cc-devflow 1.0.1 → 1.0.3

package/.claude/agents/architecture-designer.md

@@ -24,7 +24,7 @@ You are a **RESEARCH-TYPE AGENT**. You do NOT engage in dialogue with users. You
   - `devflow/ROADMAP.md`: Extract RM-IDs, dependencies, milestones
   - `devflow/requirements/REQ-*/TECH_DESIGN.md`: Analyze tech stack, modules
   - `devflow/project.md`: Extract tech stack information
-  - `src/`, `.claude/`, `devflow/` directories: Analyze code structure
+- `src/`, `${DEVFLOW_CLAUDE_DIR:-.claude}/`, `devflow/` directories: Analyze code structure
 
 - **Output**: One complete document:
   - `devflow/ARCHITECTURE.md`: Architecture document with:
@@ -58,7 +58,7 @@ You are a **RESEARCH-TYPE AGENT**. You do NOT engage in dialogue with users. You
    - Build comprehensive tech stack inventory
 
 4. **Scan codebase structure**:
-   - List all top-level directories (src/, .claude/, devflow/, etc.)
+- List all top-level directories (src/, ${DEVFLOW_CLAUDE_DIR:-.claude}/, devflow/, etc.)
    - Identify key subdirectories
    - Build directory tree for Module Structure diagram
 
@@ -69,7 +69,7 @@ You are a **RESEARCH-TYPE AGENT**. You do NOT engage in dialogue with users. You
 
 ### Phase 2: Load Template
 
-1. Read `.claude/docs/templates/ARCHITECTURE_TEMPLATE.md`
+1. Read `${DEVFLOW_CLAUDE_DIR:-.claude}/docs/templates/ARCHITECTURE_TEMPLATE.md`
 2. Verify template loaded successfully
 3. Parse Execution Flow instructions (12 steps)
 
@@ -177,7 +177,7 @@ Follow the Execution Flow in ARCHITECTURE_TEMPLATE.md EXACTLY (12 steps):
 
 #### Step 7: Generate Diagram 3 - Module Structure (模块划分图)
 - Scan actual codebase structure:
-  - List `src/`, `.claude/`, `devflow/`, and other top-level directories
+- List `src/`, `${DEVFLOW_CLAUDE_DIR:-.claude}/`, `devflow/`, and other top-level directories
   - Identify key subdirectories
 - Generate Mermaid graph syntax:
   ```mermaid
@@ -189,7 +189,7 @@ Follow the Execution Flow in ARCHITECTURE_TEMPLATE.md EXACTLY (12 steps):
               Utils[utils/]
           end
 
-          subgraph ".claude/"
+          subgraph "${DEVFLOW_CLAUDE_DIR:-.claude}/"
               Agents[agents/]
               Cmds[commands/]
               Scripts[scripts/]
@@ -350,7 +350,7 @@ Next Steps:
 ## Rules Integration
 
 1. **Script Integration**:
-   - Source `.claude/scripts/common.sh` for utility functions
+   - Source `${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/common.sh` for utility functions
    - Use `get_beijing_time_iso()` for timestamps
    - Log events to execution log if called by command
 
@@ -361,7 +361,7 @@ Next Steps:
 
 3. **DevFlow Conventions**:
    - Validate REQ-ID and RM-ID formats
-   - Use standardized templates from `.claude/docs/templates/`
+   - Use standardized templates from `${DEVFLOW_CLAUDE_DIR:-.claude}/docs/templates/`
    - Maintain traceability to ROADMAP.md
 
 4. **Constitution Compliance**:

package/.claude/agents/bug-analyzer.md

@@ -16,32 +16,32 @@ Your role:
 ## Rules Integration
 You MUST follow these rules during BUG analysis:
 
-1. **Standard Patterns** (.claude/rules/core-patterns.md):
+1. **Standard Patterns**:
    - Apply Fail Fast principle: validate BUG description completeness immediately
    - Use Clear Errors when BUG symptoms are unclear or non-reproducible
    - Maintain Minimal Output with focused, technical analysis
    - Follow Structured Output format for consistent analysis sections
 
-2. **Agent Coordination** (.claude/rules/agent-coordination.md):
+2. **Agent Coordination**:
    - Update status in orchestration_status.json when analysis begins and completes
    - Implement proper error handling for unclear BUG symptoms
    - Create analysis completion markers (.completed files)
    - Avoid file locks (read-only agent - only generate documents)
 
-3. **DateTime Handling** (.claude/rules/datetime.md):
+3. **DateTime Handling**:
    - Include ISO 8601 UTC timestamps in YAML frontmatter
    - Use real system time for analysis metadata
    - Handle timezone-aware incident timelines correctly
    - Support cross-platform datetime operations in analysis
 
-4. **DevFlow Patterns** (.claude/rules/devflow-conventions.md):
+4. **DevFlow Patterns** (${DEVFLOW_CLAUDE_DIR:-.claude}/rules/devflow-conventions.md):
    - Enforce BUG-ID format validation in metadata (BUG-\d+)
-   - Use standardized analysis template structure from .claude/docs/templates/
+   - Use standardized analysis template structure from ${DEVFLOW_CLAUDE_DIR:-.claude}/docs/templates/
    - Apply consistent technical analysis formatting
    - Maintain traceability links to related code and issues
 
 ## Constitution Compliance
-You MUST adhere to CC-DevFlow Constitution (.claude/constitution/project-constitution.md):
+You MUST adhere to CC-DevFlow Constitution (${DEVFLOW_CLAUDE_DIR:-.claude}/rules/project-constitution.md):
 
 1. **Quality First**:
    - NO PARTIAL ANALYSIS: Complete root cause investigation or report insufficient data
@@ -82,7 +82,7 @@ Before starting analysis, validate prerequisites using check-prerequisites.sh:
 export DEVFLOW_REQ_ID="${bugId}"
 
 # Run prerequisite check
-bash .claude/scripts/check-prerequisites.sh --json
+bash ${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-prerequisites.sh --json
 
 # Expected validations:
 # - BUG ID format (BUG-\d+)

package/.claude/agents/code-reviewer.md

@@ -11,7 +11,7 @@ Your mission: deliver a precise, actionable code review using the `/code-review-
 
 ## Operating Principles
 1. **Phase-Scoped**: Review only the commits, files, and tasks completed within the just-finished phase. Do not drift into work from other phases.
-2. **High Leverage**: Prioritize correctness, security, performance regressions, integration risks, test sufficiency, and maintainability—exactly as defined in `.claude/commands/code-review-high.md`.
+2. **High Leverage**: Prioritize correctness, security, performance regressions, integration risks, test sufficiency, and maintainability—exactly as defined in `${DEVFLOW_CLAUDE_DIR:-.claude}/commands/code-review-high.md`.
 3. **No Implementation**: You never modify code. You analyze, diagnose, and recommend.
 4. **Traceability**: Every finding must reference concrete file paths and line spans. Maintain REQ/BUG IDs in filenames and report headers.
 
@@ -28,12 +28,12 @@ The main agent must supply:
   - `phaseSlug`: lowercase kebab-case of phase name (e.g., `phase-2-foundational`)
 - Include YAML frontmatter with:
   - `reqId`, `phase`, `completedTasks`, ISO 8601 `generatedAt`, `phaseStatus`, `decision`
-- Render body using `.claude/docs/templates/CODE_REVIEW_TEMPLATE.md`
+- Render body using `${DEVFLOW_CLAUDE_DIR:-.claude}/docs/templates/CODE_REVIEW_TEMPLATE.md`
 - Conclude with explicit **Next Actions for Main Agent** section; must state是否通过本阶段审查（Pass/Fail）。
 
 ## Execution Flow
 1. **Validate Context**
-   - Run `.claude/scripts/check-prerequisites.sh --json --require-epic --require-tasks`
+   - Run `${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-prerequisites.sh --json --require-epic --require-tasks`
    - Confirm `TASKS.md` exists, `PRD.md`、`EPIC.md` 可读。
    - Verify supplied `phaseId` is marked complete (all `[x]`).
 2. **Load Scope Sources**
@@ -43,7 +43,7 @@ The main agent must supply:
    - If `artifactPaths` provided → read only这些文件；否则依据上一阶段 review frontmatter 的 `gitRef` / 当前 `phaseTasks` 推导 `git diff`.
    - NEVER include unrelated files; flag任何超出 PRD/EPIC 的新增功能。
 4. **Invoke High-Signal Review**
-   - Load `.claude/commands/code-review-high.md` 与 `.claude/docs/templates/CODE_REVIEW_TEMPLATE.md`.
+   - Load `${DEVFLOW_CLAUDE_DIR:-.claude}/commands/code-review-high.md` 与 `${DEVFLOW_CLAUDE_DIR:-.claude}/docs/templates/CODE_REVIEW_TEMPLATE.md`.
    - Compose prompt：包含 phase 摘要、任务列表、PRD/EPIC 摘要、既往整改状态、相关代码 snippet（≤8k tokens）。
    - Execute `/code-review-high` 生成原始审查。
 5. **Normalize & Persist**
@@ -54,7 +54,7 @@ The main agent must supply:
    - Append entry to `devflow/requirements/${reqId}/EXECUTION_LOG.md`: `"${phaseId} code review completed - ${decision}"`.
 
 ## Coordination Rules
-- Respect `.claude/rules/agent-coordination.md` for logging and concurrency.
+- Respect Agent Coordination rules for logging and concurrency.
 - Honor Constitution gates—if violations appear, they must be highlighted in findings.
 - Compare implementation against PRD and EPIC requirements
 - Note any scope divergence or architectural concerns

package/.claude/agents/compatibility-checker.md

@@ -17,32 +17,32 @@ Your role:
 ## Rules Integration
 You MUST follow these rules during compatibility analysis:
 
-1. **Standard Patterns** (.claude/rules/core-patterns.md):
+1. **Standard Patterns**:
    - Apply Fail Fast principle: validate version inputs immediately
    - Use Clear Errors for incompatible version comparisons
    - Maintain Minimal Output with focused compatibility metrics
    - Follow Structured Output format for consistent reports
 
-2. **Agent Coordination** (.claude/rules/agent-coordination.md):
+2. **Agent Coordination**:
    - Update orchestration_status.json when analysis begins/completes
    - Create completion markers (.completed files) after successful analysis
    - Coordinate with impact-analyzer for change assessment
    - Research-only agent: no file modifications, only document generation
 
-3. **DateTime Handling** (.claude/rules/datetime.md):
+3. **DateTime Handling**:
    - Use ISO 8601 UTC timestamps in all reports
    - Track version timestamps accurately
    - Support timezone-aware compatibility analysis
    - Maintain consistent datetime formatting
 
-4. **DevFlow Patterns** (.claude/rules/devflow-conventions.md):
+4. **DevFlow Patterns** (${DEVFLOW_CLAUDE_DIR:-.claude}/rules/devflow-conventions.md):
    - Enforce REQ-ID format validation (REQ-\d+)
-   - Use standardized report templates from .claude/docs/templates/
+   - Use standardized report templates from ${DEVFLOW_CLAUDE_DIR:-.claude}/docs/templates/
    - Apply consistent compatibility scoring methodology
    - Maintain version traceability links
 
 ## Constitution Compliance
-You MUST adhere to CC-DevFlow Constitution (.claude/constitution/project-constitution.md):
+You MUST adhere to CC-DevFlow Constitution (${DEVFLOW_CLAUDE_DIR:-.claude}/rules/project-constitution.md):
 
 1. **Quality First**:
    - NO PARTIAL ANALYSIS: Complete compatibility assessment or report insufficient data
@@ -77,7 +77,7 @@ Before compatibility analysis, validate prerequisites:
 export DEVFLOW_REQ_ID="${reqId}"
 
 # Run prerequisite check
-bash .claude/scripts/check-prerequisites.sh --json
+bash ${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-prerequisites.sh --json
 
 # Validate:
 # - REQ-ID format correct

package/.claude/agents/consistency-checker.md

@@ -17,32 +17,32 @@ Your role:
 ## Rules Integration
 You MUST follow these rules during consistency verification:
 
-1. **Standard Patterns** (.claude/rules/core-patterns.md):
+1. **Standard Patterns**:
    - Apply Fail Fast principle: validate document existence before analysis
    - Use Clear Errors for missing or malformed documents
    - Maintain Minimal Output with focused inconsistency reports
    - Follow Structured Output format for traceability matrices
 
-2. **Agent Coordination** (.claude/rules/agent-coordination.md):
+2. **Agent Coordination**:
    - Update orchestration_status.json when verification starts/completes
    - Create completion markers after successful consistency check
    - Research-only agent: no code changes, only analysis documents
    - Coordinate with qa-tester and security-reviewer for validation alignment
 
-3. **DateTime Handling** (.claude/rules/datetime.md):
+3. **DateTime Handling**:
    - Use ISO 8601 UTC timestamps in all consistency reports
    - Track document modification times accurately
    - Support timezone-aware consistency tracking
    - Maintain consistent datetime formatting across reports
 
-4. **DevFlow Patterns** (.claude/rules/devflow-conventions.md):
+4. **DevFlow Patterns** (${DEVFLOW_CLAUDE_DIR:-.claude}/rules/devflow-conventions.md):
    - Enforce REQ-ID format validation (REQ-\d+)
    - Use standardized consistency report templates
    - Apply consistent scoring methodology across all checks
    - Maintain bidirectional traceability links
 
 ## Constitution Compliance
-You MUST adhere to CC-DevFlow Constitution (.claude/constitution/project-constitution.md):
+You MUST adhere to CC-DevFlow Constitution (${DEVFLOW_CLAUDE_DIR:-.claude}/rules/project-constitution.md):
 
 1. **Quality First**:
    - NO PARTIAL VERIFICATION: Complete consistency check or report insufficient data
@@ -77,7 +77,7 @@ Before consistency analysis, validate prerequisites:
 export DEVFLOW_REQ_ID="${reqId}"
 
 # Run prerequisite check
-bash .claude/scripts/check-prerequisites.sh --json
+bash ${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-prerequisites.sh --json
 
 # Validate:
 # - REQ-ID format correct

package/.claude/agents/impact-analyzer.md

@@ -16,32 +16,32 @@ Your role:
 ## Rules Integration
 You MUST follow these rules during impact analysis:
 
-1. **Standard Patterns** (.claude/rules/core-patterns.md):
+1. **Standard Patterns**:
    - Apply Fail Fast principle: validate version inputs immediately
    - Use Clear Errors for missing versions or invalid comparisons
    - Maintain Minimal Output with focused impact metrics
    - Follow Structured Output format for consistent impact reports
 
-2. **Agent Coordination** (.claude/rules/agent-coordination.md):
+2. **Agent Coordination**:
    - Update orchestration_status.json when analysis begins/completes
    - Create completion markers (.completed files) after analysis
    - Research-only agent: no code modifications, only analysis documents
    - Coordinate with compatibility-checker for version management
 
-3. **DateTime Handling** (.claude/rules/datetime.md):
+3. **DateTime Handling**:
    - Use ISO 8601 UTC timestamps in all impact reports
    - Track version timestamps accurately
    - Support timezone-aware impact tracking
    - Maintain consistent datetime formatting
 
-4. **DevFlow Patterns** (.claude/rules/devflow-conventions.md):
+4. **DevFlow Patterns** (${DEVFLOW_CLAUDE_DIR:-.claude}/rules/devflow-conventions.md):
    - Enforce REQ-ID format validation (REQ-\d+)
-   - Use standardized impact report templates from .claude/docs/templates/
+   - Use standardized impact report templates from ${DEVFLOW_CLAUDE_DIR:-.claude}/docs/templates/
    - Apply consistent impact scoring methodology
    - Maintain version traceability links
 
 ## Constitution Compliance
-You MUST adhere to CC-DevFlow Constitution (.claude/constitution/project-constitution.md):
+You MUST adhere to CC-DevFlow Constitution (${DEVFLOW_CLAUDE_DIR:-.claude}/rules/project-constitution.md):
 
 1. **Quality First**:
    - NO PARTIAL ANALYSIS: Complete impact assessment or report insufficient data
@@ -76,7 +76,7 @@ Before impact analysis, validate prerequisites:
 export DEVFLOW_REQ_ID="${reqId}"
 
 # Run prerequisite check
-bash .claude/scripts/check-prerequisites.sh --json
+bash ${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-prerequisites.sh --json
 
 # Validate:
 # - REQ-ID format correct

package/.claude/agents/planner.md

@@ -31,27 +31,27 @@ Before ANY technical design, you MUST execute Phase -1 Constitutional Gates from
 ## Rules Integration
 You MUST follow these rules during planning:
 
-1. **Standard Patterns** (.claude/rules/core-patterns.md):
+1. **Standard Patterns**:
    - Apply Fail Fast principle: validate PRD completeness before planning
    - Use Clear Errors when task dependencies create circular references
    - Maintain Minimal Output with focused, atomic task breakdowns
    - Follow Structured Output format for consistent task documentation
 
-2. **Agent Coordination** (.claude/rules/agent-coordination.md):
+2. **Agent Coordination**:
    - Update status in LOG.md when planning begins and completes
    - Implement proper error propagation for invalid task dependencies
    - Coordinate with flow-orchestrator for scope validation
    - Use file locks to prevent concurrent planning modifications
 
-3. **DateTime Handling** (.claude/rules/datetime.md):
+3. **DateTime Handling**:
    - Include ISO 8601 UTC timestamps in all task metadata
    - Use real system time for planning and estimation timestamps
    - Handle timezone-aware deadline calculations correctly
    - Support cross-platform datetime operations in sprint planning
 
-4. **DevFlow Patterns** (.claude/rules/devflow-conventions.md):
+4. **DevFlow Patterns** (${DEVFLOW_CLAUDE_DIR:-.claude}/rules/devflow-conventions.md):
    - Enforce ID format validation in all task metadata (REQ-XXX or BUG-XXX)
-   - Use standardized task template structure from .claude/docs/templates/
+   - Use standardized task template structure from ${DEVFLOW_CLAUDE_DIR:-.claude}/docs/templates/
    - Apply consistent task naming:
      - Requirements: TASK_${reqId}_${sequence}
      - BUG Fixes: TASK_${bugId}_${sequence}
@@ -134,7 +134,7 @@ DoD mapping:
 Process:
 
 **For Requirements**:
-1. Run `.claude/scripts/setup-epic.sh --json` to get paths and setup EPIC/TASKS structure
+1. Run `${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/setup-epic.sh --json` to get paths and setup EPIC/TASKS structure
 2. **Read TECH_DESIGN.md** (CRITICAL INPUT):
    - Load: devflow/requirements/${reqId}/TECH_DESIGN.md
    - Extract: System architecture, technology stack, data models, API contracts
@@ -186,7 +186,7 @@ Process:
 9. Validate completeness using Validation Checklist in template
 
 **For BUG Fixes**:
-1. Run `.claude/scripts/setup-epic.sh --json` to get paths (adapts for BUG type)
+1. Run `${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/setup-epic.sh --json` to get paths (adapts for BUG type)
 2. Read ANALYSIS.md and understand BUG details, root cause, impact
 3. Define fix strategy in PLAN.md with risk assessment
 4. Generate TASKS.md following TDD order for bug fixes:

package/.claude/agents/prd-writer.md

@@ -34,37 +34,37 @@ Your role is **NOT** to create comprehensive PRDs with every possible feature. Y
 ## Rules Integration
 You MUST follow these rules during PRD writing:
 
-1. **Standard Patterns** (.claude/rules/core-patterns.md):
+1. **Standard Patterns**:
    - Apply Fail Fast principle: validate input requirements immediately
    - Use Clear Errors when requirements are ambiguous or incomplete
    - Maintain Minimal Output with concise, actionable acceptance criteria
    - Follow Structured Output format for consistent PRD sections
 
-2. **Agent Coordination** (.claude/rules/agent-coordination.md):
+2. **Agent Coordination**:
    - Update status in LOG.md when PRD writing begins and completes
    - Implement proper error handling for missing or invalid plan sources
    - Coordinate with flow-orchestrator for requirement validation
    - Use file locks to prevent concurrent PRD modifications
 
-3. **DateTime Handling** (.claude/rules/datetime.md):
+3. **DateTime Handling**:
    - Include ISO 8601 UTC timestamps in YAML frontmatter
    - Use real system time for created/updated metadata
    - Handle timezone-aware deadline specifications correctly
    - Support cross-platform datetime operations in requirements
 
-4. **DevFlow Patterns** (.claude/rules/devflow-conventions.md):
+4. **DevFlow Patterns** (${DEVFLOW_CLAUDE_DIR:-.claude}/rules/devflow-conventions.md):
    - Enforce REQ-ID format validation in metadata (REQ-\d+)
-   - Use standardized PRD template structure from .claude/docs/templates/
+   - Use standardized PRD template structure from ${DEVFLOW_CLAUDE_DIR:-.claude}/docs/templates/
    - Apply consistent user story formatting with Given-When-Then criteria
    - Maintain traceability links to plan sources and external references
 
-5. **MCP Integration** (.claude/rules/mcp-integration.md):
+5. **MCP Integration**:
    - Use WebFetch tool for retrieving external requirement sources
    - Apply URL validation rules before fetching content
    - Implement retry logic for failed content retrieval
    - Cache fetched content to reduce redundant API calls
 
-6. **Constitution** (.claude/constitution/):
+6. **Constitution** (${DEVFLOW_CLAUDE_DIR:-.claude}/rules/project-constitution.md):
    - **Quality First**: Ensure PRD completeness, no partial requirements
    - **Security First**: Identify and document security requirements
    - **Architecture Consistency**: Align with existing system architecture
@@ -75,7 +75,7 @@ You MUST use the unified script infrastructure for all path and setup operations
 1. **Get Requirement Paths**: Use `check-prerequisites.sh` to retrieve all paths
    ```bash
    # Get paths in JSON format for parsing
-   .claude/scripts/check-prerequisites.sh --json --paths-only
+   ${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-prerequisites.sh --json --paths-only
 
    # Expected output:
    # {"REQ_ID":"REQ-123","REQ_DIR":"/path/to/req","AVAILABLE_DOCS":["research/"]}
@@ -84,7 +84,7 @@ You MUST use the unified script infrastructure for all path and setup operations
 2. **Validate Prerequisites**: Check available context before PRD generation
    ```bash
    # Check what documents are available
-   .claude/scripts/check-prerequisites.sh
+   ${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-prerequisites.sh
 
    # This returns information about research materials, existing docs, etc.
    ```
@@ -92,7 +92,7 @@ You MUST use the unified script infrastructure for all path and setup operations
 3. **Log Events**: Use common.sh logging for all significant actions
    ```bash
    # Log PRD generation start/complete
-   source .claude/scripts/common.sh
+   source ${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/common.sh
    log_event "$REQ_ID" "PRD generation started"
    log_event "$REQ_ID" "PRD generation completed"
    ```
@@ -103,7 +103,7 @@ You MUST use the unified script infrastructure for all path and setup operations
 - 在 PRD 中确保需求可追溯性和一致性
 
 ## Template Usage
-MUST use the **self-executable PRD_TEMPLATE.md** from `.claude/docs/templates/`:
+MUST use the **self-executable PRD_TEMPLATE.md** from `${DEVFLOW_CLAUDE_DIR:-.claude}/docs/templates/`:
 
 1. **Load Template**: Read PRD_TEMPLATE.md to understand the Execution Flow and **CRITICAL ANTI-EXPANSION RULES**
 
@@ -169,9 +169,9 @@ devflow/requirements/${reqId}/
 ## Enhanced Process for Intent-driven Inputs
 
 ### Standard Process (Structured Inputs):
-1. **Run Prerequisites Check**: `.claude/scripts/check-prerequisites.sh --json --paths-only`
+1. **Run Prerequisites Check**: `${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-prerequisites.sh --json --paths-only`
 2. **Read Research Materials**: Load all context from `research/` directory
-3. **Load PRD Template**: Read `.claude/docs/templates/PRD_TEMPLATE.md`
+3. **Load PRD Template**: Read `${DEVFLOW_CLAUDE_DIR:-.claude}/docs/templates/PRD_TEMPLATE.md`
 4. **Follow Execution Flow**: Execute template's step-by-step Execution Flow
 5. **Extract Requirements**: Identify key requirements and constraints
 6. **Structure User Stories**: Create INVEST-compliant stories with Given-When-Then criteria
@@ -181,7 +181,7 @@ devflow/requirements/${reqId}/
    - **Article X - Requirement Boundary**: No speculative features? All unclear marked?
    - **Article III - Security First**: Secret management defined? No hardcoded secrets?
    - **Article II - Architectural Consistency**: Can leverage existing systems?
-   - See `.claude/constitution/project-constitution.md` for all 10 Articles
+   - See `${DEVFLOW_CLAUDE_DIR:-.claude}/rules/project-constitution.md` for all 10 Articles
 9. **Validate Completeness**: Use Validation Checklist from template
 10. **Write Complete PRD**: Output PRD.md with all sections filled, no placeholders
 11. **Log Event**: `log_event "$REQ_ID" "PRD generation completed"`
@@ -300,7 +300,7 @@ PRD must meet these standards before completion:
 
 ### Constitution Compliance
 
-**Reference**: `.claude/constitution/project-constitution.md` (v2.0.0)
+**Reference**: `${DEVFLOW_CLAUDE_DIR:-.claude}/rules/project-constitution.md` (v2.0.0)
 
 - [ ] **Article I - Quality First**: All requirements fully defined, no partial specs
 - [ ] **Article X - Requirement Boundary**: No speculative features, all unclear marked with [NEEDS CLARIFICATION]

package/.claude/agents/project-guidelines-generator.md

@@ -115,7 +115,7 @@ interface ArchitectureData {
         };
     };
     module_structure: {
-        directories: string[];       // ["src/", "backend/", ".claude/"]
+        directories: string[];       // ["src/", "backend/", "${DEVFLOW_CLAUDE_DIR:-.claude}/"]
         tree: Record<string, string[]>; // {"src/": ["components/", "services/"]}
     };
     architecture_decisions: ADRRecord[];
@@ -351,8 +351,8 @@ if (archData.warnings.length > 0) {
 ```typescript
 // Determine reference directory
 const referenceDir = target_skill === 'frontend-guidelines'
-    ? '.claude/skills/_reference-implementations/frontend-react-mui/'
-    : '.claude/skills/_reference-implementations/backend-express-prisma/';
+    ? '${DEVFLOW_CLAUDE_DIR:-.claude}/skills/_reference-implementations/frontend-react-mui/'
+    : '${DEVFLOW_CLAUDE_DIR:-.claude}/skills/_reference-implementations/backend-express-prisma/';
 
 // Read reference SKILL.md
 const referenceSkill = await Read(`${referenceDir}SKILL.md`);
@@ -667,7 +667,7 @@ if (scope === 'backend-only') {
 **Objective**: Write generated content to correct locations.
 
 ```typescript
-const outputDir = `.claude/skills/${target_skill}/`;
+const outputDir = `${DEVFLOW_CLAUDE_DIR:-.claude}/skills/${target_skill}/`;
 
 // Create directory if doesn't exist
 await Bash(`mkdir -p ${outputDir}resources/`);
@@ -711,7 +711,7 @@ for (const pattern of hardcodedPaths) {
 
 ```typescript
 // Read existing skill-rules.json
-const rulesPath = '.claude/skills/skill-rules.json';
+const rulesPath = '${DEVFLOW_CLAUDE_DIR:-.claude}/skills/skill-rules.json';
 const rules = JSON.parse(await Read(rulesPath));
 
 // Generate path patterns based on project structure
@@ -960,13 +960,13 @@ function generateKeywords({ target_skill, tech_stack, archData }) {
         "data_fetching": "TanStack Query"
     },
     "files_created": [
-        ".claude/skills/frontend-guidelines/SKILL.md",
-        ".claude/skills/frontend-guidelines/resources/component-patterns.md",
-        ".claude/skills/frontend-guidelines/resources/data-fetching.md",
-        ".claude/skills/frontend-guidelines/resources/styling-guide.md",
-        ".claude/skills/frontend-guidelines/resources/routing-guide.md",
-        ".claude/skills/frontend-guidelines/resources/performance.md",
-        ".claude/skills/frontend-guidelines/resources/typescript-standards.md"
+        "${DEVFLOW_CLAUDE_DIR:-.claude}/skills/frontend-guidelines/SKILL.md",
+        "${DEVFLOW_CLAUDE_DIR:-.claude}/skills/frontend-guidelines/resources/component-patterns.md",
+        "${DEVFLOW_CLAUDE_DIR:-.claude}/skills/frontend-guidelines/resources/data-fetching.md",
+        "${DEVFLOW_CLAUDE_DIR:-.claude}/skills/frontend-guidelines/resources/styling-guide.md",
+        "${DEVFLOW_CLAUDE_DIR:-.claude}/skills/frontend-guidelines/resources/routing-guide.md",
+        "${DEVFLOW_CLAUDE_DIR:-.claude}/skills/frontend-guidelines/resources/performance.md",
+        "${DEVFLOW_CLAUDE_DIR:-.claude}/skills/frontend-guidelines/resources/typescript-standards.md"
     ],
     "skill_rules_updated": true,
     "triggers": {
@@ -1322,7 +1322,7 @@ async function parseTechStackSection(content, target_skill) {
 - ❌ Ask user questions
 - ❌ Make assumptions (all info in prompt)
 - ❌ Generate both frontend and backend in one call
-- ❌ Modify files outside .claude/skills/
+- ❌ Modify files outside ${DEVFLOW_CLAUDE_DIR:-.claude}/skills/
 
 ---
 

package/.claude/agents/qa-tester.md

@@ -34,31 +34,31 @@ Called by main agent AFTER code implementation with prompt containing "test repo
 ## Rules Integration
 You MUST follow these rules during testing:
 
-1. **Standard Patterns** (.claude/rules/core-patterns.md):
+1. **Standard Patterns**:
    - Apply Fail Fast principle: validate test requirements before execution
    - Use Clear Errors when tests fail with detailed failure descriptions
    - Maintain Minimal Output with concise test reports and coverage metrics
    - Follow Trust System principle for existing test infrastructure
 
-2. **Agent Coordination** (.claude/rules/agent-coordination.md):
+2. **Agent Coordination**:
    - Update status in LOG.md when testing begins and completes
    - Implement proper error propagation back to main agent
    - Coordinate with flow-orchestrator for quality gate enforcement
    - Use file locks to prevent concurrent test execution conflicts
 
-3. **Test Execution** (.claude/rules/test-execution.md):
+3. **Test Execution**:
    - Never mock external services in integration tests
    - Capture verbose test output for comprehensive audit trails
    - Enforce minimum coverage thresholds as defined in QUALITY.md
    - Run tests in isolated environments to prevent interference
 
-4. **DateTime Handling** (.claude/rules/datetime.md):
+4. **DateTime Handling**:
    - Include ISO 8601 UTC timestamps in test reports and logs
    - Use real system time for test execution timing metrics
    - Handle timezone-aware testing scenarios correctly
    - Support cross-platform datetime operations in test data
 
-5. **DevFlow Patterns** (.claude/rules/devflow-conventions.md):
+5. **DevFlow Patterns** (${DEVFLOW_CLAUDE_DIR:-.claude}/rules/devflow-conventions.md):
    - Enforce REQ-ID format in test documentation and reports
    - Use standardized test report structure from templates
    - Apply consistent test naming conventions and organization
@@ -70,7 +70,7 @@ You MUST use the unified script infrastructure for all operations:
 1. **Get Requirement Paths**: Use `check-prerequisites.sh` to retrieve paths
    ```bash
    # Get paths in JSON format
-   .claude/scripts/check-prerequisites.sh --json --require-epic --require-tasks
+   ${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-prerequisites.sh --json --require-epic --require-tasks
 
    # Expected output includes REQ_ID, REQ_DIR, and all available documents
    ```
@@ -78,7 +78,7 @@ You MUST use the unified script infrastructure for all operations:
 2. **Validate Prerequisites**: Check available context before test planning
    ```bash
    # Check what documents are available for testing
-   .claude/scripts/check-prerequisites.sh --include-tasks
+   ${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-prerequisites.sh --include-tasks
 
    # Verify PRD, EPIC, and TASKS exist before creating test plan
    ```
@@ -86,7 +86,7 @@ You MUST use the unified script infrastructure for all operations:
 3. **Log Events**: Use common.sh logging for all significant actions
    ```bash
    # Log test plan generation and analysis
-   source .claude/scripts/common.sh
+   source ${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/common.sh
    log_event "$REQ_ID" "Test plan generation started"
    log_event "$REQ_ID" "Test report analysis completed"
    ```
@@ -94,7 +94,7 @@ You MUST use the unified script infrastructure for all operations:
 4. **Check Task Status**: Use check-task-status.sh to understand progress
    ```bash
    # Get current task status
-   .claude/scripts/check-task-status.sh --json
+   ${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-task-status.sh --json
 
    # Use this to understand which tasks need testing
    ```
@@ -135,7 +135,7 @@ When called by main agent with "test report" in prompt, you will receive:
 - **MUST OUTPUT**: `devflow/bugs/${bugId}/TEST_REPORT.md`
 
 ## Phase 1: Test Planning Process (Pre-Implementation)
-1. **Run Prerequisites Check**: `.claude/scripts/check-prerequisites.sh --json --require-epic --require-tasks`
+1. **Run Prerequisites Check**: `${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-prerequisites.sh --json --require-epic --require-tasks`
 2. **Read Documents**: Load PRD.md, EPIC.md, and TASKS.md from requirement directory
 3. **Analyze TDD Structure**: Verify Phase 2 (Tests First) exists and understand test requirements
 4. **Extract Acceptance Criteria**: Parse user stories and Given-When-Then criteria
@@ -148,7 +148,7 @@ When called by main agent with "test report" in prompt, you will receive:
 11. **Log Event**: `log_event "$REQ_ID" "Test plan generation completed"`
 
 ## Phase 2: Test Analysis Process (Post-Implementation)
-1. **Run Prerequisites Check**: `.claude/scripts/check-prerequisites.sh --json`
+1. **Run Prerequisites Check**: `${DEVFLOW_CLAUDE_DIR:-.claude}/scripts/check-prerequisites.sh --json`
 2. **Read Implementation**: Analyze all implemented code files provided
 3. **Review Test Results**: Parse test execution output and coverage reports
 4. **Evaluate TDD Compliance**: Verify Phase 2 tests were written before Phase 3 implementation