claude-symphony 0.0.1

package/template/config/memory_integration.yaml

@@ -0,0 +1,200 @@
+# claude-symphony Memory Integration Configuration
+# AI memory provider integration and sync settings
+
+memory_integration:
+  enabled: true
+  description: "AI 메모리 시스템 통합 및 컨텍스트 동기화"
+
+  # 지원 메모리 프로바이더
+  providers:
+    claude_mem:
+      enabled: true
+      description: "Claude Memory MCP 서버 통합"
+
+      # 동기화 트리거
+      sync_on:
+        - "stage_complete"
+        - "session_end"
+        - "checkpoint_create"
+        - "critical_decision"
+
+      # 저장 설정
+      storage:
+        project_prefix: "claude-symphony"
+        observation_types:
+          - "decision"
+          - "discovery"
+          - "feature"
+          - "bugfix"
+
+      # 검색 설정
+      search:
+        default_limit: 10
+        include_timeline: true
+
+    # 향후 확장용 다른 프로바이더
+    external_memory:
+      enabled: false
+      provider: null
+
+  # HANDOFF-메모리 동기화
+  handoff_memory_sync:
+    on_stage_complete:
+      - action: "save_to_memory"
+        content: "key_decisions"
+        observation_type: "decision"
+        tags:
+          - "{{stage_id}}"
+          - "{{project_name}}"
+        include_metadata:
+          - stage
+          - timestamp
+          - related_files
+
+      - action: "save_to_memory"
+        content: "blocking_issues"
+        observation_type: "bugfix"
+        tags:
+          - "issue"
+          - "{{stage_id}}"
+
+      - action: "save_to_memory"
+        content: "technical_discoveries"
+        observation_type: "discovery"
+        tags:
+          - "learning"
+          - "{{stage_id}}"
+
+    on_stage_start:
+      - action: "load_from_memory"
+        filter:
+          project: "{{project_name}}"
+          stages:
+            - "{{previous_stage}}"
+            - "{{current_stage}}"
+          limit: 20
+        inject_to: "context"
+        format: "summary"
+
+      - action: "search_related"
+        query: "{{stage_objectives}}"
+        limit: 5
+        inject_to: "relevant_context"
+
+    on_session_start:
+      - action: "load_recent"
+        project: "{{project_name}}"
+        limit: 30
+        days: 7
+        inject_to: "session_context"
+
+  # 컨텍스트 주입 설정
+  context_injection:
+    enabled: true
+    description: "메모리에서 관련 컨텍스트 자동 주입"
+
+    triggers:
+      on_task_start:
+        enabled: true
+        search_query: "{{task_description}}"
+        max_results: 5
+        inject_location: "task_context"
+
+      on_error:
+        enabled: true
+        search_query: "error {{error_type}} {{error_message}}"
+        max_results: 3
+        inject_location: "debugging_hints"
+
+    formatting:
+      header: "## Related Context from Memory"
+      item_template: |
+        ### {{title}}
+        - **Type**: {{type}}
+        - **Stage**: {{stage}}
+        - **Date**: {{date}}
+        > {{content}}
+
+  # 메모리 정리 정책
+  cleanup_policy:
+    auto_cleanup: true
+
+    retention:
+      decisions: 90  # days
+      discoveries: 180
+      bugfixes: 60
+      general: 30
+
+    deduplication:
+      enabled: true
+      similarity_threshold: 0.85
+
+# 세션 컨텍스트 복구
+session_recovery:
+  enabled: true
+  description: "이전 세션 컨텍스트 자동 복구"
+
+  on_session_start:
+    auto_load: true
+    prompt_user: false  # true면 사용자에게 확인
+
+    load_order:
+      - last_handoff
+      - recent_memories
+      - checkpoint_state
+
+  recovery_prompt:
+    enabled: true
+    template: |
+      ## 이전 세션 컨텍스트 복구됨
+
+      **마지막 스테이지**: {{last_stage}}
+      **마지막 작업**: {{last_task}}
+      **상태**: {{status}}
+
+      ### 핵심 컨텍스트
+      {{key_context}}
+
+      ### 다음 단계
+      {{next_steps}}
+
+# 크로스 스테이지 학습
+cross_stage_learning:
+  enabled: true
+  description: "스테이지 간 학습 및 패턴 인식"
+
+  patterns:
+    track:
+      - successful_approaches
+      - failed_approaches
+      - common_errors
+      - time_estimates
+
+    apply:
+      suggest_similar: true
+      warn_on_known_pitfalls: true
+
+  feedback_loop:
+    enabled: true
+    collect_on: ["stage_complete", "task_complete"]
+    metrics:
+      - success_rate
+      - time_taken
+      - iterations_needed
+
+# 통합 검증
+integration_checks:
+  on_startup:
+    - check_memory_connection
+    - verify_project_access
+    - load_configuration
+
+  health_monitoring:
+    enabled: true
+    interval: 300  # 5 minutes
+    alert_on_failure: true
+
+  fallback:
+    on_memory_unavailable: "local_cache"
+    local_cache_path: "state/memory_cache/"
+    cache_ttl: 3600  # 1 hour

package/template/config/model_enforcement.yaml

@@ -0,0 +1,141 @@
+# AI Model Enforcement Configuration
+# claude-symphony AI 모델 역할 분담 및 강제 실행 설정
+
+model_enforcement:
+  enabled: true
+
+  # ClaudeCode 직접 처리 가능 (외부 AI 호출 불필요)
+  claudecode_direct:
+    operations:
+      - "npm run lint"
+      - "npm run build"
+      - "npm run test"
+      - "npm run format"
+      - "git status"
+      - "git add"
+      - "git commit"
+      - "ls/cat/find"
+      - "파일 읽기"
+      - "간단한 수정 (< 10줄)"
+      - "설정 파일 편집"
+      - "문서 작성/수정"
+
+    characteristics:
+      - max_lines_change: 10
+      - complexity: "low"
+      - no_architectural_impact: true
+
+  # 외부 AI 강제 호출 필요
+  external_ai_required:
+    # Gemini 필수 태스크
+    gemini_tasks:
+      - "발산적 아이디어 생성"
+      - "창의적 브레인스토밍"
+      - "웹 리서치 (Reddit, HackerNews)"
+      - "시장 조사"
+      - "사용자 페르소나 생성"
+      - "UI/UX 디자인 아이디어"
+
+    # Codex 필수 태스크
+    codex_tasks:
+      - "코드 리팩토링 (> 10줄)"
+      - "성능 최적화"
+      - "복잡한 버그 수정"
+      - "아키텍처 개선"
+      - "디자인 패턴 적용"
+      - "테스트 코드 생성"
+      - "E2E 테스트 작성"
+
+# 스테이지별 자동 호출 설정
+stage_auto_invoke:
+  "01-brainstorm":
+    model: "gemini"
+    trigger: "stage_start"
+    prompt_file: "prompts/ideation.md"
+    required: true
+    wrapper: "scripts/gemini-wrapper.sh"
+
+  "03-planning":
+    model: "gemini"
+    trigger: "architecture_design"
+    prompt_file: "prompts/architecture.md"
+    required: false
+    wrapper: "scripts/gemini-wrapper.sh"
+
+  "04-ui-ux":
+    model: "gemini"
+    trigger: "design_ideation"
+    prompt_file: "prompts/ui_concepts.md"
+    required: false
+    wrapper: "scripts/gemini-wrapper.sh"
+
+  "07-refactoring":
+    model: "codex"
+    trigger: "refactoring_start"
+    prompt_file: "prompts/code_analysis.md"
+    required: true
+    wrapper: "scripts/codex-wrapper.sh"
+    flags: "--full-auto"
+
+  "09-testing":
+    model: "codex"
+    trigger: "test_generation"
+    prompt_file: "prompts/test_generation.md"
+    required: true
+    wrapper: "scripts/codex-wrapper.sh"
+    flags: "--full-auto"
+
+# 역할 분담 가이드라인
+role_guidelines:
+  claudecode:
+    strengths:
+      - "파일 시스템 조작"
+      - "빌드/테스트 실행"
+      - "Git 작업"
+      - "문서 구조화"
+      - "코드 리뷰 (읽기)"
+    avoid:
+      - "대규모 코드 생성"
+      - "창의적 아이디어 요청"
+      - "웹 리서치 기반 작업"
+
+  gemini:
+    strengths:
+      - "창의적 아이디어"
+      - "웹 리서치"
+      - "시장 분석"
+      - "UI/UX 컨셉"
+    avoid:
+      - "코드 작성/수정"
+      - "파일 시스템 조작"
+      - "Git 작업"
+
+  codex:
+    strengths:
+      - "코드 작성/수정"
+      - "리팩토링"
+      - "테스트 작성"
+      - "버그 수정"
+    avoid:
+      - "창의적 아이디어"
+      - "웹 리서치"
+      - "문서 작성"
+
+# 강제 실행 검증
+enforcement_validation:
+  warn_on_violation: true
+  block_on_violation: false
+
+  # 위반 감지 규칙
+  violation_rules:
+    - condition: "refactoring_without_codex"
+      stage: "07-refactoring"
+      message: "⚠️ 리팩토링 작업은 Codex를 통해 수행해주세요. /codex 명령 사용"
+
+    - condition: "brainstorm_without_gemini"
+      stage: "01-brainstorm"
+      message: "⚠️ 아이디어 생성은 Gemini를 통해 수행해주세요. /gemini 명령 사용"
+
+    - condition: "large_code_change_direct"
+      threshold_lines: 10
+      message: "⚠️ 10줄 이상 코드 변경은 Codex 사용을 권장합니다."

package/template/config/models.yaml

@@ -0,0 +1,242 @@
+# claude-symphony AI Models Configuration
+# Model assignments, specialization, and dynamic selection settings
+
+models:
+  # Claude Code (Anthropic)
+  claudecode:
+    provider: "anthropic"
+    model_id: "claude-sonnet-4-20250514"
+    description: "Primary coding agent with tool use"
+    capabilities:
+      - "code_generation"
+      - "code_review"
+      - "debugging"
+      - "tool_use"
+      - "mcp_integration"
+    modes:
+      plan: "Plan mode with user confirmation"
+      plan_sandbox: "Plan mode with sandboxed execution"
+      yolo: "Autonomous execution (container only)"
+      headless: "CI/CD headless mode"
+    context_limit: 200000
+
+    # AI Specialization (신규)
+    strengths:
+      - "정확한 코드 생성"
+      - "복잡한 로직 분석"
+      - "에러 디버깅"
+      - "MCP 통합"
+    best_for:
+      - "06-implementation"
+      - "08-qa"
+      - "05-task-management"
+    temperature_range:
+      min: 0.1
+      max: 0.5
+      default: 0.3
+
+  # Claude (Anthropic) - Research focused
+  claude:
+    provider: "anthropic"
+    model_id: "claude-sonnet-4-20250514"
+    description: "Research and analysis specialist"
+    capabilities:
+      - "research"
+      - "analysis"
+      - "document_synthesis"
+      - "mcp_integration"
+    modes:
+      plan: "Plan mode with structured research"
+    context_limit: 200000
+
+    # AI Specialization (신규)
+    strengths:
+      - "심층 리서치"
+      - "문서 분석 및 요약"
+      - "기술 타당성 검토"
+      - "구조화된 보고서 작성"
+    best_for:
+      - "02-research"
+    temperature_range:
+      min: 0.2
+      max: 0.6
+      default: 0.4
+
+  # Gemini (Google)
+  gemini:
+    provider: "google"
+    cli_command: "gemini"
+    description: "Brainstorming and creative planning"
+    capabilities:
+      - "brainstorming"
+      - "creative_ideation"
+      - "web_access"
+      - "multimodal"
+    modes:
+      plan: "Plan mode for structured ideation"
+      yolo: "Free-form creative exploration"
+    invocation: "tmux"
+    wrapper_script: "scripts/gemini-wrapper.sh"
+
+    # AI Specialization (신규)
+    strengths:
+      - "창의적 아이디어 생성"
+      - "다양한 관점 탐색"
+      - "빠른 아이디어 발산"
+      - "시각적 콘텐츠 분석"
+    best_for:
+      - "01-brainstorm"
+      - "03-planning"
+      - "04-ui-ux"
+    temperature_range:
+      min: 0.7
+      max: 1.0
+      default: 0.9
+
+  # Codex (OpenAI)
+  codex:
+    provider: "openai"
+    cli_command: "codex"
+    description: "Code optimization and refactoring"
+    capabilities:
+      - "code_refactoring"
+      - "test_generation"
+      - "code_optimization"
+      - "deep_analysis"
+    modes:
+      deep_dive: "Deep code analysis mode"
+      sandbox_playwright: "Sandbox with E2E testing"
+    invocation: "tmux"
+    wrapper_script: "scripts/codex-wrapper.sh"
+
+    # AI Specialization (신규)
+    strengths:
+      - "깊이 있는 코드 분석"
+      - "리팩토링 및 최적화"
+      - "테스트 생성"
+      - "기술 부채 해소"
+    best_for:
+      - "07-refactoring"
+      - "09-testing"
+    temperature_range:
+      min: 0.2
+      max: 0.6
+      default: 0.5
+
+# 동적 모델 선택 설정 (신규)
+dynamic_selection:
+  enabled: true
+  description: "태스크 특성에 따른 최적 AI 모델 자동 선택"
+
+  criteria:
+    task_type:
+      weight: 0.4
+      mapping:
+        brainstorming: "gemini"
+        creative: "gemini"
+        planning: "gemini"
+        research: "claude"
+        analysis: "claude"
+        implementation: "claudecode"
+        debugging: "claudecode"
+        refactoring: "codex"
+        testing: "codex"
+        optimization: "codex"
+
+    complexity:
+      weight: 0.3
+      thresholds:
+        simple:
+          max_lines: 100
+          prefer: "claudecode"
+        moderate:
+          max_lines: 500
+          prefer: "claudecode"
+        complex:
+          max_lines: null
+          prefer: "claudecode"
+
+    previous_performance:
+      weight: 0.3
+      enabled: true
+      lookback_days: 7
+      min_samples: 3
+
+  override_rules:
+    # Stage assignment 우선
+    stage_assignment_priority: true
+    # 사용자 지정 우선
+    user_preference_priority: true
+
+  fallback:
+    default_model: "claudecode"
+    on_error: "claudecode"
+    on_timeout: "claudecode"
+
+# Stage-to-Model Assignments
+stage_assignments:
+  "01-brainstorm":
+    primary: "gemini"
+    secondary: "claudecode"
+    collaboration: "parallel"
+
+  "02-research":
+    primary: "claude"
+    secondary: null
+    collaboration: null
+
+  "03-planning":
+    primary: "gemini"
+    secondary: null
+    collaboration: null
+
+  "04-ui-ux":
+    primary: "gemini"
+    secondary: null
+    collaboration: null
+
+  "05-task-management":
+    primary: "claudecode"
+    secondary: null
+    collaboration: null
+
+  "06-implementation":
+    primary: "claudecode"
+    secondary: null
+    collaboration: null
+
+  "07-refactoring":
+    primary: "codex"
+    secondary: "claudecode"
+    collaboration: "sequential"
+
+  "08-qa":
+    primary: "claudecode"
+    secondary: null
+    collaboration: null
+
+  "09-testing":
+    primary: "codex"
+    secondary: null
+    collaboration: null
+
+  "10-deployment":
+    primary: "claudecode"
+    secondary: null
+    collaboration: null
+
+# tmux Session Configuration
+tmux:
+  gemini_session: "ax-gemini"
+  codex_session: "ax-codex"
+  output_timeout: 300  # 5 minutes default
+  capture_method: "wait-for"  # Uses tmux wait-for for synchronization
+
+# Provider Credentials (reference only - actual values in environment)
+credentials:
+  anthropic:
+    env_var: "ANTHROPIC_API_KEY"
+  google:
+    env_var: "GOOGLE_API_KEY"
+  openai:
+    env_var: "OPENAI_API_KEY"