- Update version from 0.2.0 to 0.4.3 in README.md and cli/README.md
- Remove broken Configuration Guide link (docs/CONFIGURATION.md doesn't exist)
- Fix verifyCommands format in embedded rules (object → array)
- Document undocumented commands: detect-ai, circuit-breaker, batch, parse-result

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

- Update version from 0.2.0 to 0.4.3
- Update status from 早期开发 to 活跃开发
- Remove broken Configuration Guide link

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

- Fix embedded documentation in init command
- Correct verifyCommands format (object → array)
- Add documentation for detect-ai, circuit-breaker, batch, parse-result commands

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

- Add missing process.exit() to tasks get command
- Add missing process.exit() to detect-ai and detect-ai-save commands
- Add .unref() to auto-update timers to prevent blocking process exit

Fixes issue where commands would hang indefinitely instead of exiting.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

…ates

- Reduce phase-1-clarify from 863 lines to 107 lines (88% reduction)
- Focus on core principle: context preservation from prior conversations
- Replace rigid templates with flexible guidance
- Add context-aware clarify phase that extracts UI/UX, data models,
  API contracts, and design decisions from conversation history
- Only ask questions for gaps, not already-discussed information

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Keep concise principle-based approach while restoring essential
workflow controls:
- Explicit step numbering (Step 1-6)
- PRD backup mechanism with bash script
- State transition command (ralph-dev state update)
- AskUserQuestion tool constraints (max 4, header ≤12 chars)
- Required PHASE RESULT output format
- Error handling with recovery actions

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

…-compression resilience

Changes:
- Simplify skills following Claude Code best practices (principles over templates)
- Restore CLI bootstrap scripts to all phase skills
- Add context-compression resilience (state verification at start)
- Restore detailed descriptions with trigger hints for user-invocable skills
- Add CLI commands to phase-4-heal for proper state tracking

Key improvements:
- All skills now query state from CLI at initialization
- Reduced total lines from ~4800 to ~1200 (75% reduction)
- Maintained all essential process controls and flow
- Skills comply with official Claude Code skill guidelines

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

- Optimized skill files with best practices
- Added context-compression resilience to all phase skills
- Restored CLI bootstrap scripts and detailed descriptions

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

- Show update notification on startup (non-blocking via update-notifier)
- Ask for user confirmation (y/N) before running update
- Skip confirmation in CI mode or JSON output mode
- Update tests to match new flow

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

- Add Operating Philosophy section (Occam's Razor, service standard, ultrathink mode)
- Add Efficient Agent Usage section (delegation, parallel execution, escalation)
- Remove Chinese translations for English-only documentation

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

… preservation

- Add structured context directory (.ralph-dev/context/) for artifacts
- Preserve Plan Mode content verbatim in context/plan.md
- Track all file references in context/files-referenced.md
- Quote user's exact words in context/user-intent.md
- Add Context Index at PRD top for compression recovery
- Include domain-specific files (ui-design, data-model, api-spec, decisions)
- Update constraints to enforce context preservation rules

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Both commands were missing --json output support, causing errors when
used programmatically by agents. Now all task subcommands consistently
support JSON output format.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Complete CLI consistency - all commands now support --json output:
- detect-ai-save: added --json with proper error handling
- Updated tests to match new error message format

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

- Add idempotent phase transition support in state-entity.ts (same phase = no-op with timestamp update)
- Improve test coverage from 96.55% to 97.98%
- update.ts: 79.63% → 98.69%
- service-factory.ts: 88.37% → 100%
- state.ts: 89.01% → 100%
- Add 51 new tests (1050 total)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

- Add new update-checker.service.ts with direct npm registry check
- Cache version info in ~/.config/configstore/ralph-dev-update-check.json
- Show notification EVERY time CLI runs if update available (not just once)
- 24-hour cache interval for npm registry queries (performance)
- Replace update-notifier based approach with simpler direct implementation
- Add comprehensive tests for update-checker service

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

- fix: include context directory in archive process

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>