Changelog

/create-agent Command Added - Create custom agents with specialized skills, experience, colors, and emojis. Auto-generates YAML files in .xavier/agents/ with full integration into the colored agent system.

Missing Agent Classes Created - Added TestRunnerAgent, RubyEngineerAgent, and JavaEngineerAgent with specialized capabilities and TDD enforcement.

Fixed Agent Selection System - orchestrator._select_agent_for_task() now ALWAYS returns an agent, ensuring ALL tasks get colored agent displays during sprint execution.

Enhanced Sprint Task Delegation - Smart routing system assigns tasks to appropriate specialists based on type and tech constraints with full visual feedback.

Epic Commands Now Visible - Fixed critical issue where /create-epic, /add-to-epic, and /list-epics commands were missing from Claude Code installation scripts, making them invisible to users despite being implemented in Python code.

Roadmap Commands Fixed - Added /add-to-roadmap command MD file generation to install.sh and update.sh, resolving command visibility issues.

Command Bridge Updated - Fixed xavier_bridge.py command mapping to include all new epic and roadmap commands, ensuring proper command routing.

Update Process Fixed - Enhanced update.sh to extract and deploy new command files, ensuring users get all commands when updating.

install.sh Path Bug - Fixed critical path issues preventing source code from being copied during installation. Corrected xavier/xavier/src to xavier/src and added support for archive structures.

update.sh Detection Bug - Fixed source directory detection in update process ensuring users receive new commands when running /xavier-update.

Version Display Issues - Fixed hardcoded version references throughout framework ensuring accurate version display in greetings and help commands.

Epic Commands Available - All users can now access /create-epic, /add-to-epic, and /list-epics commands after installation or update.

Enhanced Roadmaps Working - /add-to-roadmap and enhanced /create-roadmap commands now properly delivered to all users.

/create-epic Command - Create epics to group related user stories with auto-generated business value statements, target release tracking, and initial story linking.

/add-to-epic Command - Link multiple stories to existing epics with automatic point aggregation and progress tracking.

/list-epics Command - List and filter epics by status or target release with completion percentage and story count display.

Auto-Generation - /create-roadmap now auto-generates roadmaps from project context with intelligent milestone creation based on project complexity.

/add-to-roadmap Command - Dynamically add milestones to existing roadmaps with success criteria and epic associations.

Clear Documentation - Added prominent notices throughout that Xavier stores all data exclusively in JSON format in .xavier/data/ directory.

No MD Files for Data - Clarified that Xavier does NOT create Markdown files for data storage - any MD files in projects are from external sources.

Native Sub-Agent Distribution - Fixed critical issue where users were not receiving native Claude Code sub-agents. Enhanced both update.sh and install.sh to properly copy agent files from repository ensuring ALL users get the complete sub-agent functionality.

Repository File Copying - Fixed update and installation workflows to properly distribute .claude/agents/ files with YAML front matter and comprehensive agent definitions.

Complete Coverage - Both new installations and updates now properly distribute all 6 specialized agents: project-manager, python-engineer, golang-engineer, frontend-engineer, context-manager, and test-runner.

This critical fix ensures that ALL users (both new and updating) receive the complete native Claude Code sub-agent functionality introduced in v1.1.8, providing the specialized development teams as intended.

Version Detection Bug - Fixed issue where xavier-update command showed outdated version information. Users running updates were seeing "version 1.1.5" despite newer versions being available due to missing VERSION file.

VERSION File Creation - Enhanced both update.sh and install.sh to create VERSION file ensuring accurate version detection for future updates and preventing incorrect "up to date" messages.

Story ID Uniqueness

Resolved duplicate story ID generation issue. Added collision detection with retry logic and fallback mechanism for extremely rare UUID collisions. Ensures no two stories can have the same ID.

Automatic Data Structure Initialization

Creates all required JSON files on SCRUM manager startup. Initializes stories.json, tasks.json, bugs.json, sprints.json, epics.json, and roadmaps.json while preserving existing data.

Automatic Roadmap Generation

Roadmap automatically created when creating new project. Includes 4 default milestone templates spanning 16 weeks with realistic development phases. Vision derived from project description and saves to roadmaps.json immediately.

Enhanced User Experience

Streamlined project setup process with automatic roadmap generation. Added comprehensive tests for all new features. Updated documentation to reflect automatic roadmap feature.

Complete Sprint Management Dictionary Issues - Fixed all sprint-related dict/dataclass mismatches including plan_sprint(), start_sprint(), complete_sprint(), and velocity calculations.

Sprint Status Compatibility - Fixed handling of both SprintStatus enum and string values throughout the system.

DateTime Comparison Fix - Resolved datetime vs string comparison errors in velocity calculations.

Sprint Status Compatibility Helper - Added get_sprint_status_value() for consistent status handling across enum and string formats.

Comprehensive Sprint Tests - Full test coverage for sprint operations including lifecycle, velocity, and burndown calculations.

All sprint methods in xavier_commands.py now use safe accessors for reliable operation

Sprint burndown calculations now properly handle mixed data formats

Critical Data Persistence Bug - Fixed fundamental dataclass/dictionary mismatch where models were defined as dataclasses but stored/loaded as plain dictionaries, causing AttributeError throughout the system.

Proper Serialization/Deserialization - Complete data persistence solution with serialize_dataclass() and deserialize_to_dataclass() utilities handling datetime and enum conversion.

Compatibility Layer - Safe attribute access with safe_get_attr() and safe_set_attr() working with both dataclass instances and dictionaries.

Comprehensive Unit Tests - Full test coverage for persistence, serialization, and backward compatibility scenarios.

Fixed _load_data() to properly convert JSON to dataclass instances

Fixed _save_data() to properly serialize dataclasses to JSON

Maintains backward compatibility with projects using old dictionary format

Critical Update Script Bug - Fixed command extraction failing when Xavier downloaded as archive vs git clone. Update script now properly detects install.sh location in both scenarios.

fix-commands.sh Script - Quick fix script for users affected by v1.1.2 update issue. Manually installs estimate-story and set-story-points commands without full update.

Added verbose output showing which commands are being updated during installation

Claude Code Integration - Complete integration of story estimation commands as slash commands in Claude Code.

/estimate-story Command - Now properly registered and available immediately after Xavier update.

/set-story-points Command - Manual story point setting available as Claude Code slash command.

Fixed Claude Code command registration for new estimation features

Update script now properly installs new commands in .claude directory

Xavier bridge command mapping includes all new commands

Updated all command documentation files with examples

Clear distinction between automatic and manual estimation methods

Automatic Story Estimation - New /estimate-story command uses PM agent to intelligently estimate story points for backlog stories.

Complexity Analysis Algorithm - PM agent analyzes technical complexity, CRUD operations, UI/UX requirements, and testing needs.

Batch Estimation - Estimate entire backlog at once or specific stories with visual PM agent feedback.

Renamed /estimate-story to /set-story-points for manual point setting

Enhanced PM agent with sophisticated scoring system mapping to Fibonacci points

Colored Agent Visualization System - Professional visual feedback showing which agent is working on each task with unique colors, emojis, and labels for 9 different agent types.

Real-time Status Tracking - Visual indicators for agent status (⚙️ Working, 🧪 Testing, ✅ Complete) with colored output throughout task lifecycle.

Agent Handoff Visualization - Clear visual transitions when tasks move between specialists with colored arrows and handoff reasons.

Enhanced Sprint Display - Sprint execution now shows colored progress tracking with task counters and visual success indicators.

All agents now announce themselves with colored boxes when taking over tasks

TDD process visualization showing Testing → Working → Complete flow

Orchestrator enhanced with visual feedback and handoff tracking

Added AgentColors class in ansi_art.py with distinct colors for all agent types

New display functions: display_agent_takeover(), display_agent_status(), display_agent_handoff()

Enhanced BaseAgent class with colored output methods

Beautiful ANSI Art System - Professional ASCII art displays throughout the framework with XAVIER logo in block-style similar to Claude Code.

Greeting Display System - Multiple greeting types for welcome screens, sprint announcements, daily tips, and update notifications.

Enhanced Visual Feedback - Light cyan/white color scheme matching Claude Code aesthetics with responsive terminal width detection.

Installation script now features professional ANSI art banner

Update script displays enhanced visual feedback with version comparison

Commands like /xavier-help and /start-sprint now include visual elements

Intelligent Project Creation - New /create-project command with AI-powered analysis that automatically generates user stories, tasks, and technical architecture based on project requirements.

Automatic Update System - New /xavier-update command for seamless framework updates with automatic backup of user data.

Strict Command Boundaries - Clear separation between planning and implementation phases. Commands like /create-story now only create items without auto-implementation.

Enhanced project template detection with automatic tech stack analysis

Improved agent instructions with clearer boundaries and responsibilities

Better error handling in configuration management

Optimized installation script for faster setup

Fixed update script error when run through piped execution

Resolved KeyError in configuration update for older installations

Fixed automatic story implementation issue in Claude Code

Enhanced agent communication protocols

Better sprint planning algorithms

Improved test coverage validation

Fixed sprint velocity calculation error

Resolved task dependency resolution issues

SCRUM Management - Complete implementation of user stories, tasks, bugs, and sprint planning

Test-First Development - Enforced TDD with 100% test coverage requirement

Clean Code Standards - Automatic validation of Clean Code principles and SOLID design patterns

Sequential Execution - Strict one-task-at-a-time execution with dependency management

Specialized Agents - Language-specific engineers (Python, Go, Frontend) with strict boundaries

Claude Code Integration - Seamless integration with custom commands and agents

Velocity Tracking - Automatic sprint velocity calculation and burndown charts