Release v1.6.1

🚀 Overview

This patch release enhances AI agent definitions to maximize execution precision and development quality through structured processes and clearer instructions.

✨ Key Improvements

Task Decomposition

• Verifiability Levels: L1/L2/L3 verification system for systematic task validation
• Slice Strategy: Vertical/horizontal slice determination for optimal task ordering
• Operation Verification: Mandatory verification methods with specific commands

Document Processing

• Composite Review Mode: Multi-angle document verification in single execution
• Interface Analysis: Mandatory impact analysis for interface changes
• PRD Boundaries: Clear exclusion of implementation phases from requirements

Technical Design

• ADR Management: Structured common ADR verification and creation process
• Migration Paths: Clear specification for interface conversion requirements
• Architecture Consistency: Enhanced project-wide consistency verification

🧹 Maintenance

• Agent Cleanup: Removed suspended document-fixer agent
• Dependency Optimization: 2-level maximum task dependency rule
• Bilingual Consistency: All improvements applied to both language versions

📈 Impact

• Enhanced Precision: Clearer instructions reduce AI interpretation errors
• Improved Efficiency: Streamlined review processes and verification levels
• Better Quality: Systematic verification and completion criteria
• Reduced Maintenance: Cleanup of unused components

Installation

npm install [email protected]

---

Full Changelog: v1.6.0...v1.6.1

Release v1.6.0

🚀 New Features

Code Reviewer Agent

• Added comprehensive automated code review capabilities
• Supports multi-perspective code analysis including security, performance, and maintainability
• Available in both Japanese and English versions

Review Command

• New review command for triggering code reviews
• Integrated with the code-reviewer agent for seamless workflow

✨ Improvements

Enhanced AI Agent Precision

• Optimized agent definitions for improved execution accuracy
• Refined prompt templates with clearer instructions and guidelines
• Better task decomposition and execution patterns

Documentation Updates

• Updated README with new commands and agent descriptions
• Added comprehensive design templates for technical documentation

🔧 Refactoring

Streamlined Architecture Documentation

• Removed redundant architecture pattern files (hybrid-progressive, vertical-slice)
• Consolidated documentation structure for better maintainability

Agent Definition Improvements

• Enhanced error handling guidance in quality-fixer agent
• Improved task execution flow in task-executor agent
• Refined work planning templates with better structure

📝 Changes by Component

Agents (Japanese & English)

• code-reviewer: New agent for comprehensive code review
• document-fixer: Minor refinements in review integration
• prd-creator: Updated template references and structure
• quality-fixer: Enhanced error handling and verification steps
• requirement-analyzer: Improved clarity in analysis patterns
• task-executor: Added explicit completion criteria
• technical-designer: Expanded design documentation guidance
• work-planner: Optimized task structuring patterns

Commands

• Added review command for both Japanese and English versions
• Comprehensive review workflow integration

🎯 Impact

This release significantly improves the AI agent ecosystem's accuracy and reliability, providing developers with better tools for automated code review and task execution. The streamlined documentation and optimized agent definitions enhance the overall developer experience.

Release v1.5.1

🔧 Bug Fixes

Improved Stability of Autonomous Execution Mode

This patch release addresses stability issues in the build command's autonomous execution mode that were causing inconsistent behavior during task execution.

What's Changed

Core Improvements

• Fixed structured response processing: The orchestrator now reliably detects and processes readyForQualityCheck: true signals from task-executor
• Enhanced metacognition process: Made the execution flow more explicit with mandatory steps:
  1. Execute rule-advisor to understand task essence
  2. Update TodoWrite for progress tracking
  3. Process structured responses immediately
• Improved quality gate reliability: Ensures all quality checks are properly executed before task completion

Documentation Updates

• Clarified orchestrator responsibilities in both Japanese and English command files
• Improved English translation consistency across all command documentation
• Added explicit scope definitions for better command boundary understanding

Why This Matters

The build command is critical for autonomous task execution in CI/CD pipelines and development workflows. This fix ensures:

• ✅ Consistent quality checks on all executed tasks
• ✅ Proper error detection and handling
• ✅ Reliable autonomous execution without manual intervention
• ✅ Clear execution flow that prevents task omissions

Technical Details

The issue was identified in the orchestration layer where structured responses from sub-agents were occasionally overlooked, causing the quality-fixer to not be triggered even when required. This could lead to tasks being marked as complete without proper quality validation.

Compatibility

This is a patch release with no breaking changes. All existing workflows and commands remain fully compatible.

Version

1.5.0 → 1.5.1

---

Full Changelog: v1.5.0...v1.5.1

Release v1.5.0

🎯 Overview

This release significantly improves AI execution accuracy through enhanced subagent definitions and workflow optimizations. The focus is on providing more intelligent and context-aware assistance for complex development tasks.

✨ New Features

📊 Technical Information Verification

• document-reviewer now actively verifies technical claims against latest information
• Automatic cross-referencing with official documentation and best practices
• WebSearch integration for real-time validation of technology choices

🔄 Reverse Engineering Mode

• prd-creator introduces reverse-engineer mode for existing implementations
• Automatically extracts specifications from code when PRD doesn't exist
• Enables comprehensive documentation for legacy features before major modifications

📈 Enhanced Scale-Based Workflows

• Improved PRD determination logic with three modes: create, update, and reverse-engineer
• Automatic mode selection based on project scale and existing documentation
• Clearer document requirements for small/medium/large scale projects

🔍 Latest Technology Research

• technical-designer now proactively researches latest best practices
• Automatic citation of reliable sources in ADR and Design Docs
• Integration with WebSearch for up-to-date technology comparisons

🐛 Improvements

Terminology Consistency

• Fixed scale assessment/determination terminology inconsistencies
• Unified approval recommendation language across all agents
• Improved English translations for better clarity

Workflow Optimization

• Simplified orchestrator commands for better task delegation
• Enhanced document output principles - immediate execution upon user instruction
• Improved ADR status management with clear recommendation/decision separation

📝 Documentation Updates

• Synchronized all Japanese and English agent definitions
• Updated sub-agents practical guide with clearer orchestration patterns
• Enhanced requirement analyzer with explicit PRD mode determination

🔧 Technical Details

Modified Components

• .claude/agents-en/* - All English agent definitions updated
• .claude/agents-ja/* - All Japanese agent definitions enhanced
• docs/guides/*/sub-agents.md - Orchestration guidelines improved
• .claude/commands/*/implement.md - Simplified orchestrator behavior

Compatibility

• ✅ Fully backward compatible
• ✅ No breaking changes
• ✅ Existing workflows continue to function

📊 Impact

This release enhances:

• Accuracy: Better terminology consistency reduces AI confusion
• Intelligence: Proactive research and verification capabilities
• Efficiency: Clearer workflows and automatic mode selection
• Quality: Technical claims are now verified against latest information

🚀 Upgrade Guide

Simply update to v1.5.0 - no migration needed:

npm install [email protected]

📌 Notes for Claude Code Users

These improvements are specifically optimized for Claude's new models, providing:

• More precise subagent task delegation
• Better understanding of project scale and requirements
• Proactive quality assurance and technical verification
• Reduced need for manual intervention in standard workflows

---

Full Changelog: v1.4.4...v1.5.0

Release v1.4.4

What's Changed

🔧 Improvements

• Removed rule-advisor dependencies from subagent definitions: Improved system maintainability by reducing inter-agent dependencies. Subagents now operate more independently with direct rule references instead of relying on rule-advisor selections.

• Clarified rule-advisor usage examples: Updated documentation to show examples rather than fixed usage patterns, making it clearer that these are flexible guidelines.

• Simplified task-executor configuration: Removed ambiguous "mandatory check items" section that lacked clarity, streamlining the agent's focus on implementation execution.

📝 Documentation

• Updated both Japanese and English versions of CLAUDE.md and agent definition files
• Made rule references more explicit and self-contained in each agent definition

Technical Details

• Changed "rule-advisor selected" references to "project" references throughout agent definitions
• Removed cross-dependencies between subagents to improve system reliability
• Maintained backward compatibility - no breaking changes

Full Changelog: v1.4.3...v1.4.4

Release v1.4.3

Breaking Changes / Adaptations

• Adapted to Claude Code behavior change where subagents can no longer invoke other subagents through the Task tool
• Removed Task tool from all 18 subagent definitions (both Japanese and English)
• Replaced rule-advisor Task invocations with direct rule file references
• Addressed the new limitation preventing self-referencing Task tool usage in subagents

Documentation Updates

• Updated all agent definitions to use bullet point format instead of Task syntax for code examples
• Synchronized Japanese and English agent documentation
• Removed quality assurance line from /design command description
• Marked document-fixer as suspended in README due to Task tool limitations

Technical Improvements

• Streamlined subagent tool configurations by removing tools that cannot be used within subagents
• Ensured all agent definitions work correctly within the subagent execution context
• Simplified agent implementation examples for better clarity

Release v1.4.2

Release Notes - npx Setup Feature

🚀 What's New

NPX Support for Quick Project Setup

You can now initialize a new AI-powered project with a single command:

npx github:shinpr/ai-coding-project-boilerplate my-project

✨ Features

• Instant Setup: Create a new project without manual cloning
• Interactive Configuration: Guided setup process with language selection
• Automatic Dependencies: Post-setup script handles all npm installations
• Clean Project Structure: Removes unnecessary boilerplate files automatically

📦 Implementation Details

• Added bin/create-project.js for npx execution
• Created setup scripts for project initialization
• Updated package.json with bin configuration
• Enhanced README with npx usage instructions

🎯 Benefits

• Faster project initialization
• Reduced setup friction for new users
• Consistent project structure across teams
• Better developer experience

---

This feature enables developers to quickly bootstrap AI-assisted coding projects with all the necessary configurations and best practices built-in.

Release v1.4.1

What's Changed

• 🔄 Restructured reference management: Removed redundant reference sections from rule files and centralized them in rules-index.yaml
• 🎯 Enhanced rule-advisor: Shifted from "minimal necessary" to "comprehensive quality" approach for better task completion
• 📊 Improved efficiency: Achieved ~18% reduction in rule file sizes for better context window utilization

Key Improvements

• CLAUDE.md basic rules are now always included in rule selection
• Quality assurance rules automatically included for code modifications
• Better handling of tasks like "fix type errors" with appropriate testing rules

Stats

• Japanese rules: -383 lines (-19.2%) / -6,568 characters (-17.2%)
• English rules: -382 lines (-19.1%) / -11,275 characters (-18.3%)

Release v1.4.0

Overview

This release introduces a comprehensive dynamic rule selection system that significantly improves AI agent efficiency and accuracy. The system reduces rule content by 15-18% while maintaining or improving functionality through intelligent rule selection and enhanced agent capabilities.

Major Features

🎯 Dynamic Rule Selection System

• New rules-index.yaml files in both docs/rules-ja/ and docs/rules-en/
  • Provides metadata for each rule file (description, priority, keywords)
  • Enables AI agents to select only relevant rules based on task context
  • Optimizes context window usage for better performance

🤖 New rule-advisor Subagent

• Intelligent rule selection agent that maximizes AI execution accuracy
• Selects minimal effective ruleset based on task requirements
• Provides structured guidance including:
  • Task essence understanding
  • Applicable rules selection
  • Past failure pattern recognition
  • First action recommendations

📝 Rule Content Optimization

• Japanese rules: Reduced by 353 lines (17.7%) / 5,330 characters (13.9%)
• English rules: Reduced by 353 lines (17.7%) / 9,586 characters (15.5%)
• CLAUDE.ja.md: Reduced by 39 lines (22.3%) / 1,040 characters (25.9%)
• CLAUDE.en.md: Reduced by 38 lines (21.8%) / 2,655 characters (32.1%)
• Removed canonical-phrases.md - content integrated into other files

📚 Enhanced Documentation

• Updated README files with rule index system explanation
• Added detailed descriptions for all 10 subagents
• Documented the dynamic rule selection benefits
• Improved project structure documentation

Technical Improvements

Rule System Enhancements

• Mandatory integration between TodoWrite and rule-advisor
• Automatic rule selection before task execution
• Meta-cognition support through structured task analysis
• Past failure pattern recognition and avoidance

Agent Behavior Controls

• Stricter implementation approval requirements
• Automatic stops for large-scale changes (5+ files)
• Forced rule-advisor execution on repeated errors
• Enhanced progress tracking through TodoWrite integration

Breaking Changes

• canonical-phrases.md has been removed - content integrated into other rule files
• Rule loading now prioritizes the rule index system over loading all files

Migration Guide

1. Update language settings: npm run lang:en or npm run lang:ja
2. Review new rules-index.yaml files for rule organization
3. Utilize rule-advisor for optimal rule selection in tasks
4. Update any custom scripts that reference canonical-phrases.md

Commit History

• 6872f86 feat: Release v1.4.0 - Enhanced documentation and rule-advisor integration
• ec766f5 feat: Complete English translation of rule files from Japanese sources
• 29a418e feat: Complete English translation of agent definitions and commands
• 0dc7b63 refactor(rules): Rule system evaluation and optimization
• eac63ea refactor(rules): CLAUDE.ja.md accuracy improvement and rule-advisor integration
• 5001d38 refactor: Complete migration to dynamic rule selection system

Statistics

• Total rule content reduction: ~15-18% across all languages
• 10 specialized subagents now available
• 6 core rule files (down from 7)
• Full bilingual support (Japanese/English)

Future Considerations

• Further optimization opportunities exist in consolidating common concepts across files
• Potential for more granular rule selection based on specific task types
• Enhanced integration between agents for complex workflows

---

This release represents a significant step forward in making AI-assisted development more efficient and accurate through intelligent rule management and enhanced agent capabilities.

Release v1.3.1

Overview

This release enhances the pre-implementation planning process for Claude Code by introducing self-reflection questions that improve metacognitive abilities and reduce implementation errors.

New Features

Enhanced Pre-Implementation Planning

• Task Command Files: Added new command files (task.md) for both Japanese and English environments in .claude/commands-ja/ and .claude/commands-en/ directories
• Self-Reflection Questions: Updated the implementation workflow to include three critical self-reflection questions that must be addressed before presenting an implementation plan:
  1. Task Essence & Rules: Identify the core purpose of the task and determine which development rules should guide the approach
  2. Rule-Based Prioritization: Based on applicable rules, determine what should be tackled first
  3. Failure Pattern Recognition: Leverage past experience to identify common failures associated with similar tasks

Improvements

Metacognitive Enhancement

• The new self-reflection mechanism encourages deeper thinking before implementation
• Helps prevent common mistakes by forcing consideration of past failure patterns
• Ensures rule compliance from the very beginning of task execution

Bilingual Support

• Both Japanese (CLAUDE.ja.md) and English (CLAUDE.en.md) documentation have been updated
• Command files are available in both languages to support international development teams

Technical Details

• Version bumped from 1.3.0 to 1.3.1
• No breaking changes
• Backward compatible with existing workflows

Benefits

1. Reduced Implementation Errors: By thinking through potential pitfalls before coding
2. Better Rule Adherence: Explicit identification of applicable rules before starting
3. Improved Planning Quality: More thoughtful and comprehensive implementation plans
4. Learning from Experience: Systematic consideration of past failures prevents repetition

Migration Guide

No migration required. The changes are additive and enhance the existing workflow without breaking compatibility.