecture.md` - Workflow-integrated agents - `src/modules/bmb/docs/agent-compilation.md` - YAML → XML process - `src/modules/bmb/docs/agent-menu-patterns.md` - Menu design patterns - `src/modules/bmb/docs/index.md` - Documentation hub **Net Result:** ~1,930 line reduction while adding MORE value through focused content ### 3. Create-Agent Workflow Enhancements **Critical Persona Field Guidance Added to Step 4:** Explains how the LLM interprets each persona field when the agent activates: - **role** → "What knowledge, skills, and capabilities do I possess?" - **identity** → "What background, experience, and context shape my responses?" - **communication_style** → "What verbal patterns, word choice, quirks, and phrasing do I use?" - **principles** → "What beliefs and operating philosophy drive my choices?" **Key Insight:** `communication_style` should ONLY describe HOW the agent talks, not restate role/identity/principles. The `communication-presets.csv` provides 60 pure communication styles with NO role/identity/principles mixed in. **Files Updated:** - `src/modules/bmb/workflows/create-agent/instructions.md` - Added persona field interpretation guide - `src/modules/bmb/workflows/create-agent/brainstorm-context.md` - Refined to 137 lines - `src/modules/bmb/workflows/create-agent/communication-presets.csv` - 60 styles across 13 categories ### 4. Reference Agent Cleanup **Removed install_config Personality Bloat:** Understanding: Future installer will handle personality customization, so stripped all personality toggles from reference agents. **commit-poet.agent.yaml** (Simple Agent): - BEFORE: 36 personality combinations (3 enthusiasm × 3 depths × 4 styles) = decision fatigue - AFTER: Single concise persona with pure communication style - Changed from verbose conditionals to: "Poetic drama and flair with every turn of a phrase. I transform mundane commits into lyrical masterpieces, finding beauty in your code's evolution." - Reduction: 248 lines → 153 lines (38% reduction) **journal-keeper.agent.yaml** (Expert Agent): - Stripped install_config, simplified communication_style - Shows proper Expert agent structure with sidecar files **security-engineer.agent.yaml & trend-analyst.agent.yaml** (Module Agents): - Added header comments explaining WHY Module Agent (design intent, not just location) - Clarified: Module agents are designed FOR ecosystem integration, not capability-limited **Files Updated:** - `src/modules/bmb/reference/agents/simple-examples/commit-poet.agent.yaml` - `src/modules/bmb/reference/agents/expert-examples/journal-keeper/journal-keeper.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/security-engineer.agent.yaml` - `src/modules/bmb/reference/agents/module-examples/trend-analyst.agent.yaml` ### 5. BMM Agent Voice Enhancement **Gave all 9 BMM agents distinct, memorable communication voices:** **Mary (analyst)** - The favorite! Changed from generic "systematic and probing" to: "Treats analysis like a treasure hunt - excited by every clue, thrilled when patterns emerge. Asks questions that spark 'aha!' moments while structuring insights with precision." **Other Notable Voices:** - **John (pm):** "Asks 'WHY?' relentlessly like a detective on a case. Direct and data-sharp, cuts through fluff to what actually matters." - **Winston (architect):** "Speaks in calm, pragmatic tones, balancing 'what could be' with 'what should be.' Champions boring technology that actually works." - **Amelia (dev):** "Ultra-succinct. Speaks in file paths and AC IDs - every statement citable. No fluff, all precision." - **Bob (sm):** "Crisp and checklist-driven. Every word has a purpose, every requirement crystal clear. Zero tolerance for ambiguity." - **Sally (ux-designer):** "Paints pictures with words, telling user stories that make you FEEL the problem. Empathetic advocate with creative storytelling flair." **Pattern Applied:** Moved behaviors from communication_style to principles, keeping communication_style as PURE verbal patterns. **Files Updated:** - `src/modules/bmm/agents/analyst.agent.yaml` - `src/modules/bmm/agents/pm.agent.yaml` - `src/modules/bmm/agents/architect.agent.yaml` - `src/modules/bmm/agents/dev.agent.yaml` - `src/modules/bmm/agents/sm.agent.yaml` - `src/modules/bmm/agents/tea.agent.yaml` - `src/modules/bmm/agents/tech-writer.agent.yaml` - `src/modules/bmm/agents/ux-designer.agent.yaml` - `src/modules/bmm/agents/frame-expert.agent.yaml` ### 6. Linting Fixes **ESLint Compliance:** - Replaced all `'utf-8'` with `'utf8'` (unicorn/text-encoding-identifier-case) - Changed `variables.hasOwnProperty(varName)` to `Object.hasOwn(variables, varName)` (unicorn/prefer-object-has-own) - Replaced `JSON.parse(JSON.stringify(...))` with `structuredClone(...)` (unicorn/prefer-structured-clone) - Fixed empty YAML mapping values in sample files **Files Fixed:** - 7 JavaScript files across agent tooling (compiler, installer, commands, IDE integration) - 1 YAML sample file ## Architecture Decisions ### Agent Types Are About Architecture, Not Capability - **Simple:** Self-contained in single YAML (NOT limited in capability) - **Expert:** Includes sidecar files (templates, docs, etc.) - **Module:** Designed for BMAD ecosystem integration (workflows, cross-agent coordination) ### Persona Field Separation Critical for LLM Interpretation The LLM needs distinct fields to understand its role: - Mixing role/identity/principles into communication_style confuses the persona - Pure communication styles (from communication-presets.csv) have ZERO role/identity/principles content - Example DON'T: "Experienced analyst who uses systematic approaches..." (mixing identity + style) - Example DO: "Systematic and probing. Structures findings hierarchically." (pure style) ### Install-Time vs Runtime Configuration - Template variables ({{var}}) resolve at compile-time - Runtime variables ({user_name}, {bmad_folder}) resolve when agent activates - Future installer will handle personality customization, so agents should ship with single default persona ## Testing - All linting passes (ESLint with max-warnings=0) - Agent compilation tested with commit-poet, journal-keeper examples - Install workflow validated with Simple and Expert agent types - Manifest tracking and IDE integration verified ## Impact This establishes BMAD as having a complete, production-ready agent creation and installation system with: - Clear documentation for all agent types - Automated compilation and installation - Strong persona design guidance - Reference implementations showing best practices - Distinct, memorable agent voices throughout BMM module Co-Authored-By: BMad Builder <builder@bmad.dev> Co-Authored-By: Mary the Analyst <analyst@bmad.dev> Co-Authored-By: Paige the Tech Writer <tech-writer@bmad.dev>
2025-11-17 22:25:15 -06:00
manifestData.path = relativePath;
addToManifest(manifestFile, manifestData);
}
// Create IDE slash commands (async function)
await createIdeSlashCommands(projectDir, finalAgentName, relativePath, metadata);
// Update manifest.yaml
if (await fs.pathExists(manifestYamlFile)) {
updateManifestYaml(manifestYamlFile, finalAgentName, agentType);
}
results.count++;
results.agents.push(finalAgentName);
} catch (agentError) {
console.log(chalk.yellow(` ⚠️ Failed to reinstall ${agent.name}: ${agentError.message}`));
}
}
} catch (error) {
console.log(chalk.yellow(` ⚠️ Error reinstalling custom agents: ${error.message}`));
}
return results;
}
/**
* Copy IDE-specific documentation to BMAD docs
* @param {Array} ides - List of selected IDEs
* @param {string} bmadDir - BMAD installation directory
*/
async copyIdeDocumentation(ides, bmadDir) {
const docsDir = path.join(bmadDir, 'docs');
await fs.ensureDir(docsDir);
for (const ide of ides) {
const sourceDocPath = path.join(getProjectRoot(), 'docs', 'ide-info', `${ide}.md`);
const targetDocPath = path.join(docsDir, `${ide}-instructions.md`);
if (await fs.pathExists(sourceDocPath)) {
await this.copyFileWithPlaceholderReplacement(sourceDocPath, targetDocPath, this.bmadFolderName || 'bmad');
}
}
}
/**
* Scan for legacy/obsolete files in BMAD installation
* @param {string} bmadDir - BMAD installation directory
* @returns {Object} Categorized files for cleanup
*/
async scanForLegacyFiles(bmadDir) {
const legacyFiles = {
backup: [],
documentation: [],
deprecated_task: [],
unknown: [],
};
try {
// Load files manifest to understand what should exist
const manifestPath = path.join(bmadDir, 'files-manifest.csv');
const manifestFiles = new Set();
if (await fs.pathExists(manifestPath)) {
const manifestContent = await fs.readFile(manifestPath, 'utf8');
const lines = manifestContent.split('\n').slice(1); // Skip header
for (const line of lines) {
if (line.trim()) {
const relativePath = line.split(',')[0];
if (relativePath) {
manifestFiles.add(relativePath);
}
}
}
}
// Scan all files recursively
const allFiles = await this.getAllFiles(bmadDir);
for (const filePath of allFiles) {
const relativePath = path.relative(bmadDir, filePath);
// Skip expected files
if (this.isExpectedFile(relativePath, manifestFiles)) {
continue;
}
// Categorize legacy files
if (relativePath.endsWith('.bak')) {
legacyFiles.backup.push({
path: filePath,
relativePath: relativePath,
size: (await fs.stat(filePath)).size,
mtime: (await fs.stat(filePath)).mtime,
});
} else if (this.isDocumentationFile(relativePath)) {
legacyFiles.documentation.push({
path: filePath,
relativePath: relativePath,
size: (await fs.stat(filePath)).size,
mtime: (await fs.stat(filePath)).mtime,
});
} else if (this.isDeprecatedTaskFile(relativePath)) {
const suggestedAlternative = this.suggestAlternative(relativePath);
legacyFiles.deprecated_task.push({
path: filePath,
relativePath: relativePath,
size: (await fs.stat(filePath)).size,
mtime: (await fs.stat(filePath)).mtime,
suggestedAlternative,
});
} else {
legacyFiles.unknown.push({
path: filePath,
relativePath: relativePath,
size: (await fs.stat(filePath)).size,
mtime: (await fs.stat(filePath)).mtime,
});
}
}
} catch (error) {
console.warn(`Warning: Could not scan for legacy files: ${error.message}`);
}
return legacyFiles;
}
/**
* Get all files in directory recursively
* @param {string} dir - Directory to scan
* @returns {Array} Array of file paths
*/
async getAllFiles(dir) {
const files = [];
async function scan(currentDir) {
const entries = await fs.readdir(currentDir);
for (const entry of entries) {
const fullPath = path.join(currentDir, entry);
const stat = await fs.stat(fullPath);
if (stat.isDirectory()) {
// Skip certain directories
if (!['node_modules', '.git', 'dist', 'build'].includes(entry)) {
await scan(fullPath);
}
} else {
files.push(fullPath);
}
}
}
await scan(dir);
return files;
}
/**
* Check if file is expected in installation
* @param {string} relativePath - Relative path from BMAD dir
* @param {Set} manifestFiles - Files from manifest
* @returns {boolean} True if expected file
*/
isExpectedFile(relativePath, manifestFiles) {
// Core files in manifest
if (manifestFiles.has(relativePath)) {
return true;
}
// Configuration files
if (relativePath.startsWith('_cfg/') || relativePath === 'config.yaml') {
return true;
}
// Custom files
if (relativePath.startsWith('custom/') || relativePath === 'manifest.yaml') {
return true;
}
// Generated files
if (relativePath === 'manifest.csv' || relativePath === 'files-manifest.csv') {
return true;
}
// IDE-specific files
const ides = ['vscode', 'cursor', 'windsurf', 'claude-code', 'github-copilot', 'zsh', 'bash', 'fish'];
if (ides.some((ide) => relativePath.includes(ide))) {
return true;
}
// BMAD MODULE STRUCTURES - recognize valid module content
const modulePrefixes = ['bmb/', 'bmm/', 'cis/', 'core/', 'bmgd/'];
const validExtensions = ['.yaml', '.yml', '.json', '.csv', '.md', '.xml', '.svg', '.png', '.jpg', '.gif', '.excalidraw', '.js'];
// Check if this file is in a recognized module directory
for (const modulePrefix of modulePrefixes) {
if (relativePath.startsWith(modulePrefix)) {
// Check if it has a valid extension
const hasValidExtension = validExtensions.some((ext) => relativePath.endsWith(ext));
if (hasValidExtension) {
return true;
}
}
}
// Special case for core module resources
if (relativePath.startsWith('core/resources/')) {
return true;
}
// Special case for docs directory
if (relativePath.startsWith('docs/')) {
return true;
}
return false;
}
/**
* Check if file is documentation
* @param {string} relativePath - Relative path
* @returns {boolean} True if documentation
*/
isDocumentationFile(relativePath) {
const docExtensions = ['.md', '.txt', '.pdf'];
const docPatterns = ['docs/', 'README', 'CHANGELOG', 'LICENSE'];
return docExtensions.some((ext) => relativePath.endsWith(ext)) || docPatterns.some((pattern) => relativePath.includes(pattern));
}
/**
* Check if file is deprecated task file
* @param {string} relativePath - Relative path
* @returns {boolean} True if deprecated
*/
isDeprecatedTaskFile(relativePath) {
// Known deprecated files
const deprecatedFiles = ['adv-elicit-methods.csv', 'game-resources.json', 'ux-workflow.json'];
return deprecatedFiles.some((dep) => relativePath.includes(dep));
}
/**
* Suggest alternative for deprecated file
* @param {string} relativePath - Deprecated file path
* @returns {string} Suggested alternative
*/
suggestAlternative(relativePath) {
const alternatives = {
'adv-elicit-methods.csv': 'Use the new structured workflows in src/modules/',
'game-resources.json': 'Resources are now integrated into modules',
'ux-workflow.json': 'UX workflows are now in src/modules/bmm/workflows/',
};
for (const [deprecated, alternative] of Object.entries(alternatives)) {
if (relativePath.includes(deprecated)) {
return alternative;
}
}
return 'Check src/modules/ for new alternatives';
}
/**
* Perform interactive cleanup of legacy files
* @param {string} bmadDir - BMAD installation directory
* @param {boolean} skipInteractive - Skip interactive prompts
* @returns {Object} Cleanup results
*/
async performCleanup(bmadDir, skipInteractive = false) {
const inquirer = require('inquirer');
const yaml = require('js-yaml');
// Load user retention preferences
const retentionPath = path.join(bmadDir, '_cfg', 'user-retained-files.yaml');
let retentionData = { retainedFiles: [], history: [] };
if (await fs.pathExists(retentionPath)) {
const retentionContent = await fs.readFile(retentionPath, 'utf8');
retentionData = yaml.load(retentionContent) || retentionData;
}
// Scan for legacy files
const legacyFiles = await this.scanForLegacyFiles(bmadDir);
const allLegacyFiles = [...legacyFiles.backup, ...legacyFiles.documentation, ...legacyFiles.deprecated_task, ...legacyFiles.unknown];
if (allLegacyFiles.length === 0) {
return { deleted: 0, retained: 0, message: 'No legacy files found' };
}
let deletedCount = 0;
let retainedCount = 0;
const filesToDelete = [];
if (skipInteractive) {
// Auto-delete all non-retained files
for (const file of allLegacyFiles) {
if (!retentionData.retainedFiles.includes(file.relativePath)) {
filesToDelete.push(file);
}
}
} else {
// Interactive cleanup
console.log(chalk.cyan('\n🧹 Legacy File Cleanup\n'));
console.log(chalk.dim('The following obsolete files were found:\n'));
// Group files by category
const categories = [];
if (legacyFiles.backup.length > 0) {
categories.push({ name: 'Backup Files (.bak)', files: legacyFiles.backup });
}
if (legacyFiles.documentation.length > 0) {
categories.push({ name: 'Documentation', files: legacyFiles.documentation });
}
if (legacyFiles.deprecated_task.length > 0) {
categories.push({ name: 'Deprecated Task Files', files: legacyFiles.deprecated_task });
}
if (legacyFiles.unknown.length > 0) {
categories.push({ name: 'Unknown Files', files: legacyFiles.unknown });
}
for (const category of categories) {
console.log(chalk.yellow(`${category.name}:`));
for (const file of category.files) {
const size = (file.size / 1024).toFixed(1);
const date = file.mtime.toLocaleDateString();
let line = ` - ${file.relativePath} (${size}KB, ${date})`;
if (file.suggestedAlternative) {
line += chalk.dim(`${file.suggestedAlternative}`);
}
console.log(chalk.dim(line));
}
console.log();
}
const prompt = await inquirer.prompt([
{
type: 'confirm',
name: 'proceed',
message: 'Would you like to review these files for cleanup?',
default: true,
},
]);
if (!prompt.proceed) {
return { deleted: 0, retained: allLegacyFiles.length, message: 'Cleanup cancelled by user' };
}
// Show selection interface
const selectionPrompt = await inquirer.prompt([
{
type: 'checkbox',
name: 'filesToDelete',
message: 'Select files to delete (use SPACEBAR to select, ENTER to continue):',
choices: allLegacyFiles.map((file) => {
const isRetained = retentionData.retainedFiles.includes(file.relativePath);
const description = `${file.relativePath} (${(file.size / 1024).toFixed(1)}KB)`;
return {
name: description,
value: file,
checked: !isRetained && !file.relativePath.includes('.bak'),
};
}),
pageSize: Math.min(allLegacyFiles.length, 15),
},
]);
filesToDelete.push(...selectionPrompt.filesToDelete);
}
// Delete selected files
for (const file of filesToDelete) {
try {
await fs.remove(file.path);
deletedCount++;
} catch (error) {
console.warn(`Warning: Could not delete ${file.relativePath}: ${error.message}`);
}
}
// Count retained files
retainedCount = allLegacyFiles.length - deletedCount;
// Update retention data
const newlyRetained = allLegacyFiles.filter((f) => !filesToDelete.includes(f)).map((f) => f.relativePath);
retentionData.retainedFiles = [...new Set([...retentionData.retainedFiles, ...newlyRetained])];
retentionData.history.push({
date: new Date().toISOString(),
deleted: deletedCount,
retained: retainedCount,
files: filesToDelete.map((f) => f.relativePath),
});
// Save retention data
await fs.ensureDir(path.dirname(retentionPath));
await fs.writeFile(retentionPath, yaml.dump(retentionData), 'utf8');
return { deleted: deletedCount, retained: retainedCount };
}
}
module.exports = { Installer };