ai-workflow-init 6.2.3 → 6.3.1

package/.claude/settings.local.json

@@ -25,15 +25,6 @@
       }
     ],
     "PostToolUse": [
-      {
-        "matcher": "Write",
-        "hooks": [
-          {
-            "type": "prompt",
-            "prompt": "If the file path contains 'docs/ai/planning/feature-' and ends with '.md', validate this planning document has all required sections: 1. Goal & Acceptance Criteria, 2. Risks & Assumptions, 3. Definition of Done, 4. Implementation Plan (with Summary and Phases), 5. Follow-ups. Return JSON: {\"valid\": true/false, \"missing\": [list of missing sections], \"message\": \"brief validation result\"}. If file path doesn't match, return {\"valid\": true, \"message\": \"skipped - not a planning doc\"}"
-          }
-        ]
-      },
       {
         "matcher": "Edit",
         "hooks": [
@@ -44,26 +35,6 @@
         ]
       }
     ],
-    "PreToolUse": [
-      {
-        "matcher": "Bash",
-        "hooks": [
-          {
-            "type": "prompt",
-            "prompt": "GIT SAFETY CHECK: If command contains 'git ' (git commands like git add, git commit, git push, git pull, git checkout, git branch, git merge, git rebase, git reset, git stash, etc.): Check the ORIGINAL USER PROMPT - did the user EXPLICITLY request git operations? Keywords indicating user wants git: 'commit', 'push', 'git', 'version control', 'save changes', 'create branch'. If user did NOT explicitly request git operations, BLOCK with message 'Git operations require explicit user request. Ask user first.' Return JSON: {\"allow\": true/false, \"reason\": \"explanation\"}. If command does not contain git, return {\"allow\": true, \"reason\": \"not a git command\"}"
-          }
-        ]
-      },
-      {
-        "matcher": "Write|Edit",
-        "hooks": [
-          {
-            "type": "prompt",
-            "prompt": "SECURITY CHECK: If file path matches any sensitive pattern: '.env', '.env.*', 'credentials', 'secrets', 'api_key', 'apikey', 'password', 'private_key', 'token', '.pem', '.key', 'auth.json', 'config/prod': WARN user with message '⚠️ SENSITIVE FILE: [filename] - Are you sure you want to modify this file?'. Return JSON: {\"isSensitive\": true/false, \"warning\": \"message if sensitive\", \"allow\": true}. Always allow but warn for sensitive files."
-          }
-        ]
-      }
-    ],
     "Stop": [
       {
         "matcher": "",

package/AGENTS.md

@@ -2,123 +2,89 @@
 
 ## Core Coding Philosophy
 
-Apply these principles when providing solutions, generating code, or making technical decisions:
-
-### 1. Simplicity First
-
-- Choose the simplest solution that meets requirements
-- Avoid over-engineering and unnecessary abstractions
-- Simple code = fewer bugs, easier to maintain, easier to understand
-- Ask: "Can this be done more simply while still working?"
-- Complexity is a last resort, not a first choice
+### 1. Simplicity First (with Strategic Exceptions)
+- **Default: Keep it simple**
+  - Choose simplest solution that meets requirements
+  - Avoid over-engineering and unnecessary abstractions
+  - Don't build for hypothetical futures
+
+- **Think ahead ONLY for:**
+  - **Security**: Input validation, authentication, authorization
+  - **Performance**: Scalability bottlenecks, query optimization
+  - All other cases → Choose simplicity
+
+- **Examples:**
+  - ✅ Use array methods instead of custom loops
+  - ✅ Add input validation for user data (security)
+  - ✅ Consider pagination for large datasets (performance)
+  - ❌ Don't create abstractions for one-time operations
 
 ### 2. Deep Understanding
-
-- Understand requirements fully before planning or coding
-- If unclear about requirement, flow, edge cases, or expected behavior → Ask the user
+- If unclear about requirements, edge cases, or expected behavior → **Ask first**
 - Never assume or guess - clarification prevents wasted effort
-- Ask questions like:
+- Key questions:
   - "What should happen when X occurs?"
-  - "How should the system behave if Y fails?"
   - "Is this the expected flow: A → B → C?"
 
-### 3. Multiple Options
-
-- Provide 2-5 solution options for each problem when appropriate
-- Present trade-offs clearly: pros/cons, complexity, performance, maintainability
+### 3. Multiple Options When Appropriate
+- Present 2-3 solution options with clear trade-offs
 - Format: "Option 1: [approach] - Pros: [...] Cons: [...]"
-- Let user choose based on their context and priorities
-- Not every problem has one "best" solution
-
-### 4. Think Ahead
-
-- While keeping solutions simple, consider future implications:
-  - How will this scale with more data/users?
-  - What if requirements change slightly?
-  - Are there security vulnerabilities?
-  - What are performance bottlenecks?
-- Balance: Simple now + adaptable for reasonable future needs
-- Do not build for hypothetical futures, but be aware of likely changes
-
-**Philosophy in practice:**
-
-- Simplicity: Use built-in array methods instead of custom loop logic
-- Deep Understanding: "Should this API return 404 or 400 for invalid IDs?"
-- Multiple Options: "We can use: 1) localStorage (simple), 2) IndexedDB (scalable), or 3) Backend API (persistent)"
-- Think Ahead: "This works for 100 items, but consider pagination for 10,000+"
+- Let user choose based on their priorities
 
 ---
 
-## Core Workflow: Plan → Implement → Test → Review
-
-### Workflow Alignment
+## Workflow Guidelines
 
-- **Plan:** Create feature planning doc at `docs/ai/planning/feature-{name}.md` before coding. Do not start until planning exists and is agreed.
-- **Implement:** Provide 1-3 sentence status updates before operations. Use file editing tools, not copy-paste. Update checkboxes `[ ]` → `[x]` in planning doc.
-- **Test:** Run linter/type-check/build on changed files after each batch. Auto-fix issues (up to 3 attempts) before asking for help.
-- **Review:** When complete, validate against planning doc acceptance criteria and CODE_CONVENTIONS.md.
+**Tooling:**
+- Prefer semantic search; grep for exact matches only
+- Run independent operations in parallel
+- Search for files matching patterns when exploring codebase
+- Search content for patterns when looking for specific code
 
-## File Structure
+**Communication:**
+- Use Markdown minimally; backticks for `files/functions/classes`
+- Mirror user's language; code/comments in English
+- Status updates before/after key actions
 
-### Planning Documents
+**Code Presentation:**
+- Existing code: `startLine:endLine:filepath`
+- New code: fenced blocks with language tag
 
-- Location: `docs/ai/planning/feature-{name}.md` (kebab-case)
-- Must include: Goal, Acceptance Criteria (GWT), Risks, Implementation Phases, Follow-ups
-- Template: `docs/ai/planning/feature-template.md`
+**TODO Management:**
+- Create todos for medium/large tasks (≤14 words, verb-led)
+- Keep ONE `in_progress` item only
+- Update immediately; mark completed when done
 
-### Project Standards
-
-- Coding conventions: `docs/ai/project/CODE_CONVENTIONS.md`
-- Architecture guide: `docs/ai/project/PROJECT_STRUCTURE.md`
-- Language templates: `docs/ai/project/template-convention/`
-
-## Tooling Strategy
-
-- Prefer semantic search across codebase; use grep only for exact matches
-- Default to parallel execution for independent operations
-- Quality tools: ESLint, TypeScript, Prettier (auto-format/auto-fix when possible)
-
-## Communication
-
-- Use Markdown only when necessary; backticks for `files/dirs/functions/classes`
-- Status updates before/after important actions
-- Mirror user's chat language; code/comments always in English
-
-## Code Presentation
+---
 
-- Existing code: cite with `startLine:endLine:filepath` (no language tag)
-- New code: fenced blocks with language tag, no line numbers
+## Skill Reporting (MANDATORY)
 
-## TODO Policy
+**CRITICAL REQUIREMENT - ALWAYS follow this:**
 
-- For medium/large tasks: create todos (≤14 words, verb-led)
-- Keep only ONE `in_progress` item
-- Update immediately after progress; mark completed upon finish
+At the START of EVERY response, BEFORE any other content, report skills:
 
-## Git Workflow
+```
+📚 Skills: skill-name-1, skill-name-2
+```
 
-- Feature branches: `feature/{name}` (match planning doc name)
-- Commit format: `[phase] brief description`
-- Examples: `[planning] create user auth plan`, `[phase-1] implement database schema`
+**Rules:**
+- If skills were activated → List them
+- If NO skills activated → Write: `📚 Skills: none`
+- This line MUST appear in EVERY response, no exceptions
+- Place BEFORE greeting, explanation, or any other content
 
-## Slash Commands
+**Example responses:**
 
-- `/create-plan` - Generate planning doc
-- `/execute-plan` - Implement tasks from planning doc
-- `/modify-plan` - Modify plan after implementation
-- `/code-review` - Validate against standards
-- `/generate-standards` - Update CODE_CONVENTIONS.md
-- `/writing-test` - Generate tests from acceptance criteria
-- `/init-chat` - Load project rules (AGENTS.md)
+```
+📚 Skills: frontend-design-fundamentals, frontend-design-theme-factory
 
-## Quality Gates
+I'll help you create a modern login page...
+```
 
-### Before marking task complete:
+```
+📚 Skills: none
 
-- Code matches planning doc specification
-- Linting passes (no warnings)
-- Type checking passes (if applicable)
-- Build succeeds (if applicable)
-- Checkbox updated in planning doc: `[x]`
+Sure, I can help you fix that bug...
+```
 
----
+Skills are defined in the project's skills directory.

package/README.md

@@ -4,7 +4,7 @@ A standardized AI workflow system for modern AI coding assistants. Initialize st
 
 ## Features
 
-- **Multi-Platform Support**: Works with Cursor, GitHub Copilot, Claude Code, and OpenCode
+- **Multi-Platform Support**: Works with Cursor, GitHub Copilot, Claude Code, OpenCode, and Factory Droid
 - **Structured Workflows**: Plan → Implement → Test → Review methodology
 - **14 Pre-built Commands**: Create plans, execute tasks, run tests, code reviews, and more
 - **7 Reusable Skills**: Design fundamentals, accessibility, theme generation, quality checks
@@ -36,6 +36,7 @@ Choose from:
 - **GitHub Copilot** → `.github/prompts/` and `.github/copilot-instructions.md`
 - **Claude Code** → `.claude/commands/`, `.claude/skills/`, `.claude/themes/`
 - **OpenCode** → `.opencode/command/`, `.opencode/skill/`, `.opencode/agent/`
+- **Factory Droid** → `.factory/commands/`, `.factory/skills/`, `.factory/droids/`
 
 ### Install Specific Tool
 
@@ -49,6 +50,9 @@ npx ai-workflow-init --tool cursor
 # Install only OpenCode
 npx ai-workflow-init --tool opencode
 
+# Install only Factory Droid
+npx ai-workflow-init --tool factory
+
 # Install only GitHub Copilot
 npx ai-workflow-init --tool copilot
 ```
@@ -372,6 +376,7 @@ AGENTS.md               # Universal AI instructions
 | **GitHub Copilot** | `.github/prompts/*.prompt.md` | - | `.github/copilot-instructions.md` |
 | **Claude Code** | `.claude/commands/*.md` | `.claude/skills/*/SKILL.md` | `.claude/CLAUDE.md`, `.claude/themes/` |
 | **OpenCode** | `.opencode/command/*.md` | `.opencode/skill/*/SKILL.md` | `.opencode/agent/`, `opencode.json` |
+| **Factory Droid** | `.factory/commands/*.md` | `.factory/skills/*/SKILL.md` | `.factory/droids/*.md` |
 
 ---
 
@@ -393,13 +398,13 @@ Skills provide specialized knowledge that AI agents can load on-demand:
 
 ## Platform Compatibility
 
-| Feature | Cursor | Copilot | Claude | OpenCode |
-|---------|--------|---------|--------|----------|
-| Commands | ✅ | ✅ | ✅ | ✅ |
-| Skills | ✅ | ❌ | ✅ | ✅ |
-| Custom Agents | ❌ | ❌ | ❌ | ✅ |
-| AGENTS.md | ✅ | ✅ | ✅ | ✅ |
-| Path-specific rules | ✅ | ✅ | ❌ | ❌ |
+| Feature | Cursor | Copilot | Claude | OpenCode | Factory Droid |
+|---------|--------|---------|--------|----------|---------------|
+| Commands | ✅ | ✅ | ✅ | ✅ | ✅ |
+| Skills | ✅ | ❌ | ✅ | ✅ | ✅ |
+| Custom Agents | ❌ | ❌ | ❌ | ✅ | ✅ (Droids) |
+| AGENTS.md | ✅ | ✅ | ✅ | ✅ | ✅ |
+| Path-specific rules | ✅ | ✅ | ❌ | ❌ | ❌ |
 
 ---
 
@@ -436,7 +441,7 @@ Skills provide specialized knowledge that AI agents can load on-demand:
 
 ## Contributing
 
-This project maintains workflows for 4 AI coding tools. When adding commands:
+This project maintains workflows for 5 AI coding tools. When adding commands:
 
 1. Add to `.claude/commands/` (source of truth)
 2. Run `/sync-workflow` to propagate to other tools

package/cli.js

@@ -220,6 +220,12 @@ const AI_TOOLS = [
     description: "Terminal-based AI coding agent",
     folders: [".opencode/command", ".opencode/skill", ".opencode/agent"],
   },
+  {
+    id: "factory",
+    name: "Factory Droid",
+    description: "Factory AI coding assistant",
+    folders: [".factory/commands", ".factory/skills", ".factory/droids"],
+  },
 ];
 
 // Interactive multi-select using arrow keys
@@ -384,17 +390,13 @@ function installClaudeCode() {
   }
   run(`npx degit ${REPO}/.claude/commands .claude/commands --force`);
 
-  // Download CLAUDE.md (context memory) - only if not exists
+  // Download CLAUDE.md (context memory) - always overwrite to get latest
   step("🚚 Downloading Claude Code context memory (.claude/CLAUDE.md)...");
   const claudeMdPath = ".claude/CLAUDE.md";
-  if (existsSync(claudeMdPath)) {
-    skip(`Skipping (already exists): ${claudeMdPath}`);
-  } else {
-    try {
-      run(`curl -fsSL ${RAW_BASE}/.claude/CLAUDE.md -o ${claudeMdPath}`);
-    } catch (_) {
-      run(`wget -qO ${claudeMdPath} ${RAW_BASE}/.claude/CLAUDE.md`);
-    }
+  try {
+    run(`curl -fsSL ${RAW_BASE}/.claude/CLAUDE.md -o ${claudeMdPath}`);
+  } catch (_) {
+    run(`wget -qO ${claudeMdPath} ${RAW_BASE}/.claude/CLAUDE.md`);
   }
 
   // Create settings.json with hooks (project-level, shareable with team)
@@ -417,15 +419,6 @@ function installClaudeCode() {
           }
         ],
         PostToolUse: [
-          {
-            matcher: "Write",
-            hooks: [
-              {
-                type: "prompt",
-                prompt: "If the file path contains 'docs/ai/planning/feature-' and ends with '.md', validate this planning document has all required sections: 1. Goal & Acceptance Criteria, 2. Risks & Assumptions, 3. Definition of Done, 4. Implementation Plan (with Summary and Phases), 5. Follow-ups. Return JSON: {\"valid\": true/false, \"missing\": [list of missing sections], \"message\": \"brief validation result\"}. If file path doesn't match, return {\"valid\": true, \"message\": \"skipped - not a planning doc\"}"
-              }
-            ]
-          },
           {
             matcher: "Edit",
             hooks: [
@@ -436,35 +429,6 @@ function installClaudeCode() {
             ]
           }
         ],
-        PreToolUse: [
-          {
-            matcher: "Edit",
-            hooks: [
-              {
-                type: "prompt",
-                prompt: "If the file path contains 'docs/ai/planning/feature-' and ends with '.md': Check if the agent is starting a new phase. If yes, verify all tasks in the previous phase are marked [x]. Return JSON: {\"canProceed\": true/false, \"message\": \"reason\"}. If file path doesn't match, return {\"canProceed\": true, \"message\": \"skipped - not a planning doc\"}"
-              }
-            ]
-          },
-          {
-            matcher: "Bash",
-            hooks: [
-              {
-                type: "prompt",
-                prompt: "GIT SAFETY CHECK: If command contains 'git ' (git commands like git add, git commit, git push, git pull, git checkout, git branch, git merge, git rebase, git reset, git stash, etc.): Check the ORIGINAL USER PROMPT - did the user EXPLICITLY request git operations? Keywords indicating user wants git: 'commit', 'push', 'git', 'version control', 'save changes', 'create branch'. If user did NOT explicitly request git operations, BLOCK with message 'Git operations require explicit user request. Ask user first.' Return JSON: {\"allow\": true/false, \"reason\": \"explanation\"}. If command does not contain git, return {\"allow\": true, \"reason\": \"not a git command\"}"
-              }
-            ]
-          },
-          {
-            matcher: "Write|Edit",
-            hooks: [
-              {
-                type: "prompt",
-                prompt: "SECURITY CHECK: If file path matches any sensitive pattern: '.env', '.env.*', 'credentials', 'secrets', 'api_key', 'apikey', 'password', 'private_key', 'token', '.pem', '.key', 'auth.json', 'config/prod': WARN user with message '⚠️ SENSITIVE FILE: [filename] - Are you sure you want to modify this file?'. Return JSON: {\"isSensitive\": true/false, \"warning\": \"message if sensitive\", \"allow\": true}. Always allow but warn for sensitive files."
-              }
-            ]
-          }
-        ],
         Stop: [
           {
             matcher: "",
@@ -565,6 +529,45 @@ function installOpenCode() {
   }
 }
 
+// Install Factory Droid
+function installFactoryDroid() {
+  step("🚚 Downloading Factory Droid commands (.factory/commands)...");
+  if (!existsSync(".factory/commands")) {
+    mkdirSync(".factory/commands", { recursive: true });
+  }
+
+  // Check if source exists, if not create from Claude commands
+  try {
+    run(`npx degit ${REPO}/.factory/commands .factory/commands --force`);
+  } catch (e) {
+    console.log(`${colors.yellow}⚠️  Factory Droid commands not found in repo, will be synced from Claude Code${colors.reset}`);
+  }
+
+  step("🚚 Downloading Factory Droid skills (.factory/skills)...");
+  if (!existsSync(".factory/skills")) {
+    mkdirSync(".factory/skills", { recursive: true });
+  }
+
+  // Check if source exists, if not create from Claude skills
+  try {
+    run(`npx degit ${REPO}/.factory/skills .factory/skills --force`);
+  } catch (e) {
+    console.log(`${colors.yellow}⚠️  Factory Droid skills not found in repo, will be synced from Claude Code${colors.reset}`);
+  }
+
+  step("🚚 Downloading Factory Droid droids (.factory/droids)...");
+  if (!existsSync(".factory/droids")) {
+    mkdirSync(".factory/droids", { recursive: true });
+  }
+
+  // Check if source exists
+  try {
+    run(`npx degit ${REPO}/.factory/droids .factory/droids --force`);
+  } catch (e) {
+    console.log(`${colors.yellow}⚠️  Factory Droid droids not found in repo, using built-in droids${colors.reset}`);
+  }
+}
+
 async function main() {
   console.log(`
 ${colors.cyan}╔═══════════════════════════════════════════════════════════╗
@@ -633,6 +636,10 @@ ${colors.cyan}╔═════════════════════
     installOpenCode();
   }
 
+  if (toolIds.includes("factory")) {
+    installFactoryDroid();
+  }
+
   // Download AGENTS.md (luôn ghi đè)
   step("🚚 Downloading AGENTS.md...");
   try {

package/docs/sync-workflow/common.md

@@ -0,0 +1,131 @@
+# Common Analysis for Workflow Sync
+
+This file contains shared logic for analyzing source files and detecting existing targets.
+
+---
+
+## Step 1: Fetch Latest Documentation (Optional)
+
+**CRITICAL: Fetch latest docs before syncing to ensure format compliance.**
+
+Only fetch documentation for the selected platforms. Use WebSearch or WebFetch.
+
+**Error handling:**
+- Web search fails: Use cached knowledge + warn user docs may be outdated
+- URL fetch fails: Try alternative URLs or search queries
+- Format changed significantly: Alert user, show differences
+
+---
+
+## Step 2: Analyze Source (Claude Code)
+
+### 2a: Analyze Commands
+
+**Tools:**
+- Glob(pattern=".claude/commands/*.md")
+- Read(file_path=...) for each command
+
+**For each Claude command, extract:**
+- Frontmatter: name, description
+- Goal section
+- Step-by-step instructions
+- Tool references (AskUserQuestion, Read, Write, Edit, Task, etc.)
+- Skill references (`.claude/skills/...`)
+- Notes and guidelines
+
+### 2b: Analyze Skills
+
+**Tools:**
+- Glob(pattern=".claude/skills/**/SKILL.md")
+- Read(file_path=...) for each skill
+
+**For each Claude skill, extract:**
+- Skill name and category (from path)
+- SKILL.md content
+- Triggers and usage conditions
+- Instructions and guidelines
+
+### 2c: Analyze Base Instructions
+
+**Tools:**
+- Read(file_path=".claude/CLAUDE.md")
+- Read(file_path="AGENTS.md")
+
+**Build inventory:**
+```
+| Type    | Name              | Description                          | Location              |
+|---------|-------------------|--------------------------------------|-----------------------|
+| Command | create-plan       | Generates feature planning doc       | .claude/commands/     |
+| Command | execute-plan      | Implements tasks from planning doc   | .claude/commands/     |
+| Skill   | figma-extraction  | Extract design from Figma            | .claude/skills/design/|
+| Skill   | quality-code-check| Linting and type checking            | .claude/skills/arch/  |
+| Rules   | CLAUDE.md         | Base instructions                    | .claude/              |
+| Rules   | AGENTS.md         | Agent instructions                   | root                  |
+```
+
+---
+
+## Step 3: Detect Existing Target Files
+
+**Tools (run for selected platforms only):**
+
+### Cursor:
+- Glob(pattern=".cursor/commands/*.md")
+- Glob(pattern=".cursor/rules/*.md")
+
+### GitHub Copilot:
+- Glob(pattern=".github/prompts/*.prompt.md")
+- Read(file_path=".github/copilot-instructions.md")
+
+### OpenCode:
+- Glob(pattern=".opencode/command/*.md")
+- Glob(pattern=".opencode/skill/*/SKILL.md")
+- Glob(pattern=".opencode/agent/*.md")
+
+### Factory Droid:
+- Glob(pattern=".factory/commands/*.md")
+- Glob(pattern=".factory/skills/*/SKILL.md")
+- Glob(pattern=".factory/droids/*.md")
+
+---
+
+## Step 4: Classify Files
+
+**Compare and classify:**
+- **MISSING**: Source exists but not in target
+- **OUTDATED**: Target exists but content differs significantly
+- **CURRENT**: Target matches source (skip)
+
+**Output classification example:**
+```
+[Platform Name]:
+  - create-plan: OUTDATED (source modified)
+  - execute-plan: CURRENT (skip)
+  - sync-workflow: MISSING (new)
+```
+
+---
+
+## Tool Reference Conversion (Universal)
+
+| Claude Code | Generic Instruction |
+|-------------|---------------------|
+| `AskUserQuestion(...)` | Ask user for input |
+| `Task(subagent_type='Explore')` | Search codebase |
+| `Task(subagent_type='General')` | General analysis |
+| `Read(file_path=...)` | Read file |
+| `Write(file_path=...)` | Write file |
+| `Edit(file_path=...)` | Edit file |
+| `Glob(pattern=...)` | Find files matching pattern |
+| `Grep(pattern=...)` | Search content for pattern |
+| `Bash(command=...)` | Run terminal command |
+| `WebFetch(url=...)` | Fetch URL content |
+
+---
+
+## Error Handling
+
+- Directory doesn't exist: Create it first with `mkdir -p`
+- File write fails: Retry once, then notify user
+- Conversion uncertain: Add TODO comment in output file
+- Skill name validation fails: Adjust name to meet requirements