- Use word boundaries in delta operation detection to avoid false matches
- Standardize name extraction to use directory name after specs/changes
- Combine archive validation warning into single prompt for better UX
- Add change.md validation to archive command before archiving
- Replace structured {given, when, then} with {rawText} field
- Remove complex Given/When/Then parsing logic
- Preserve original formatting and whitespace
- Update all tests to use rawText field
- Remove unnecessary validation checks
This change reduces complexity while preserving all content exactly
as written, making the system more maintainable and flexible.
- Extract magic numbers to named constants in validation/constants.ts
- Update task 3.2 to reflect actual implementation (constants.ts)
- Add comprehensive documentation for validation system
- Add migration guide for future command integration
- Update CLI help text for diff command
- Add Zod schemas for specs, changes, requirements, and scenarios
- Implement markdown parser for extracting structured data
- Create validation infrastructure with error/warning/info levels
- Enhance archive command with pre-archive validation
- Add --no-validate flag with confirmation prompt for emergencies
- Enhance diff command with non-blocking validation warnings
- Add JSON converters for spec and change formats
- Add comprehensive test coverage for all validation components
The --deltas flag was ambiguous. Since it shows only the requirement
changes (ADDED/MODIFIED/REMOVED/RENAMED sections), rename it to
--requirements-only to be explicit about what it displays.
Add three OpenSpec change proposals for enhancing the CLI:
- add-spec-commands: Resource-based spec commands with JSON output
- add-change-commands: Resource-based change commands with JSON output
- add-zod-validation: Runtime validation with detailed error reporting
These proposals enable programmatic access to specs and changes,
improving integration with CI/CD pipelines and external tooling.
- Update openspec-conventions spec with delta-based approach
- Add Header-Based Requirement Identification for programmatic matching
- Define ADDED/MODIFIED/REMOVED/RENAMED sections format
- Document standard output symbols (+ ~ - →)
- Update openspec/README.md with delta conventions and examples
- Update init command template to use delta format
- Mark completed tasks in adopt-delta-based-changes/tasks.md
This implements the first part of the delta-based changes proposal,
updating all documentation and conventions to support the new format.
- Added back tasks to update the actual specs (not just proposals)
- Included validation implementation tasks
- Kept implementation-focused structure
- Clarified that specs in changes folder are proposals, not current truth
- CLI specs now reference openspec-conventions for shared concepts
- Added standard output symbols definition to conventions
- Simplified tasks.md to focus on implementation only
- Fixed terminology to consistently use 'normalized header'
- Condense 'What Changes' section to core concepts only
- Simplify Impact section to essentials
- Make Conflict Resolution one concise paragraph
- Remove inline comment from example
- Focus on the key benefit: readable GitHub diffs
- Remove phased migration timeline (project not in use yet)
- Remove deprecation notices from CLI commands
- Keep simple backward compatibility for both formats
- Replace full future state storage with delta-based approach
- Store only ADDED, MODIFIED, RENAMED, and REMOVED requirements
- Use headers as unique identifiers for programmatic matching
- Enable cleaner GitHub reviews showing only actual changes
- Add comprehensive examples and implementation tasks
- Added try-catch error handling for configurator failures
- Improved console output to be more specific about what was updated
- Added TODO comment for future multi-configurator test enhancement
- Added test for error handling when configurator fails
- Console now shows 'Updated OpenSpec instructions (README.md)' for clarity
The update command now only updates existing AI tool configuration
files instead of forcing CLAUDE.md creation. This allows team members
to use different AI tools without conflicts.
- Modified update.ts to check for existing files before updating
- Added comprehensive tests for the new behavior
- Updated spec and documentation to reflect team-friendly approach
- Created project README documenting the behavior
- Add --skip-specs flag to archive command to skip spec update operations
- Fix confirmation behavior: declining spec updates now continues with archiving instead of cancelling
- Add comprehensive tests for new functionality
- Update task documentation to reflect completed implementation
The update command now only updates existing AI tool configuration
files instead of forcing CLAUDE.md creation. This allows team members
to use different AI tools without conflicts.
Add change proposal to enable skipping spec updates during archive operation.
This allows archiving changes that don't modify specs (tooling, docs, etc.)
and fixes the confirmation behavior to continue archiving even when users
decline spec updates.
- Add Specification Format section to openspec-conventions with:
- Requirement headers for consistent structure
- Scenario headers with bold WHEN/THEN/AND keywords
- Format flexibility for different content types
- Update all CLI command specs to use structured format:
- cli-init: Convert all behavioral sections
- cli-list: Apply structured format throughout
- cli-update: Restructure requirements and edge cases
- cli-diff: Update all behavior sections
- cli-archive: Convert complex behaviors to scenarios
- Update openspec-conventions spec itself to follow its own format
- Mark all tasks as completed
- Remove migration.md as project has no existing users to migrate
- Remove Migration Support section from tasks.md
- Structured format only applies to behavioral specs, not convention definitions
Tabish Bidiwale