From 4bb4bdc124e6d14eab3f0e63158867a05308fcfc Mon Sep 17 00:00:00 2001
From: catlog22 <catlog22@github.com>
Date: Sun, 23 Nov 2025 22:06:13 +0800
Subject: [PATCH] refactor: migrate agents to path-based context loading

- Update action-planning-agent and task-generate-agent to load context
  via file paths instead of embedded context packages
- Streamline test-cycle-execute workflow execution
- Remove redundant content from conflict-resolution and context-gather
- Update SKILL.md and tdd-plan documentation
---
 .claude/agents/action-planning-agent.md       | 162 ++++++----
 .../commands/workflow/test-cycle-execute.md   | 290 +++++++++--------
 .../workflow/tools/conflict-resolution.md     |   5 -
 .../commands/workflow/tools/context-gather.md |   2 -
 .../workflow/tools/task-generate-agent.md     | 292 ++++++------------
 .claude/skills/command-guide/SKILL.md         |   4 +-
 .../reference/commands/workflow/tdd-plan.md   |  46 ---
 7 files changed, 348 insertions(+), 453 deletions(-)

diff --git a/.claude/agents/action-planning-agent.md b/.claude/agents/action-planning-agent.md
index 38a98932..a4ad0251 100644
--- a/.claude/agents/action-planning-agent.md
+++ b/.claude/agents/action-planning-agent.md
@@ -21,15 +21,15 @@ You are a pure execution agent specialized in creating actionable implementation
 ## Execution Process
 
 ### Input Processing
-**What you receive:**
-- **Execution Context Package**: Structured context from command layer
+
+**What you receive from command layer:**
+- **Session Paths**: File paths to load content autonomously
+  - `session_metadata_path`: Session configuration and user input
+  - `context_package_path`: Context package with brainstorming artifacts catalog
+- **Metadata**: Simple values
   - `session_id`: Workflow session identifier (WFS-[topic])
-  - `session_metadata`: Session configuration and state
-  - `analysis_results`: Analysis recommendations and task breakdown
-  - `artifacts_inventory`: Detected brainstorming outputs (role analyses, guidance-specification, role analyses)
-  - `context_package`: Project context and assets
-  - `mcp_capabilities`: Available MCP tools (exa-code, exa-web)
-  - `mcp_analysis`: Optional pre-executed MCP analysis results
+  - `execution_mode`: agent-mode | cli-execute-mode
+  - `mcp_capabilities`: Available MCP tools (exa_code, exa_web, code_index)
 
 **Legacy Support** (backward compatibility):
 - **pre_analysis configuration**: Multi-step array format with action, template, method fields
@@ -37,73 +37,103 @@ You are a pure execution agent specialized in creating actionable implementation
 - **Task requirements**: Direct task description
 
 ### Execution Flow (Two-Phase)
+
 ```
-Phase 1: Context Validation & Enhancement (Discovery Results Provided)
-1. Receive and validate execution context package
-2. Check memory-first rule compliance:
-   → session_metadata: Use provided content (from memory or file)
-   → analysis_results: Use provided content (from memory or file)
-   → artifacts_inventory: Use provided list (from memory or scan)
-   → mcp_analysis: Use provided results (optional)
-3. Optional MCP enhancement (if not pre-executed):
+Phase 1: Content Loading & Context Assembly
+1. Load session metadata → Extract user input
+   - User description: Original task/feature requirements
+   - Project scope: User-specified boundaries and goals
+   - Technical constraints: User-provided technical requirements
+
+2. Load context package → Extract key fields
+   - brainstorm_artifacts: Catalog of brainstorming outputs
+     - guidance_specification: Path to overall framework
+     - role_analyses[]: Array of role analysis files with priorities
+     - synthesis_output: Path to synthesis results (if exists)
+     - conflict_resolution: Conflict status and affected files
+   - focus_areas: Target directories for implementation
+   - assets: Existing code patterns to reuse
+   - conflict_risk: Risk level (low/medium/high)
+
+3. Load brainstorming artifacts (in priority order)
+   a. guidance-specification.md (Highest Priority)
+      → Overall design framework and architectural decisions
+   b. Role analyses (High Priority - load ALL files)
+      → system-architect/analysis.md
+      → subject-matter-expert/analysis.md
+      → (Other roles as listed in context package)
+   c. Synthesis output (if exists)
+      → Integrated view with clarifications
+   d. Conflict resolution (if conflict_risk ≥ medium)
+      → Review resolved conflicts in artifacts
+
+4. Optional MCP enhancement
    → mcp__exa__get_code_context_exa() for best practices
    → mcp__exa__web_search_exa() for external research
-4. Assess task complexity (simple/medium/complex) from analysis
+
+5. Assess task complexity (simple/medium/complex)
 
 Phase 2: Document Generation (Autonomous Output)
-1. Extract task definitions from analysis_results
-2. Generate task JSON files with 5-field schema + artifacts
+1. Synthesize requirements from all sources (user input + brainstorming artifacts)
+2. Generate task JSON files with 6-field schema + artifacts integration
 3. Create IMPL_PLAN.md with context analysis and artifact references
 4. Generate TODO_LIST.md with proper structure (▸, [ ], [x])
 5. Update session state for execution readiness
 ```
 
-### Context Package Usage
+### Context Package Fields to Load
 
-**Standard Context Structure**:
+**Load from `context_package_path` - fields defined by context-search-agent**:
+
+**Always Present**:
+- `metadata.task_description`: User's original task description
+- `metadata.keywords`: Extracted technical keywords
+- `metadata.complexity`: Task complexity level (simple/medium/complex)
+- `metadata.session_id`: Workflow session identifier
+- `project_context.architecture_patterns`: Architecture patterns (MVC, Service layer, etc.)
+- `project_context.tech_stack`: Language, frameworks, libraries
+- `project_context.coding_conventions`: Naming, error handling, async patterns
+- `assets.source_code[]`: Relevant existing files with paths and metadata
+- `assets.documentation[]`: Reference docs (CLAUDE.md, API docs)
+- `assets.config[]`: Configuration files (package.json, .env.example)
+- `assets.tests[]`: Test files
+- `dependencies.internal[]`: Module dependencies
+- `dependencies.external[]`: Package dependencies
+- `conflict_detection.risk_level`: Conflict risk (low/medium/high)
+
+**Conditionally Present** (check existence before loading):
+- `brainstorm_artifacts.guidance_specification`: Overall design framework (if exists)
+  - Check: `brainstorm_artifacts?.guidance_specification?.exists === true`
+  - Content: Use `content` field if present, else load from `path`
+- `brainstorm_artifacts.role_analyses[]`: Role-specific analyses (if array not empty)
+  - Each role: `role_analyses[i].files[j]` has `path` and `content`
+- `brainstorm_artifacts.synthesis_output`: Synthesis results (if exists)
+  - Check: `brainstorm_artifacts?.synthesis_output?.exists === true`
+  - Content: Use `content` field if present, else load from `path`
+- `conflict_detection.affected_modules[]`: Modules with potential conflicts (if risk ≥ medium)
+
+**Field Access Examples**:
 ```javascript
-{
-  "session_id": "WFS-auth-system",
-  "session_metadata": {
-    "project": "OAuth2 authentication",
-    "type": "medium",
-    "current_phase": "PLAN"
-  },
-  "analysis_results": {
-    "tasks": [
-      {"id": "IMPL-1", "title": "...", "requirements": [...]}
-    ],
-    "complexity": "medium",
-    "dependencies": [...]
-  },
-  "artifacts_inventory": {
-    "synthesis_specification": ".workflow/WFS-auth/.brainstorming/role analysis documents",
-    "topic_framework": ".workflow/WFS-auth/.brainstorming/guidance-specification.md",
-    "role_analyses": [
-      ".workflow/WFS-auth/.brainstorming/system-architect/analysis.md",
-      ".workflow/WFS-auth/.brainstorming/subject-matter-expert/analysis.md"
-    ]
-  },
-  "context_package": {
-    "assets": [...],
-    "focus_areas": [...]
-  },
-  "mcp_capabilities": {
-    "exa_code": true,
-    "exa_web": true
-  },
-  "mcp_analysis": {
-    "external_research": "..."
-  }
+// Always safe - direct field access
+const techStack = contextPackage.project_context.tech_stack;
+const riskLevel = contextPackage.conflict_detection.risk_level;
+const existingCode = contextPackage.assets.source_code; // Array of files
+
+// Conditional - use content if available, else load from path
+if (contextPackage.brainstorm_artifacts?.guidance_specification?.exists) {
+  const spec = contextPackage.brainstorm_artifacts.guidance_specification;
+  const content = spec.content || Read(spec.path);
+}
+
+if (contextPackage.brainstorm_artifacts?.role_analyses?.length > 0) {
+  contextPackage.brainstorm_artifacts.role_analyses.forEach(role => {
+    role.files.forEach(file => {
+      const analysis = file.content || Read(file.path);
+    });
+  });
 }
 ```
 
-**Using Context in Task Generation**:
-1. **Extract Tasks**: Parse `analysis_results.tasks` array
-2. **Map Artifacts**: Use `artifacts_inventory` to add artifact references to task.context
-3. **Assess Complexity**: Use `analysis_results.complexity` for document structure decision
-4. **Session Paths**: Use `session_id` to construct output paths (.workflow/active/{session_id}/)
-
 ### MCP Integration Guidelines
 
 **Exa Code Context** (`mcp_capabilities.exa_code = true`):
@@ -503,13 +533,13 @@ Use `analysis_results.complexity` or task count to determine structure:
 - Flat structure: IMPL_PLAN.md + TODO_LIST.md + task JSONs
 - No container tasks, all leaf tasks
 
-**Medium Tasks** (6-10 tasks):
+**Medium Tasks** (6-12 tasks):
 - Two-level hierarchy: IMPL_PLAN.md + TODO_LIST.md + task JSONs
 - Optional container tasks for grouping
 
-**Complex Tasks** (>10 tasks):
-- **Re-scope required**: Maximum 10 tasks hard limit
-- If analysis_results contains >10 tasks, consolidate or request re-scoping
+**Complex Tasks** (>12 tasks):
+- **Re-scope required**: Maximum 12 tasks hard limit
+- If analysis_results contains >12 tasks, consolidate or request re-scoping
 
 ## Quantification Requirements (MANDATORY)
 
@@ -566,7 +596,7 @@ Use `analysis_results.complexity` or task count to determine structure:
 - **Follow 5-field schema**: All task JSONs must have id, title, status, meta, context, flow_control
 - **Map artifacts**: Use artifacts_inventory to populate task.context.artifacts array
 - **Add MCP integration**: Include MCP tool steps in flow_control.pre_analysis when capabilities available
-- **Validate task count**: Maximum 10 tasks hard limit, request re-scope if exceeded
+- **Validate task count**: Maximum 12 tasks hard limit, request re-scope if exceeded
 - **Use session paths**: Construct all paths using provided session_id
 - **Link documents properly**: Use correct linking format (📋 for JSON, ✅ for summaries)
 - **Run validation checklist**: Verify all quantification requirements before finalizing task JSONs
@@ -575,6 +605,6 @@ Use `analysis_results.complexity` or task count to determine structure:
 - Load files directly (use provided context package instead)
 - Assume default locations (always use session_id in paths)
 - Create circular dependencies in task.depends_on
-- Exceed 10 tasks without re-scoping
+- Exceed 12 tasks without re-scoping
 - Skip artifact integration when artifacts_inventory is provided
 - Ignore MCP capabilities when available
diff --git a/.claude/commands/workflow/test-cycle-execute.md b/.claude/commands/workflow/test-cycle-execute.md
index 6a0155e9..d385f1c2 100644
--- a/.claude/commands/workflow/test-cycle-execute.md
+++ b/.claude/commands/workflow/test-cycle-execute.md
@@ -278,57 +278,85 @@ Task(
   subagent_type="cli-planning-agent",
   description=`Analyze test failures and generate fix task (iteration ${currentIteration})`,
   prompt=`
-    ## Context Package
-    {
-      "session_id": "${sessionId}",
-      "iteration": ${currentIteration},
-      "analysis_type": "test-failure",
-      "failure_context": {
-        "failed_tests": ${JSON.stringify(failedTests)},
-        "error_messages": ${JSON.stringify(errorMessages)},
-        "test_output": "${testOutputPath}",
-        "pass_rate": ${passRate},
-        "previous_attempts": ${JSON.stringify(previousAttempts)}
-      },
-      "cli_config": {
-        "tool": "gemini",
-        "model": "gemini-3-pro-preview-11-2025",
-        "template": "01-diagnose-bug-root-cause.txt",
-        "timeout": 2400000,
-        "fallback": "qwen"
-      },
-      "task_config": {
-        "agent": "@test-fix-agent",
-        "type": "test-fix-iteration",
-        "max_iterations": ${maxIterations},
-        "use_codex": false
-      }
-    }
+    ## Task Objective
+    Analyze test failures and generate structured fix task JSON for iteration ${currentIteration}
 
-    ## Your Task
-    1. Execute CLI analysis using Gemini (fallback to Qwen if needed)
-    2. Parse CLI output and extract fix strategy with specific modification points
-    3. Generate IMPL-fix-${currentIteration}.json using your internal task template
-    4. Save analysis report to .process/iteration-${currentIteration}-analysis.md
-    5. Report success and task ID back to orchestrator
+    ## MANDATORY FIRST STEPS
+    1. Read test results: {session.test_results_path}
+    2. Read test output: {session.test_output_path}
+    3. Read iteration state: {session.iteration_state_path}
+    4. Read fix history (if exists): {session.fix_history_path}
+
+    ## Session Paths
+    - Workflow Dir: {session.workflow_dir}
+    - Test Results: {session.test_results_path}
+    - Test Output: {session.test_output_path}
+    - Iteration State: {session.iteration_state_path}
+    - Fix History: {session.fix_history_path}
+    - Task Output Dir: {session.task_dir}
+    - Analysis Output: {session.process_dir}/iteration-${currentIteration}-analysis.md
+    - CLI Output: {session.process_dir}/iteration-${currentIteration}-cli-output.txt
+
+    ## Context Metadata
+    - Session ID: ${sessionId}
+    - Current Iteration: ${currentIteration}
+    - Max Iterations: ${maxIterations}
+    - Current Pass Rate: ${passRate}%
+
+    ## CLI Configuration
+    - Tool: gemini (fallback: qwen)
+    - Model: gemini-3-pro-preview-11-2025
+    - Template: 01-diagnose-bug-root-cause.txt
+    - Timeout: 2400000ms
+
+    ## Expected Deliverables
+    1. Task JSON file: {session.task_dir}/IMPL-fix-${currentIteration}.json
+    2. Analysis report: {session.process_dir}/iteration-${currentIteration}-analysis.md
+    3. CLI raw output: {session.process_dir}/iteration-${currentIteration}-cli-output.txt
+    4. Return task ID to orchestrator
+
+    ## Quality Standards
+    - Fix strategy must include specific modification points (file:function:lines)
+    - Analysis must identify root causes, not just symptoms
+    - Task JSON must be valid and complete with all required fields
+    - All deliverables saved to specified paths
+
+    ## Success Criteria
+    Generate valid IMPL-fix-${currentIteration}.json with:
+    - Concrete fix strategy with modification points
+    - Root cause analysis from CLI tool
+    - All required task JSON fields (id, title, status, meta, context, flow_control)
+    - Return task ID for orchestrator to queue
   `
 )
 ```
 
-#### Agent Response
+#### Agent Response Format
+**Agent must return structured response with deliverable paths:**
+
 ```javascript
 {
   "status": "success",
   "task_id": "IMPL-fix-${iteration}",
-  "task_path": ".workflow/${session}/.task/IMPL-fix-${iteration}.json",
-  "analysis_report": ".process/iteration-${iteration}-analysis.md",
-  "cli_output": ".process/iteration-${iteration}-cli-output.txt",
-  "summary": "Fix authentication token validation and null check issues",
-  "modification_points_count": 2,
-  "estimated_complexity": "low"
+  "deliverables": {
+    "task_json": ".workflow/${session}/.task/IMPL-fix-${iteration}.json",
+    "analysis_report": ".workflow/${session}/.process/iteration-${iteration}-analysis.md",
+    "cli_output": ".workflow/${session}/.process/iteration-${iteration}-cli-output.txt"
+  },
+  "summary": {
+    "root_causes": ["Authentication token validation missing", "Null check missing"],
+    "modification_points": [
+      "src/auth/client.ts:sendRequest:45-50",
+      "src/validators/user.ts:validateUser:23-25"
+    ],
+    "estimated_complexity": "low",
+    "expected_pass_rate_improvement": "85% → 95%"
+  }
 }
 ```
 
+**Orchestrator validates all deliverable paths exist before proceeding.**
+
 #### Generated Analysis Report Structure
 The @cli-planning-agent generates `.process/iteration-N-analysis.md`:
 
@@ -503,54 +531,55 @@ TodoWrite({
 4. **Iteration Complete**: Mark iteration item completed
 5. **All Complete**: Mark parent task completed
 
-## Agent Context Package
+## Agent Context Loading
 
-**Generated by test-cycle-execute orchestrator before launching agents.**
+**Orchestrator provides file paths, agents load content themselves.**
 
-The orchestrator assembles this context package from:
-- Task JSON file (IMPL-*.json)
-- Iteration state files
-- Test results and failure context
-- Session metadata
+### Path References Provided to Agents
 
-This package is passed to agents via the Task tool's prompt context.
+**Orchestrator passes these paths via Task tool prompt:**
 
-### Enhanced Context for Test-Fix Agent
-```json
+```javascript
 {
-  "task": { /* IMPL-fix-N.json */ },
-  "iteration_context": {
-    "current_iteration": N,
-    "max_iterations": 5,
-    "previous_attempts": [
-      {
-        "iteration": N-1,
-        "failures": ["test1", "test2"],
-        "fixes_attempted": ["fix1", "fix2"],
-        "result": "partial_success"
-      }
-    ],
-    "failure_analysis": {
-      "source": "gemini_cli",
-      "analysis_file": ".process/iteration-N-analysis.md",
-      "fix_strategy": { /* from CLI */ }
-    }
-  },
-  "test_context": {
-    "test_framework": "jest|pytest|...",
-    "test_files": ["path/to/test1.test.ts"],
-    "test_command": "npm test",
-    "coverage_target": 80
-  },
-  "session": {
-    "workflow_dir": ".workflow/active/WFS-test-{session}/",
-    "iteration_state_file": ".process/iteration-state.json",
-    "test_results_file": ".process/test-results.json",
-    "fix_history_file": ".process/fix-history.json"
-  }
+  // Primary Task Definition
+  "task_json_path": ".workflow/active/WFS-test-{session}/.task/IMPL-fix-N.json",
+
+  // Iteration Context Paths
+  "iteration_state_path": ".workflow/active/WFS-test-{session}/.process/iteration-state.json",
+  "fix_history_path": ".workflow/active/WFS-test-{session}/.process/fix-history.json",
+
+  // Test Context Paths
+  "test_results_path": ".workflow/active/WFS-test-{session}/.process/test-results.json",
+  "test_output_path": ".workflow/active/WFS-test-{session}/.process/test-output.log",
+  "test_context_path": ".workflow/active/WFS-test-{session}/.process/TEST_ANALYSIS_RESULTS.md",
+
+  // Analysis & Strategy Paths (for fix-iteration tasks)
+  "analysis_path": ".workflow/active/WFS-test-{session}/.process/iteration-N-analysis.md",
+  "cli_output_path": ".workflow/active/WFS-test-{session}/.process/iteration-N-cli-output.txt",
+
+  // Session Management Paths
+  "workflow_dir": ".workflow/active/WFS-test-{session}/",
+  "summaries_dir": ".workflow/active/WFS-test-{session}/.summaries/",
+  "todo_list_path": ".workflow/active/WFS-test-{session}/TODO_LIST.md",
+
+  // Metadata (simple values, not file content)
+  "session_id": "WFS-test-{session}",
+  "current_iteration": N,
+  "max_iterations": 5
 }
 ```
 
+### Agent Loading Sequence
+
+**Agents must load files in this order:**
+
+1. **Task JSON** (`task_json_path`) - Get task definition, requirements, fix strategy
+2. **Iteration State** (`iteration_state_path`) - Understand current iteration context
+3. **Test Results** (`test_results_path`) - Analyze current test status
+4. **Test Output** (`test_output_path`) - Review detailed test execution logs
+5. **Analysis Report** (`analysis_path`, for fix tasks) - Load CLI-generated fix strategy
+6. **Fix History** (`fix_history_path`) - Review previous fix attempts to avoid repetition
+
 ## File Structure
 
 ### Test-Fix Session Files
@@ -616,77 +645,66 @@ This package is passed to agents via the Task tool's prompt context.
 
 ## Agent Prompt Template
 
-**Unified template for all agent tasks (orchestrator invokes with Task tool):**
+**Dynamic Generation**: Before agent invocation, orchestrator reads task JSON and extracts key requirements.
 
 ```bash
 Task(subagent_type="{meta.agent}",
-     prompt="**TASK EXECUTION: {task.title}**
+     prompt="Execute task: {task.title}
 
-     ## STEP 1: Load Complete Task JSON
-     **MANDATORY**: First load the complete task JSON from: {session.task_json_path}
+     {[FLOW_CONTROL]}
 
-     cat {session.task_json_path}
+     **Task Objectives** (from task JSON):
+     {task.context.requirements}
 
-     **CRITICAL**: Validate all required fields present
+     **Expected Deliverables**:
+     - For test-gen: Test files in target directories, test coverage report
+     - For test-fix: Test execution results saved to test-results.json, test-output.log
+     - For test-fix-iteration: Fixed code files, updated test results, iteration summary
 
-     ## STEP 2: Task Context (From Loaded JSON)
-     **ID**: {task.id}
-     **Type**: {task.meta.type}
-     **Agent**: {task.meta.agent}
+     **Quality Standards**:
+     - All tests must execute without errors
+     - Test results must be saved in structured JSON format
+     - All deliverables must be saved to specified paths
+     - Task status must be updated in task JSON
 
-     ## STEP 3: Execute Task Based on Type
+     **MANDATORY FIRST STEPS**:
+     1. Read complete task JSON: {session.task_json_path}
+     2. Load iteration state (if applicable): {session.iteration_state_path}
+     3. Load test context: {session.test_context_path}
 
-     ### For test-gen (IMPL-001):
-     - Generate tests based on TEST_ANALYSIS_RESULTS.md
-     - Follow test framework conventions
-     - Create test files in target_files
+     Follow complete execution guidelines in @.claude/agents/{meta.agent}.md
 
-     ### For test-fix (IMPL-002):
-     - Run test suite: {test_command}
-     - Collect results to .process/test-results.json
-     - Report results to orchestrator (do NOT analyze failures)
-     - Orchestrator will handle failure detection and iteration decisions
-     - If success: Mark complete
+     **Session Paths** (use these for all file operations):
+     - Workflow Dir: {session.workflow_dir}
+     - Task JSON: {session.task_json_path}
+     - TODO List: {session.todo_list_path}
+     - Summaries Dir: {session.summaries_dir}
+     - Test Results: {session.test_results_path}
+     - Test Output Log: {session.test_output_path}
+     - Iteration State: {session.iteration_state_path}
+     - Fix History: {session.fix_history_path}
 
-     ### For test-fix-iteration (IMPL-fix-N):
-     - Load fix strategy from context.fix_strategy (CONTENT, not path)
-     - Apply surgical fixes to identified files
-     - Return results to orchestrator
-     - Do NOT run tests independently - orchestrator manages all test execution
-     - Do NOT handle failures - orchestrator analyzes and decides next iteration
+     **Critical Rules**:
+     - For test-fix tasks: Run tests and save results, do NOT analyze failures
+     - For fix-iteration tasks: Apply fixes from task JSON, do NOT run tests independently
+     - Orchestrator manages iteration loop and failure analysis
+     - Return results to orchestrator for next-step decisions
 
-     ## STEP 4: Implementation Context (From JSON)
-     **Requirements**: {context.requirements}
-     **Fix Strategy**: {context.fix_strategy} (full content provided in task JSON)
-     **Failure Context**: {context.failure_context}
-     **Iteration History**: {context.inherited.iteration_history}
-
-     ## STEP 5: Flow Control Execution
-     If flow_control.pre_analysis exists, execute steps sequentially
-
-     ## STEP 6: Agent Completion
-     1. Execute task following implementation_approach
-     2. Update task status in JSON
-     3. Update TODO_LIST.md
-     4. Generate summary in .summaries/
-     5. **CRITICAL**: Save results for orchestrator to analyze
-
-     **Output Requirements**:
-     - test-results.json: Structured test results
-     - test-output.log: Full test output
-     - iteration-state.json: Current iteration state (if applicable)
-     - task-summary.md: Completion summary
-
-     **Return to Orchestrator**: Agent completes and returns. Orchestrator decides next action.
-     "),
-     description="Execute {task.type} task with JSON validation")
+     **Success Criteria**:
+     - Complete all task objectives as specified in task JSON
+     - Deliver all required outputs to specified paths
+     - Update task status and TODO_LIST.md
+     - Generate completion summary in .summaries/
+     ",
+     description="Executing: {task.title}")
 ```
 
-**Key Points**:
-- Agent executes single task and returns
-- Orchestrator analyzes results and decides next step
-- Fix strategy content (not path) embedded in task JSON by orchestrator
-- Agent does not manage iteration loop
+**Key Changes from Previous Version**:
+1. **Paths over Content**: Provide JSON paths for agent to read, not embedded content
+2. **MANDATORY FIRST STEPS**: Explicit requirement to load task JSON and context
+3. **Complete Session Paths**: All file paths provided for agent operations
+4. **Emphasized Deliverables**: Clear deliverable requirements per task type
+5. **Simplified Structure**: Removed type-specific instructions (agent reads from JSON)
 
 ## Error Handling & Recovery
 
diff --git a/.claude/commands/workflow/tools/conflict-resolution.md b/.claude/commands/workflow/tools/conflict-resolution.md
index 02d58e90..868586e2 100644
--- a/.claude/commands/workflow/tools/conflict-resolution.md
+++ b/.claude/commands/workflow/tools/conflict-resolution.md
@@ -12,11 +12,6 @@ examples:
 ## Purpose
 Analyzes conflicts between implementation plans and existing codebase, **including module scenario uniqueness detection**, generating multiple resolution strategies with **iterative clarification until boundaries are clear**.
 
-**Key Enhancements**:
-- **Scenario Uniqueness Detection**: Agent searches all existing modules to identify functional overlaps
-- **Iterative Clarification Loop**: Unlimited questions per conflict until scenario boundaries are uniquely defined (max 10 rounds)
-- **Dynamic Re-analysis**: Agent updates strategies based on user clarifications
-
 **Scope**: Detection and strategy generation only - NO code modification or task creation.
 
 **Trigger**: Auto-executes in `/workflow:plan` Phase 3 when `conflict_risk ≥ medium`.
diff --git a/.claude/commands/workflow/tools/context-gather.md b/.claude/commands/workflow/tools/context-gather.md
index 06efdb92..a8ed7ec4 100644
--- a/.claude/commands/workflow/tools/context-gather.md
+++ b/.claude/commands/workflow/tools/context-gather.md
@@ -57,8 +57,6 @@ Task(
   subagent_type="context-search-agent",
   description="Gather comprehensive context for plan",
   prompt=`
-You are executing as context-search-agent (.claude/agents/context-search-agent.md).
-
 ## Execution Mode
 **PLAN MODE** (Comprehensive) - Full Phase 1-3 execution
 
diff --git a/.claude/commands/workflow/tools/task-generate-agent.md b/.claude/commands/workflow/tools/task-generate-agent.md
index 485f9e45..2cd1f3ce 100644
--- a/.claude/commands/workflow/tools/task-generate-agent.md
+++ b/.claude/commands/workflow/tools/task-generate-agent.md
@@ -17,240 +17,140 @@ Autonomous task JSON and IMPL_PLAN.md generation using action-planning-agent wit
 - **Two-Phase Flow**: Discovery (context gathering) → Output (document generation)
 - **Memory-First**: Reuse loaded documents from conversation memory
 - **MCP-Enhanced**: Use MCP tools for advanced code analysis and research
-- **Pre-Selected Templates**: Command selects correct template based on `--cli-execute` flag **before** invoking agent
-- **Agent Simplicity**: Agent receives pre-selected template and focuses only on content generation
 - **Path Clarity**: All `focus_paths` prefer absolute paths (e.g., `D:\\project\\src\\module`), or clear relative paths from project root (e.g., `./src/module`)
 
 ## Execution Lifecycle
 
-### Phase 1: Discovery & Context Loading
-**⚡ Memory-First Rule**: Skip file loading if documents already in conversation memory
+### Phase 1: Context Preparation (Command Responsibility)
 
-**Agent Context Package**:
-```javascript
-{
-  "session_id": "WFS-[session-id]",
-  "execution_mode": "agent-mode" | "cli-execute-mode",  // Determined by flag
-  "task_json_template_path": "~/.claude/workflows/cli-templates/prompts/workflow/task-json-agent-mode.txt"
-                           | "~/.claude/workflows/cli-templates/prompts/workflow/task-json-cli-mode.txt",
-  // Path selected by command based on --cli-execute flag, agent reads it
-  "session_metadata": {
-    // If in memory: use cached content
-    // Else: Load from .workflow/active//{session-id}/workflow-session.json
-  },
-  "brainstorm_artifacts": {
-    // Loaded from context-package.json → brainstorm_artifacts section
-    "role_analyses": [
-      {
-        "role": "system-architect",
-        "files": [{"path": "...", "type": "primary|supplementary"}]
-      }
-    ],
-    "guidance_specification": {"path": "...", "exists": true},
-    "synthesis_output": {"path": "...", "exists": true},
-    "conflict_resolution": {"path": "...", "exists": true}  // if conflict_risk >= medium
-  },
-  "context_package_path": ".workflow/active//{session-id}/.process/context-package.json",
-  "context_package": {
-    // If in memory: use cached content
-    // Else: Load from .workflow/active//{session-id}/.process/context-package.json
-  },
-  "mcp_capabilities": {
-    "code_index": true,
-    "exa_code": true,
-    "exa_web": true
-  }
-}
+**Command prepares session paths and metadata, agent loads content autonomously.**
+
+**Session Path Structure**:
+```
+.workflow/active/WFS-{session-id}/
+├── workflow-session.json          # Session metadata
+├── .process/
+│   └── context-package.json       # Context package with artifact catalog
+├── .task/                         # Output: Task JSON files
+├── IMPL_PLAN.md                   # Output: Implementation plan
+└── TODO_LIST.md                   # Output: TODO list
 ```
 
-**Discovery Actions**:
-1. **Load Session Context** (if not in memory)
-   ```javascript
-   if (!memory.has("workflow-session.json")) {
-     Read(.workflow/active//{session-id}/workflow-session.json)
-   }
-   ```
+**Command Preparation**:
+1. **Assemble Session Paths** for agent prompt:
+   - `session_metadata_path`
+   - `context_package_path`
+   - Output directory paths
 
-2. **Load Context Package** (if not in memory)
-   ```javascript
-   if (!memory.has("context-package.json")) {
-     Read(.workflow/active//{session-id}/.process/context-package.json)
-   }
-   ```
+2. **Provide Metadata** (simple values):
+   - `session_id`
+   - `execution_mode` (agent-mode | cli-execute-mode)
+   - `mcp_capabilities` (available MCP tools)
 
-3. **Extract & Load Role Analyses** (from context-package.json)
-   ```javascript
-   // Extract role analysis paths from context package
-   const roleAnalysisPaths = contextPackage.brainstorm_artifacts.role_analyses
-     .flatMap(role => role.files.map(f => f.path));
-
-   // Load each role analysis file
-   roleAnalysisPaths.forEach(path => Read(path));
-   ```
-
-4. **Load Conflict Resolution** (from context-package.json, if exists)
-   ```javascript
-   if (contextPackage.brainstorm_artifacts.conflict_resolution?.exists) {
-     Read(contextPackage.brainstorm_artifacts.conflict_resolution.path)
-   }
-   ```
-
-5. **Code Analysis with Native Tools** (optional - enhance understanding)
-   ```bash
-   # Find relevant files for task context
-   find . -name "*auth*" -type f
-   rg "authentication|oauth" -g "*.ts"
-   ```
-
-6. **MCP External Research** (optional - gather best practices)
-   ```javascript
-   // Get external examples for implementation
-   mcp__exa__get_code_context_exa(
-     query="TypeScript JWT authentication best practices",
-     tokensNum="dynamic"
-   )
-   ```
+**Note**: Agent autonomously loads files based on context package content (dynamic, not fixed template). Brainstorming artifacts only loaded if they exist in session.
 
 ### Phase 2: Agent Execution (Document Generation)
 
-**Pre-Agent Template Selection** (Command decides path before invoking agent):
-```javascript
-// Command checks flag and selects template PATH (not content)
-const templatePath = hasCliExecuteFlag
-  ? "~/.claude/workflows/cli-templates/prompts/workflow/task-json-cli-mode.txt"
-  : "~/.claude/workflows/cli-templates/prompts/workflow/task-json-agent-mode.txt";
-```
-
 **Agent Invocation**:
 ```javascript
 Task(
   subagent_type="action-planning-agent",
   description="Generate task JSON and implementation plan",
   prompt=`
-## Execution Context
+## Task Objective
+Generate implementation plan (IMPL_PLAN.md), task JSONs, and TODO list for workflow session
 
-**Session ID**: WFS-{session-id}
-**Execution Mode**: {agent-mode | cli-execute-mode}
-**Task JSON Template Path**: {template_path}
+## MANDATORY FIRST STEPS
+1. Read session metadata: {session.session_metadata_path}
+2. Load context package: {session.context_package_path}
+3. **Dynamically load files based on context package content** (see below)
 
-## Phase 1: Discovery Results (Provided Context)
+## Dynamic Content Loading Strategy
 
-### Session Metadata
-{session_metadata_content}
+**Load files based on what exists in context package - NOT a fixed template**
 
-### Role Analyses (Enhanced by Synthesis)
-{role_analyses_content}
-- Includes requirements, design specs, enhancements, and clarifications from synthesis phase
+### Step 1: Always Load (Required)
+- **Session Metadata** → Extract user input
+  - User description: Original task requirements
+  - Project scope and boundaries
+  - Technical constraints
 
-### Artifacts Inventory
-- **Guidance Specification**: {guidance_spec_path}
-- **Role Analyses**: {role_analyses_list}
+### Step 2: Check Context Package (Conditional Loading)
 
-### Context Package
-{context_package_summary}
-- Includes conflict_risk assessment
+**If `brainstorm_artifacts` exists in context package:**
+- Load artifacts **in priority order** as listed below
+- **If `brainstorm_artifacts` does NOT exist**: Skip to Step 3
 
-### Conflict Resolution (Conditional)
-If conflict_risk was medium/high, modifications have been applied to:
-- **guidance-specification.md**: Design decisions updated to resolve conflicts
-- **Role analyses (*.md)**: Recommendations adjusted for compatibility
-- **context-package.json**: Marked as "resolved" with conflict IDs
-- NO separate CONFLICT_RESOLUTION.md file (conflicts resolved in-place)
+**Priority Loading (when artifacts exist):**
+1. **guidance-specification.md** (if `guidance_specification.exists = true`)
+   - Overall design framework - use as primary reference
 
-### MCP Analysis Results (Optional)
-**Code Structure**: {mcp_code_index_results}
-**External Research**: {mcp_exa_research_results}
+2. **Role Analyses** (if `role_analyses[]` array exists)
+   - Load ALL role analysis files listed in array
+   - Each file path: `role_analyses[i].files[j].path`
 
-## Phase 2: Document Generation Task
+3. **Synthesis Output** (if `synthesis_output.exists = true`)
+   - Integrated view with clarifications
 
-**Agent Configuration Reference**: All task generation rules, quantification requirements, quality standards, and execution details are defined in action-planning-agent.
+4. **Conflict Resolution** (if `conflict_risk` = "medium" or "high")
+   - Check `conflict_resolution.status`
+   - If "resolved": Use updated artifacts (conflicts pre-addressed)
 
-Refer to: @.claude/agents/action-planning-agent.md for:
-- Task Decomposition Standards
-- Quantification Requirements (MANDATORY)
-- 5-Field Task JSON Schema
-- IMPL_PLAN.md Structure
-- TODO_LIST.md Format
-- Execution Flow & Quality Validation
+### Step 3: Extract Project Context
+- `focus_areas`: Target directories for implementation
+- `assets`: Existing code patterns to reuse
 
-### Required Outputs Summary
+## Session Paths
+- Session Metadata: .workflow/active/{session-id}/workflow-session.json
+- Context Package: .workflow/active/{session-id}/.process/context-package.json
+- Output Task Dir: .workflow/active/{session-id}/.task/
+- Output IMPL_PLAN: .workflow/active/{session-id}/IMPL_PLAN.md
+- Output TODO_LIST: .workflow/active/{session-id}/TODO_LIST.md
 
-#### 1. Task JSON Files (.task/IMPL-*.json)
-- **Location**: `.workflow/active//{session-id}/.task/`
-- **Template**: Read from `{template_path}` (pre-selected by command based on `--cli-execute` flag)
-- **Schema**: 5-field structure (id, title, status, meta, context, flow_control) with artifacts integration
-- **Details**: See action-planning-agent.md § Task JSON Generation
+## Context Metadata
+- Session ID: {session-id}
+- Execution Mode: {agent-mode | cli-execute-mode}
+- MCP Capabilities Available: {exa_code, exa_web, code_index}
 
-#### 2. IMPL_PLAN.md
-- **Location**: `.workflow/active//{session-id}/IMPL_PLAN.md`
-- **Template**: `~/.claude/workflows/cli-templates/prompts/workflow/impl-plan-template.txt`
-- **Details**: See action-planning-agent.md § Implementation Plan Creation
+**Note**: Content loading is **dynamic** based on actual files in session, not a fixed template
 
-#### 3. TODO_LIST.md
-- **Location**: `.workflow/active//{session-id}/TODO_LIST.md`
-- **Format**: Hierarchical task list with status indicators (▸, [ ], [x]) and JSON links
-- **Details**: See action-planning-agent.md § TODO List Generation
+## Expected Deliverables
+1. **Task JSON Files** (.task/IMPL-*.json)
+   - 6-field schema (id, title, status, context_package_path, meta, context, flow_control)
+   - Quantified requirements with explicit counts
+   - Artifacts integration from context package
+   - Flow control with pre_analysis steps
 
-### Agent Execution Summary
+2. **Implementation Plan** (IMPL_PLAN.md)
+   - Context analysis and artifact references
+   - Task breakdown and execution strategy
+   - Complete structure per agent definition
 
-**Key Steps** (Detailed instructions in action-planning-agent.md):
-1. Load task JSON template from provided path
-2. Extract and decompose tasks with quantification
-3. Generate task JSON files enforcing quantification requirements
-4. Create IMPL_PLAN.md using template
-5. Generate TODO_LIST.md matching task JSONs
-6. Update session state
+3. **TODO List** (TODO_LIST.md)
+   - Hierarchical structure with status indicators (▸, [ ], [x])
+   - Links to task JSONs and summaries
+   - Matches task JSON hierarchy
 
-**Quality Gates** (Full checklist in action-planning-agent.md):
-- ✓ Quantification requirements enforced (explicit counts, measurable acceptance, exact targets)
-- ✓ Task count ≤10 (hard limit)
-- ✓ Artifact references mapped correctly
-- ✓ MCP tool integration added
-- ✓ Documents follow template structure
+## Quality Standards
+- Task count ≤12 (hard limit)
+- All requirements quantified (explicit counts and lists)
+- Acceptance criteria measurable (verification commands)
+- Artifact references mapped from context package
+- All documents follow agent-defined structure
 
-## Output
-
-Generate all three documents and report completion status:
-- Task JSON files created: N files
-- Artifacts integrated: synthesis-spec, guidance-specification, N role analyses
+## Success Criteria
+- All task JSONs valid and saved to .task/ directory
+- IMPL_PLAN.md created with complete structure
+- TODO_LIST.md generated matching task JSONs
+- Return completion status with file count
 `
 )
 ```
 
-
-### Agent Context Passing
-
-**Memory-Aware Context Assembly**:
-```javascript
-// Assemble context package for agent
-const agentContext = {
-  session_id: "WFS-[id]",
-
-  // Use memory if available, else load
-  session_metadata: memory.has("workflow-session.json")
-    ? memory.get("workflow-session.json")
-    : Read(.workflow/active/WFS-[id]/workflow-session.json),
-
-  context_package_path: ".workflow/active/WFS-[id]/.process/context-package.json",
-
-  context_package: memory.has("context-package.json")
-    ? memory.get("context-package.json")
-    : Read(".workflow/active/WFS-[id]/.process/context-package.json"),
-
-  // Extract brainstorm artifacts from context package
-  brainstorm_artifacts: extractBrainstormArtifacts(context_package),
-
-  // Load role analyses using paths from context package
-  role_analyses: brainstorm_artifacts.role_analyses
-    .flatMap(role => role.files)
-    .map(file => Read(file.path)),
-
-  // Load conflict resolution if exists (from context package)
-  conflict_resolution: brainstorm_artifacts.conflict_resolution?.exists
-    ? Read(brainstorm_artifacts.conflict_resolution.path)
-    : null,
-
-  // Optional MCP enhancements
-  mcp_analysis: executeMcpDiscovery()
-}
-```
+**Key Changes from Previous Version**:
+1. **Paths over Content**: Provide file paths for agent to read, not embedded content
+2. **MANDATORY FIRST STEPS**: Explicit requirement to load session metadata and context package
+3. **Complete Session Paths**: All file paths provided for agent operations
+4. **Emphasized Deliverables**: Clear deliverable requirements with quality standards
+5. **No Agent Self-Reference**: Removed "Refer to action-planning-agent.md" (agent knows its own definition)
+6. **No Template Paths**: Removed all template references (agent has complete schema/structure definitions)
diff --git a/.claude/skills/command-guide/SKILL.md b/.claude/skills/command-guide/SKILL.md
index 245cbb70..26a0106e 100644
--- a/.claude/skills/command-guide/SKILL.md
+++ b/.claude/skills/command-guide/SKILL.md
@@ -32,9 +32,9 @@ Comprehensive command guide for Claude DMS3 workflow system covering 75 commands
 - Creates feature-specific SKILL packages for code understanding
 - Progressive loading (2K → 30K tokens) for efficient context management
 
-### Agent Enhancements
+### Agent 
 
-- **cli-explore-agent** (New) - Specialized code exploration with Deep Scan mode (Bash + Gemini)
+- **cli-explore-agent**  - Specialized code exploration with Deep Scan mode (Bash + Gemini)
 - **cli-planning-agent** - Enhanced task generation with improved context handling
 - **ui-design-agent** - Major refactoring for better design system extraction
 
diff --git a/.claude/skills/command-guide/reference/commands/workflow/tdd-plan.md b/.claude/skills/command-guide/reference/commands/workflow/tdd-plan.md
index 182b2881..0cb4f630 100644
--- a/.claude/skills/command-guide/reference/commands/workflow/tdd-plan.md
+++ b/.claude/skills/command-guide/reference/commands/workflow/tdd-plan.md
@@ -417,52 +417,6 @@ Convert user input to TDD-structured format:
 - **Command failure**: Keep phase in_progress, report error
 - **TDD validation failure**: Report incomplete chains or wrong dependencies
 
-## TDD Workflow Enhancements
-
-### Overview
-The TDD workflow has been significantly enhanced by integrating best practices from both traditional `plan --agent` and `test-gen` workflows, creating a hybrid approach that bridges the gap between idealized TDD and real-world development complexity.
-
-### Key Improvements
-
-#### 1. Test Coverage Analysis (Phase 3)
-**Adopted from test-gen workflow**
-
-Before planning TDD tasks, the workflow now analyzes the existing codebase:
-- Detects existing test patterns and conventions
-- Identifies current test coverage
-- Discovers related components and integration points
-- Detects test framework automatically
-
-**Benefits**:
-- Context-aware TDD planning
-- Avoids duplicate test creation
-- Enables integration with existing tests
-- No longer assumes greenfield scenarios
-
-#### 2. Iterative Green Phase with Test-Fix Cycle
-**Adopted from test-gen workflow**
-
-IMPL (Green phase) tasks now include automatic test-fix cycle for resilient implementation:
-
-**Enhanced IMPL Task Flow**:
-```
-1. Write minimal implementation code
-2. Execute test suite
-3. IF tests pass → Complete task
-4. IF tests fail → Enter fix cycle:
-   a. Gemini diagnoses with bug-fix template
-   b. Apply fix (manual or Codex)
-   c. Retest
-   d. Repeat (max 3 iterations)
-5. IF max iterations → Auto-revert changes 🔄
-```
-
-**Benefits**:
-- Faster feedback within Green phase
-- Autonomous recovery from implementation errors
-- Systematic debugging with Gemini
-- Safe rollback prevents broken state
-
 #### 3. Agent-Driven Planning
 **From plan --agent workflow**