All notable changes to SpecPulse will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
- Script Path Detection: Simplified and fixed project root detection in all shell scripts
- Clear Directory Structure: Scripts now consistently use
$SCRIPT_DIR/..for project root - Documentation: Added comprehensive directory structure explanation in CLAUDE.md
- All shell scripts now use consistent path detection logic
- Improved clarity in script comments about directory structure
- Critical Template Bug: Fixed template files being created with wrong names (spec-001.md instead of spec.md)
- Script Path Bug: Fixed shell scripts looking in wrong directory (../../templates instead of ../templates)
- Init Command: Now correctly creates spec.md, plan.md, and task.md templates
- Project Root Detection: Scripts now properly detect project root directory
- Continuous Task Execution: New
/sp-executecommand for non-stop task completion - Flow State Development: Execute all tasks without interruptions or context switching
- Batch Processing Mode: Complete entire task lists with a single command (
/sp-execute all) - Smart Task Progression: Automatic advancement through pending, in-progress, and blocked tasks
- Task Status Detection: Intelligent identification of next tasks to execute
- Service-Aware Execution: Support for both monolithic (T001) and decomposed (AUTH-T001) task formats
- New Shell Script: Added
sp-pulse-execute.shfor continuous execution orchestration - Enhanced Productivity: 10x faster task completion through elimination of manual confirmations
- Updated README: Added comprehensive section on continuous task execution with examples
- Enhanced Workflow: Streamlined development process from spec to completion
- Command Documentation: Added
/sp-executeto all command listings and workflows
- Template Naming: Fixed
sp-pulse-init.shto use correct file naming (spec-001.md instead of spec-001.md) - Directory Structure: Ensured consistent feature ID usage in file creation
- Updated README.md: Added Important Security Rules section explaining protected vs editable directories
- Documentation: Clarified that templates are copied to working directories, never modified directly
- Clarified File Edit Permissions: Edit tool restored to Claude commands with clear restrictions
- Added Security Notes: All commands now explicitly state which directories are protected
- Directory Protection: templates/, scripts/, commands/ folders are read-only after init
- Claude commands can use Edit tool but ONLY in specs/, plans/, tasks/, memory/ folders
- Added CRITICAL security notes to all command files
- Clarified that templates are COPIED to working directories, then edited there
- Critical Security Fix: Prevented modification of template files
- Script Improvements: Fixed template file paths in all shell scripts
- AI Command Security: Removed Edit tool from Claude commands to prevent template modification
- Template Protection: Ensured all scripts only COPY templates, never modify originals
- All scripts now use correct template names (spec.md, plan.md, task.md)
- Claude commands explicitly state templates must be COPIED, not edited
- Enhanced security by preventing any modification of resources/templates files
- Simplified script execution: Removed Python script fallback, now using only Bash scripts across all platforms
- Streamlined packaging: Removed Python scripts from package distribution
- Updated documentation: Clarified Bash requirement for all platforms (Git Bash on Windows)
- All Python scripts (.py) from resources/scripts directory
- Python fallback references from all Claude and Gemini command files
- Python script references from package data in pyproject.toml and setup.py
- Version display: Fixed
__version__in__init__.pyto show correct version - Now
specpulse --versiondisplays the correct version number
- Critical: Fixed syntax error in main.py line 480 - unterminated string literal
- Multi-line strings in decompose method now use proper triple quotes
- Microservice Decomposition: New
/sp-decomposecommand to break down large specifications into microservices - Service-Based Planning: Automatic detection of decomposed specs and creation of service-specific plans
- Integration Planning: New integration plan template for coordinating between services
- Service Task Management: Task IDs with service prefixes (AUTH-T001, USER-T001, INT-T001)
- Smart Workflow Detection: Plans and tasks automatically adapt to monolithic vs decomposed architecture
- New Templates: Added templates for microservices, API contracts, interfaces, and integration plans
- Decomposition CLI Command:
specpulse decompose [spec-id] [--microservices] [--apis] [--interfaces] - Enhanced Context Tracking:
memory/context.mdnow tracks decomposition status and service architecture
- Workflow Adaptation:
/sp-planand/sp-taskcommands now detect decomposition and generate appropriate artifacts - Template Structure: Added
templates/decomposition/directory with specialized templates - Script Updates: Added
sp-pulse-decompose.shandsp-pulse-decompose.pyfor decomposition orchestration
- Template Loading: Proper handling of decomposition templates in SpecPulse core
- CLI Integration: Decompose command properly integrated with existing workflow
- Command prefix system: All commands now use
sp-prefix to avoid reserved name conflicts - Multi-spec workflow: Support for multiple specifications, plans, and tasks within the same feature
- Versioned file system: Automatic naming with spec-001.md, plan-001.md, task-001.md format
- Context detection: Automatic feature detection using memory/context.md and git branch names
- Manual workflow control:
/sp-pulseonly creates directories, manual control over/spec,/plan,/task - Enhanced AI integration: Improved workflow coordination between Claude and Gemini assistants
- Comprehensive codebase review: Fixed all old command references throughout the project
- Template system cleanup: Removed nonsensical shell script execution commands from templates
- File path consistency: Updated all hardcoded file paths to use versioned naming system
- Script naming: All scripts now consistently use
sp-prefix (sp-pulse-.py, sp-pulse-.sh) - Documentation updates: Updated all README references and examples to use new command structure
- Breaking change: All command names changed from
/pulse,/spec,/plan,/taskto/sp-pulse,/sp-spec,/sp-plan,/sp-task - Removed PowerShell scripts: Now only supporting .sh and .py scripts for cross-platform compatibility
- Improved project structure: Better organization of scripts and templates with consistent naming
- Enhanced error handling: Better validation and error messages throughout the system
- Critical cross-platform template bug: Fixed
specpulse initcommand to use proper cross-platform scripts from@specpulse/resourcesinstead of embedded templates - Package data configuration: Added missing
.pyand.ps1script files to package manifest inpyproject.toml - CLI script creation: Updated
_create_scriptsmethod to copy all cross-platform scripts from resources directory - Resource path resolution: Enhanced resource path detection using
pkg_resources.resource_filename()
- Complete cross-platform script availability: All 12 scripts (4 types × 3 platforms) now properly installed
- Better error handling for script copying operations
- Improved script count reporting during initialization
- Complete cross-platform compatibility with Windows, Linux, and macOS support
- Python script equivalents of all bash scripts (.py files)
- PowerShell script equivalents for Windows systems (.ps1 files)
- Automatic platform detection in AI commands with fallback mechanisms
- Comprehensive cross-platform test suite with 100% success rate
- Unicode character support for international feature names (≤, ≥, →, ←)
- Enhanced template system with proper path resolution across platforms
- AI commands now detect operating system automatically and choose appropriate script type
- Template paths updated to use consistent
/resources/templates/structure - Error handling improved with platform-specific logging and fallback mechanisms
- Package data updated to include all script variants (Python, PowerShell, Bash)
- Template path resolution issues across different platforms
- File permission handling for cross-platform compatibility
- Unicode encoding issues on Windows systems
- Script execution failures on non-Unix platforms
- Comprehensive test suite with 95% code coverage (193+ tests)
- Proper Gemini command format following official documentation
- Claude command frontmatter with YAML metadata
- Support for command arguments in both Claude (
$ARGUMENTS) and Gemini ({{args}}) - Test files for AI command validation
- CHANGELOG.md for tracking version changes
- Updated Claude command files to include proper frontmatter
- Converted Gemini command files to official simple TOML format
- Improved README documentation with accurate command examples
- Enhanced test coverage from 69% to 95%
- AI commands now properly accept and handle arguments
- Command documentation matches actual implementation
- Gemini commands use correct
{{args}}placeholder - Claude commands use correct
$ARGUMENTSvariable
- Initial PyPI release
- Core SpecPulse framework
- CLI interface with init, validate, sync, doctor commands
- Claude and Gemini AI integration
- Constitution-based development principles
- Phase Gates validation system
- Improved project structure
- Enhanced validation logic
- Various bug fixes and improvements
- Basic test coverage
- Documentation improvements
- Minor bug fixes
- Initial beta release
- Basic functionality
- Setup and installation issues