refactor: Replace CLI execution flags with semantic-driven tool selection

- Remove --cli-execute flag from plan.md, tdd-plan.md, task-generate-agent.md, task-generate-tdd.md - Remove --use-codex flag from test-gen.md, test-fix-gen.md, test-task-generate.md - Remove meta.use_codex from task JSON schema in action-planning-agent.md and cli-planning-agent.md - Add "Semantic CLI Tool Selection" section to action-planning-agent.md - Document explicit source: metadata.task_description from context-package.json - Update test-fix-agent.md execution mode documentation - Update action-plan-verify.md to remove use_codex validation - Sync SKILL reference copies via analyze_commands.py CLI tool usage now determined semantically from user's task description (e.g., "use Codex for implementation") instead of explicit flags. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
2026-02-05 01:50:27 +08:00 · 2025-11-29 15:59:01 +08:00
parent 09114f59c8
commit 132eec900c
32 changed files with 1080 additions and 1050 deletions
--- a/.claude/agents/action-planning-agent.md
+++ b/.claude/agents/action-planning-agent.md
@@ -43,7 +43,6 @@ You are a pure execution agent specialized in creating actionable implementation
  - `context_package_path`: Context package with brainstorming artifacts catalog
 - **Metadata**: Simple values
  - `session_id`: Workflow session identifier (WFS-[topic])
-  - `execution_mode`: agent-mode | cli-execute-mode
  - `mcp_capabilities`: Available MCP tools (exa_code, exa_web, code_index)

 **Legacy Support** (backward compatibility):
@@ -244,8 +243,7 @@ Generate individual `.task/IMPL-*.json` files with the following structure:
    "type": "test-gen|test-fix",
    "agent": "@code-developer|@test-fix-agent",
    "test_framework": "jest|vitest|pytest|junit|mocha",
-    "coverage_target": "80%",
-    "use_codex": true|false
+    "coverage_target": "80%"
  }
 }
 ```
@@ -253,7 +251,8 @@ Generate individual `.task/IMPL-*.json` files with the following structure:
 **Test-Specific Fields**:
 - `test_framework`: Existing test framework from project (required for test tasks)
 - `coverage_target`: Target code coverage percentage (optional)
- `use_codex`: Whether to use Codex for automated fixes in test-fix tasks (optional, default: false)
+
+**Note**: CLI tool usage for test-fix tasks is now controlled via `flow_control.implementation_approach` steps with `command` fields, not via `meta.use_codex`.

 #### Context Object

@@ -485,15 +484,31 @@ The `implementation_approach` supports **two execution modes** based on the pres
     - `bash(codex --full-auto exec '[task]' resume --last --skip-git-repo-check -s danger-full-access)` (multi-step)
     - `bash(cd [path] && gemini -p '[prompt]' --approval-mode yolo)` (write mode)

-**Mode Selection Strategy**:
- **Default to agent execution** for most tasks
- **Use CLI mode** when:
-  - User explicitly requests CLI tool (codex/gemini/qwen)
-  - Task requires multi-step autonomous reasoning beyond agent capability
-  - Complex refactoring needs specialized tool analysis
-  - Building on previous CLI execution context (use `resume --last`)
+**Semantic CLI Tool Selection**:

-**Key Principle**: The `command` field is **optional**. Agent must decide based on task complexity and user preference.
+Agent determines CLI tool usage per-step based on user semantics and task nature.
+
+**Source**: Scan `metadata.task_description` from context-package.json for CLI tool preferences.
+
+**User Semantic Triggers** (patterns to detect in task_description):
+- "use Codex/codex" → Add `command` field with Codex CLI
+- "use Gemini/gemini" → Add `command` field with Gemini CLI
+- "use Qwen/qwen" → Add `command` field with Qwen CLI
+- "CLI execution" / "automated" → Infer appropriate CLI tool
+
+**Task-Based Selection** (when no explicit user preference):
+- **Implementation/coding**: Codex preferred for autonomous development
+- **Analysis/exploration**: Gemini preferred for large context analysis
+- **Documentation**: Gemini/Qwen with write mode (`--approval-mode yolo`)
+- **Testing**: Depends on complexity - simple=agent, complex=Codex
+
+**Default Behavior**: Agent always executes the workflow. CLI commands are embedded in `implementation_approach` steps:
+- Agent orchestrates task execution
+- When step has `command` field, agent executes it via Bash
+- When step has no `command` field, agent implements directly
+- This maintains agent control while leveraging CLI tool power
+
+**Key Principle**: The `command` field is **optional**. Agent decides based on user semantics and task complexity.

 **Examples**:

--- a/.claude/agents/cli-planning-agent.md
+++ b/.claude/agents/cli-planning-agent.md
@@ -66,8 +66,7 @@ You are a specialized execution agent that bridges CLI analysis tools with task
  "task_config": {
    "agent": "@test-fix-agent",
    "type": "test-fix-iteration",
-    "max_iterations": 5,
-    "use_codex": false
+    "max_iterations": 5
  }
 }
 ```
@@ -263,7 +262,6 @@ function extractModificationPoints() {
    "analysis_report": ".process/iteration-{iteration}-analysis.md",
    "cli_output": ".process/iteration-{iteration}-cli-output.txt",
    "max_iterations": "{task_config.max_iterations}",
-    "use_codex": "{task_config.use_codex}",
    "parent_task": "{parent_task_id}",
    "created_by": "@cli-planning-agent",
    "created_at": "{timestamp}"
--- a/.claude/agents/code-developer.md
+++ b/.claude/agents/code-developer.md
@@ -24,8 +24,6 @@ You are a code execution specialist focused on implementing high-quality, produc
 - **Context-driven** - Use provided context and existing code patterns
 - **Quality over speed** - Write boring, reliable code that works

-
-
 ## Execution Process

 ### 1. Context Assessment
--- a/.claude/agents/test-fix-agent.md
+++ b/.claude/agents/test-fix-agent.md
@@ -142,9 +142,9 @@ run_test_layer "L1-unit" "$UNIT_CMD"

 ### 3. Failure Diagnosis & Fixing Loop

-**Execution Modes**:
+**Execution Modes** (determined by `flow_control.implementation_approach`):

-**A. Manual Mode (Default, meta.use_codex=false)**:
+**A. Agent Mode (Default, no `command` field in steps)**:
 ```
 WHILE tests are failing AND iterations < max_iterations:
    1. Use Gemini to diagnose failure (bug-fix template)
@@ -155,17 +155,17 @@ WHILE tests are failing AND iterations < max_iterations:
 END WHILE
 ```

-**B. Codex Mode (meta.use_codex=true)**:
+**B. CLI Mode (`command` field present in implementation_approach steps)**:
 ```
 WHILE tests are failing AND iterations < max_iterations:
    1. Use Gemini to diagnose failure (bug-fix template)
-    2. Use Codex to apply fixes automatically with resume mechanism
+    2. Execute `command` field (e.g., Codex) to apply fixes automatically
    3. Re-run test suite
    4. Verify fix doesn't break other tests
 END WHILE
 ```

-**Codex Resume in Test-Fix Cycle** (when `meta.use_codex=true`):
+**Codex Resume in Test-Fix Cycle** (when step has `command` with Codex):
 - First iteration: Start new Codex session with full context
 - Subsequent iterations: Use `resume --last` to maintain fix history and apply consistent strategies

--- a/.claude/commands/workflow/action-plan-verify.md
+++ b/.claude/commands/workflow/action-plan-verify.md
@@ -101,7 +101,7 @@ Load only minimal necessary context from each artifact:
 - Dependencies (depends_on, blocks)
 - Context (requirements, focus_paths, acceptance, artifacts)
 - Flow control (pre_analysis, implementation_approach)
- Meta (complexity, priority, use_codex)
+- Meta (complexity, priority)

 ### 3. Build Semantic Models

--- a/.claude/commands/workflow/plan.md
+++ b/.claude/commands/workflow/plan.md
@@ -1,7 +1,7 @@
 ---
 name: plan
-description: 5-phase planning workflow with action-planning-agent task generation, outputs IMPL_PLAN.md and task JSONs with optional CLI auto-execution
-argument-hint: "[--cli-execute] \"text description\"|file.md"
+description: 5-phase planning workflow with action-planning-agent task generation, outputs IMPL_PLAN.md and task JSONs
+argument-hint: "\"text description\"|file.md"
 allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 ---

@@ -69,7 +69,7 @@ Phase 3: Conflict Resolution (conditional)
      └─ conflict_risk < medium → Skip to Phase 4

 Phase 4: Task Generation
-   └─ /workflow:tools:task-generate-agent --session sessionId [--cli-execute]
+   └─ /workflow:tools:task-generate-agent --session sessionId
      └─ Output: IMPL_PLAN.md, task JSONs, TODO_LIST.md

 Return:
@@ -273,15 +273,10 @@ SlashCommand(command="/workflow:tools:conflict-resolution --session [sessionId]
 **Step 4.1: Dispatch** - Generate implementation plan and task JSONs

 ```javascript
-// Default (agent mode)
 SlashCommand(command="/workflow:tools:task-generate-agent --session [sessionId]")
-
-// With CLI execution (if --cli-execute flag present)
-SlashCommand(command="/workflow:tools:task-generate-agent --session [sessionId] --cli-execute")
 ```

-**Flag**:
- `--cli-execute`: Generate tasks with Codex execution commands
+**CLI Execution Note**: CLI tool usage is now determined semantically by action-planning-agent based on user's task description. If user specifies "use Codex/Gemini/Qwen for X", the agent embeds `command` fields in relevant `implementation_approach` steps.

 **Input**: `sessionId` from Phase 1

@@ -423,7 +418,7 @@ Phase 3: conflict-resolution [AUTO-TRIGGERED if conflict_risk ≥ medium]
    ↓ Output: Modified brainstorm artifacts (NO report file)
    ↓ Skip if conflict_risk is none/low → proceed directly to Phase 4
    ↓
-Phase 4: task-generate-agent --session sessionId [--cli-execute]
+Phase 4: task-generate-agent --session sessionId
    ↓ Input: sessionId + resolved brainstorm artifacts + session memory
    ↓ Output: IMPL_PLAN.md, task JSONs, TODO_LIST.md
    ↓
@@ -504,9 +499,7 @@ Return summary to user
 - **If conflict_risk ≥ medium**: Launch Phase 3 conflict-resolution with sessionId and contextPath
 - Wait for Phase 3 to finish executing (if executed), verify CONFLICT_RESOLUTION.md created
 - **If conflict_risk is none/low**: Skip Phase 3, proceed directly to Phase 4
- **Build Phase 4 command**:
-  - Base command: `/workflow:tools:task-generate-agent --session [sessionId]`
-  - Add `--cli-execute` if flag present
+- **Build Phase 4 command**: `/workflow:tools:task-generate-agent --session [sessionId]`
 - Pass session ID to Phase 4 command
 - Verify all Phase 4 outputs
 - Update TodoWrite after each phase (dynamically adjust for Phase 3 presence)
--- a/.claude/commands/workflow/tdd-plan.md
+++ b/.claude/commands/workflow/tdd-plan.md
@@ -1,7 +1,7 @@
 ---
 name: tdd-plan
 description: TDD workflow planning with Red-Green-Refactor task chain generation, test-first development structure, and cycle tracking
-argument-hint: "[--cli-execute] \"feature description\"|file.md"
+argument-hint: "\"feature description\"|file.md"
 allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 ---

@@ -11,9 +11,7 @@ allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)

 **This command is a pure orchestrator**: Dispatches 6 slash commands in sequence, parse outputs, pass context, and ensure complete TDD workflow creation with Red-Green-Refactor task generation.

-**Execution Modes**:
- **Agent Mode** (default): Use `/workflow:tools:task-generate-tdd` (autonomous agent-driven)
- **CLI Mode** (`--cli-execute`): Use `/workflow:tools:task-generate-tdd --cli-execute` (Gemini/Qwen)
+**CLI Tool Selection**: CLI tool usage is determined semantically from user's task description. Include "use Codex/Gemini/Qwen" in your request for CLI execution.

 **Task Attachment Model**:
 - SlashCommand dispatch **expands workflow** by attaching sub-tasks to current TodoWrite
@@ -235,13 +233,11 @@ SlashCommand(command="/workflow:tools:conflict-resolution --session [sessionId]
 **Step 5.1: Dispatch** - TDD task generation via action-planning-agent

 ```javascript
-// Agent Mode (default)
 SlashCommand(command="/workflow:tools:task-generate-tdd --session [sessionId]")
-
-// CLI Mode (--cli-execute flag)
-SlashCommand(command="/workflow:tools:task-generate-tdd --session [sessionId] --cli-execute")
 ```

+**Note**: CLI tool usage is determined semantically from user's task description.
+
 **Parse**: Extract feature count, task count (not chain count - tasks now contain internal TDD cycles)

 **Validate**:
@@ -454,8 +450,7 @@ Convert user input to TDD-structured format:
 - `/workflow:tools:test-context-gather` - Phase 3: Analyze existing test patterns and coverage
 - `/workflow:tools:conflict-resolution` - Phase 4: Detect and resolve conflicts (auto-triggered if conflict_risk ≥ medium)
 - `/compact` - Phase 4: Memory optimization (if context approaching limits)
- `/workflow:tools:task-generate-tdd` - Phase 5: Generate TDD tasks with agent-driven approach (default, autonomous)
- `/workflow:tools:task-generate-tdd --cli-execute` - Phase 5: Generate TDD tasks with CLI tools (Gemini/Qwen, when `--cli-execute` flag used)
+- `/workflow:tools:task-generate-tdd` - Phase 5: Generate TDD tasks (CLI tool usage determined semantically)

 **Follow-up Commands**:
 - `/workflow:action-plan-verify` - Recommended: Verify TDD plan quality and structure before execution
--- a/.claude/commands/workflow/test-fix-gen.md
+++ b/.claude/commands/workflow/test-fix-gen.md
@@ -1,7 +1,7 @@
 ---
 name: test-fix-gen
 description: Create test-fix workflow session from session ID, description, or file path with test strategy generation and task planning
-argument-hint: "[--use-codex] [--cli-execute] (source-session-id | \"feature description\" | /path/to/file.md)"
+argument-hint: "(source-session-id | \"feature description\" | /path/to/file.md)"
 allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 ---

@@ -43,7 +43,7 @@ fi
 - **Session Isolation**: Creates independent `WFS-test-[slug]` session
 - **Context-First**: Gathers implementation context via appropriate method
 - **Format Reuse**: Creates standard `IMPL-*.json` tasks with `meta.type: "test-fix"`
- **Manual First**: Default to manual fixes, use `--use-codex` for automation
+- **Semantic CLI Selection**: CLI tool usage determined from user's task description
 - **Automatic Detection**: Input pattern determines execution mode

 ### Coordinator Role
@@ -79,16 +79,14 @@ This command is a **pure planning coordinator**:

 ```bash
 # Basic syntax
-/workflow:test-fix-gen [FLAGS] <INPUT>
-
-# Flags (optional)
--use-codex        # Enable Codex automated fixes in IMPL-002
--cli-execute      # Enable CLI execution in IMPL-001
+/workflow:test-fix-gen <INPUT>

 # Input
 <INPUT>            # Session ID, description, or file path
 ```

+**Note**: CLI tool usage is determined semantically from the task description. To request CLI execution, include it in your description (e.g., "use Codex for automated fixes").
+
 ### Usage Examples

 #### Session Mode
@@ -96,11 +94,8 @@ This command is a **pure planning coordinator**:
 # Test validation for completed implementation
 /workflow:test-fix-gen WFS-user-auth-v2

-# With automated fixes
-/workflow:test-fix-gen --use-codex WFS-api-endpoints
-
-# With CLI execution
-/workflow:test-fix-gen --cli-execute --use-codex WFS-payment-flow
+# With semantic CLI request
+/workflow:test-fix-gen WFS-api-endpoints  # Add "use Codex" in description for automated fixes
 ```

 #### Prompt Mode - Text Description
@@ -108,17 +103,14 @@ This command is a **pure planning coordinator**:
 # Generate tests from feature description
 /workflow:test-fix-gen "Test the user authentication API endpoints in src/auth/api.ts"

-# With automated fixes
-/workflow:test-fix-gen --use-codex "Test user registration and login flows"
+# With CLI execution (semantic)
+/workflow:test-fix-gen "Test user registration and login flows, use Codex for automated fixes"
 ```

 #### Prompt Mode - File Reference
 ```bash
 # Generate tests from requirements file
 /workflow:test-fix-gen ./docs/api-requirements.md
-
-# With flags
-/workflow:test-fix-gen --use-codex --cli-execute ./specs/feature.md
 ```

 ### Mode Comparison
@@ -143,7 +135,7 @@ This command is a **pure planning coordinator**:
 5. **Complete All Phases**: Do not return until Phase 5 completes
 6. **Track Progress**: Update TodoWrite dynamically with task attachment/collapse pattern
 7. **Automatic Detection**: Mode auto-detected from input pattern
-8. **Parse Flags**: Extract `--use-codex` and `--cli-execute` flags for Phase 4
+8. **Semantic CLI Detection**: CLI tool usage determined from user's task description for Phase 4
 9. **Task Attachment Model**: SlashCommand dispatch **attaches** sub-tasks to current workflow. Orchestrator **executes** these attached tasks itself, then **collapses** them after completion
 10. **⚠️ CRITICAL: DO NOT STOP**: Continuous multi-phase workflow. After executing all attached tasks, immediately collapse them and execute next phase

@@ -283,13 +275,13 @@ For each targeted file/function, Gemini MUST generate:
 **Step 4.1: Dispatch** - Generate test task JSONs

 ```javascript
-SlashCommand(command="/workflow:tools:test-task-generate [--use-codex] [--cli-execute] --session [testSessionId]")
+SlashCommand(command="/workflow:tools:test-task-generate --session [testSessionId]")
 ```

 **Input**:
 - `testSessionId` from Phase 1
- `--use-codex` flag (if present) - Controls IMPL-002 fix mode
- `--cli-execute` flag (if present) - Controls IMPL-001 generation mode
+
+**Note**: CLI tool usage is determined semantically from user's task description.

 **Expected Behavior**:
 - Parse TEST_ANALYSIS_RESULTS.md from Phase 3 (multi-layered test plan)
@@ -422,7 +414,7 @@ CRITICAL - Next Steps:
 - **Phase 2**: Mode-specific context gathering (session summaries vs codebase analysis)
 - **Phase 3**: Multi-layered test requirements analysis (L0: Static, L1: Unit, L2: Integration, L3: E2E)
 - **Phase 4**: Multi-task generation with quality gate (IMPL-001, IMPL-001.5-review, IMPL-002)
- **Fix Mode Configuration**: `--use-codex` flag controls IMPL-002 fix mode (manual vs automated)
+- **Fix Mode Configuration**: CLI tool usage determined semantically from user's task description


 ---
@@ -521,16 +513,15 @@ If quality gate fails:
 - Task ID: `IMPL-002`
 - `meta.type: "test-fix"`
 - `meta.agent: "@test-fix-agent"`
- `meta.use_codex: true|false` (based on `--use-codex` flag)
 - `context.depends_on: ["IMPL-001"]`
 - `context.requirements`: Execute and fix tests

 **Test-Fix Cycle Specification**:
 **Note**: This specification describes what test-cycle-execute orchestrator will do. The agent only executes single tasks.
- **Cycle Pattern** (orchestrator-managed): test → gemini_diagnose → manual_fix (or codex) → retest
+- **Cycle Pattern** (orchestrator-managed): test → gemini_diagnose → fix (agent or CLI) → retest
 - **Tools Configuration** (orchestrator-controlled):
  - Gemini for analysis with bug-fix template → surgical fix suggestions
-  - Manual fix application (default) OR Codex if `--use-codex` flag (resume mechanism)
+  - Agent fix application (default) OR CLI if `command` field present in implementation_approach
 - **Exit Conditions** (orchestrator-enforced):
  - Success: All tests pass
  - Failure: Max iterations reached (5)
@@ -674,8 +665,7 @@ Key Points:
 4. **Mode Selection**:
   - Use **Session Mode** for completed workflow validation
   - Use **Prompt Mode** for ad-hoc test generation
-   - Use `--use-codex` for autonomous fix application
-   - Use `--cli-execute` for enhanced generation capabilities
+   - Include "use Codex" in description for autonomous fix application

 ## Related Commands

@@ -688,9 +678,7 @@ Key Points:
 - `/workflow:tools:test-context-gather` - Phase 2 (Session Mode): Gather source session context
 - `/workflow:tools:context-gather` - Phase 2 (Prompt Mode): Analyze codebase directly
 - `/workflow:tools:test-concept-enhanced` - Phase 3: Generate test requirements using Gemini
- `/workflow:tools:test-task-generate` - Phase 4: Generate test task JSONs using action-planning-agent (autonomous, default)
- `/workflow:tools:test-task-generate --use-codex` - Phase 4: With automated Codex fixes for IMPL-002 (when `--use-codex` flag used)
- `/workflow:tools:test-task-generate --cli-execute` - Phase 4: With CLI execution mode for IMPL-001 test generation (when `--cli-execute` flag used)
+- `/workflow:tools:test-task-generate` - Phase 4: Generate test task JSONs (CLI tool usage determined semantically)

 **Follow-up Commands**:
 - `/workflow:status` - Review generated test tasks
--- a/.claude/commands/workflow/test-gen.md
+++ b/.claude/commands/workflow/test-gen.md
@@ -1,7 +1,7 @@
 ---
 name: test-gen
 description: Create independent test-fix workflow session from completed implementation session, analyzes code to generate test tasks
-argument-hint: "[--use-codex] [--cli-execute] source-session-id"
+argument-hint: "source-session-id"
 allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 ---

@@ -16,7 +16,7 @@ allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 - **Context-First**: Prioritizes gathering code changes and summaries from source session
 - **Format Reuse**: Creates standard `IMPL-*.json` task, using `meta.type: "test-fix"` for agent assignment
 - **Parameter Simplification**: Tools auto-detect test session type via metadata, no manual cross-session parameters needed
- **Manual First**: Default to manual fixes, use `--use-codex` flag for automated Codex fix application
+- **Semantic CLI Selection**: CLI tool usage is determined by user's task description (e.g., "use Codex for fixes")

 **Task Attachment Model**:
 - SlashCommand dispatch **expands workflow** by attaching sub-tasks to current TodoWrite
@@ -48,7 +48,7 @@ allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 5. **Complete All Phases**: Do not return to user until Phase 5 completes (summary returned)
 6. **Track Progress**: Update TodoWrite dynamically with task attachment/collapse pattern
 7. **Automatic Detection**: context-gather auto-detects test session and gathers source session context
-8. **Parse --use-codex Flag**: Extract flag from arguments and pass to Phase 4 (test-task-generate)
+8. **Semantic CLI Selection**: CLI tool usage determined from user's task description, passed to Phase 4
 9. **Command Boundary**: This command ends at Phase 5 summary. Test execution is NOT part of this command.
 10. **Task Attachment Model**: SlashCommand dispatch **attaches** sub-tasks to current workflow. Orchestrator **executes** these attached tasks itself, then **collapses** them after completion
 11. **⚠️ CRITICAL: DO NOT STOP**: Continuous multi-phase workflow. After executing all attached tasks, immediately collapse them and execute next phase
@@ -224,13 +224,13 @@ SlashCommand(command="/workflow:tools:test-concept-enhanced --session [testSessi
 **Step 4.1: Dispatch** - Generate test task JSON files and planning documents

 ```javascript
-SlashCommand(command="/workflow:tools:test-task-generate [--use-codex] [--cli-execute] --session [testSessionId]")
+SlashCommand(command="/workflow:tools:test-task-generate --session [testSessionId]")
 ```

 **Input**:
 - `testSessionId` from Phase 1
- `--use-codex` flag (if present in original command) - Controls IMPL-002 fix mode
- `--cli-execute` flag (if present in original command) - Controls IMPL-001 generation mode
+
+**Note**: CLI tool usage for fixes is determined semantically from user's task description (e.g., "use Codex for automated fixes").

 **Expected Behavior**:
 - Parse TEST_ANALYSIS_RESULTS.md from Phase 3
@@ -260,16 +260,15 @@ SlashCommand(command="/workflow:tools:test-task-generate [--use-codex] [--cli-ex
 - Task ID: `IMPL-002`
 - `meta.type: "test-fix"`
 - `meta.agent: "@test-fix-agent"`
- `meta.use_codex: true|false` (based on --use-codex flag)
 - `context.depends_on: ["IMPL-001"]`
 - `context.requirements`: Execute and fix tests
 - `flow_control.implementation_approach.test_fix_cycle`: Complete cycle specification
-  - **Cycle pattern**: test → gemini_diagnose → manual_fix (or codex if --use-codex) → retest
-  - **Tools configuration**: Gemini for analysis with bug-fix template, manual or Codex for fixes
+  - **Cycle pattern**: test → gemini_diagnose → fix (agent or CLI based on `command` field) → retest
+  - **Tools configuration**: Gemini for analysis with bug-fix template, agent or CLI for fixes
  - **Exit conditions**: Success (all pass) or failure (max iterations)
 - `flow_control.implementation_approach.modification_points`: 3-phase execution flow
  - Phase 1: Initial test execution
-  - Phase 2: Iterative Gemini diagnosis + manual/Codex fixes (based on flag)
+  - Phase 2: Iterative Gemini diagnosis + fixes (agent or CLI based on step's `command` field)
  - Phase 3: Final validation and certification

 <!-- TodoWrite: When test-task-generate dispatched, INSERT 3 test-task-generate tasks -->
@@ -327,7 +326,7 @@ Artifacts Created:

 Test Framework: [detected framework]
 Test Files to Generate: [count]
-Fix Mode: [Manual|Codex Automated] (based on --use-codex flag)
+Fix Mode: [Agent|CLI] (based on `command` field in implementation_approach steps)

 Review Generated Artifacts:
 - Test plan: .workflow/[testSessionId]/IMPL_PLAN.md
@@ -373,7 +372,7 @@ Ready for execution. Use appropriate workflow commands to proceed.
 - **Phase 2**: Cross-session context gathering from source implementation session
 - **Phase 3**: Test requirements analysis with Gemini for generation strategy
 - **Phase 4**: Dual-task generation (IMPL-001 for test generation, IMPL-002 for test execution)
- **Fix Mode Configuration**: `--use-codex` flag controls IMPL-002 fix mode (manual vs automated)
+- **Fix Mode Configuration**: CLI tool usage determined semantically from user's task description



@@ -444,7 +443,7 @@ Generates two task definition files:
  - Agent: @test-fix-agent
  - Dependency: IMPL-001 must complete first
  - Max iterations: 5
-  - Fix mode: Manual or Codex (based on --use-codex flag)
+  - Fix mode: Agent or CLI (based on `command` field in implementation_approach)

 See `/workflow:tools:test-task-generate` for complete task JSON schemas.

@@ -481,11 +480,10 @@ Created in `.workflow/active/WFS-test-[session]/`:
 **IMPL-002.json Structure**:
 - `meta.type: "test-fix"`
 - `meta.agent: "@test-fix-agent"`
- `meta.use_codex`: true/false (based on --use-codex flag)
 - `context.depends_on: ["IMPL-001"]`
 - `flow_control.implementation_approach.test_fix_cycle`: Complete cycle specification
  - Gemini diagnosis template
-  - Fix application mode (manual/codex)
+  - Fix application mode (agent or CLI based on `command` field)
  - Max iterations: 5
 - `flow_control.implementation_approach.modification_points`: 3-phase flow

@@ -503,13 +501,11 @@ See `/workflow:tools:test-task-generate` for complete JSON schemas.
 **Prerequisite Commands**:
 - `/workflow:plan` or `/workflow:execute` - Complete implementation session that needs test validation

-**Dispatched by This Command** (5 phases):
+**Dispatched by This Command** (4 phases):
 - `/workflow:session:start` - Phase 1: Create independent test workflow session
 - `/workflow:tools:test-context-gather` - Phase 2: Analyze test coverage and gather source session context
 - `/workflow:tools:test-concept-enhanced` - Phase 3: Generate test requirements and strategy using Gemini
- `/workflow:tools:test-task-generate` - Phase 4: Generate test task JSONs using action-planning-agent (autonomous, default)
- `/workflow:tools:test-task-generate --use-codex` - Phase 4: With automated Codex fixes for IMPL-002 (when `--use-codex` flag used)
- `/workflow:tools:test-task-generate --cli-execute` - Phase 4: With CLI execution mode for IMPL-001 test generation (when `--cli-execute` flag used)
+- `/workflow:tools:test-task-generate` - Phase 4: Generate test task JSONs (CLI tool usage determined semantically)

 **Follow-up Commands**:
 - `/workflow:status` - Review generated test tasks
--- a/.claude/commands/workflow/tools/task-generate-agent.md
+++ b/.claude/commands/workflow/tools/task-generate-agent.md
@@ -1,10 +1,9 @@
 ---
 name: task-generate-agent
 description: Generate implementation plan documents (IMPL_PLAN.md, task JSONs, TODO_LIST.md) using action-planning-agent - produces planning artifacts, does NOT execute code implementation
-argument-hint: "--session WFS-session-id [--cli-execute]"
+argument-hint: "--session WFS-session-id"
 examples:
  - /workflow:tools:task-generate-agent --session WFS-auth
-  - /workflow:tools:task-generate-agent --session WFS-auth --cli-execute
 ---

 # Generate Implementation Plan Command
@@ -26,7 +25,7 @@ Generate implementation planning documents (IMPL_PLAN.md, task JSONs, TODO_LIST.

 ```
 Input Parsing:
-   ├─ Parse flags: --session, --cli-execute
+   ├─ Parse flags: --session
   └─ Validation: session_id REQUIRED

 Phase 1: Context Preparation (Command)
@@ -65,9 +64,10 @@ Phase 2: Planning Document Generation (Agent)

 2. **Provide Metadata** (simple values):
   - `session_id`
-   - `execution_mode` (agent-mode | cli-execute-mode)
   - `mcp_capabilities` (available MCP tools)

+**Note**: CLI tool usage is now determined semantically by action-planning-agent based on user's task description, not by flags.
+
 ### Phase 2: Planning Document Generation (Agent Responsibility)

 **Purpose**: Generate IMPL_PLAN.md, task JSONs, and TODO_LIST.md - planning documents only, NOT code implementation.
@@ -97,9 +97,13 @@ Output:

 ## CONTEXT METADATA
 Session ID: {session-id}
-Planning Mode: {agent-mode | cli-execute-mode}
 MCP Capabilities: {exa_code, exa_web, code_index}

+## CLI TOOL SELECTION
+Determine CLI tool usage per-step based on user's task description:
+- If user specifies "use Codex/Gemini/Qwen for X" → Add command field to relevant steps
+- Default: Agent execution (no command field) unless user explicitly requests CLI
+
 ## EXPLORATION CONTEXT (from context-package.exploration_results)
 - Load exploration_results from context-package.json
 - Use aggregated_insights.critical_files for focus_paths generation
--- a/.claude/commands/workflow/tools/task-generate-tdd.md
+++ b/.claude/commands/workflow/tools/task-generate-tdd.md
@@ -1,24 +1,23 @@
 ---
 name: task-generate-tdd
 description: Autonomous TDD task generation using action-planning-agent with Red-Green-Refactor cycles, test-first structure, and cycle validation
-argument-hint: "--session WFS-session-id [--cli-execute]"
+argument-hint: "--session WFS-session-id"
 examples:
  - /workflow:tools:task-generate-tdd --session WFS-auth
-  - /workflow:tools:task-generate-tdd --session WFS-auth --cli-execute
 ---

 # Autonomous TDD Task Generation Command

 ## Overview
-Autonomous TDD task JSON and IMPL_PLAN.md generation using action-planning-agent with two-phase execution: discovery and document generation. Supports both agent-driven execution (default) and CLI tool execution modes. Generates complete Red-Green-Refactor cycles contained within each task.
+Autonomous TDD task JSON and IMPL_PLAN.md generation using action-planning-agent with two-phase execution: discovery and document generation. Generates complete Red-Green-Refactor cycles contained within each task.

 ## Core Philosophy
 - **Agent-Driven**: Delegate execution to action-planning-agent for autonomous operation
 - **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 TDD template based on `--cli-execute` flag **before** invoking agent
- **Agent Simplicity**: Agent receives pre-selected template and focuses only on content generation
+- **Semantic CLI Selection**: CLI tool usage determined from user's task description, not flags
+- **Agent Simplicity**: Agent generates content with semantic CLI detection
 - **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`)
 - **TDD-First**: Every feature starts with a failing test (Red phase)
 - **Feature-Complete Tasks**: Each task contains complete Red-Green-Refactor cycle
@@ -57,7 +56,7 @@ Autonomous TDD task JSON and IMPL_PLAN.md generation using action-planning-agent

 ```
 Input Parsing:
-   ├─ Parse flags: --session, --cli-execute
+   ├─ Parse flags: --session
   └─ Validation: session_id REQUIRED

 Phase 1: Discovery & Context Loading (Memory-First)
@@ -69,7 +68,7 @@ Phase 1: Discovery & Context Loading (Memory-First)
   └─ Optional: MCP external research

 Phase 2: Agent Execution (Document Generation)
-   ├─ Pre-agent template selection (agent-mode OR cli-execute-mode)
+   ├─ Pre-agent template selection (semantic CLI detection)
   ├─ Invoke action-planning-agent
   ├─ Generate TDD Task JSON Files (.task/IMPL-*.json)
   │  └─ Each task: complete Red-Green-Refactor cycle internally
@@ -86,11 +85,8 @@ Phase 2: Agent Execution (Document Generation)
 ```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
  "workflow_type": "tdd",
+  // Note: CLI tool usage is determined semantically by action-planning-agent based on user's task description
  "session_metadata": {
    // If in memory: use cached content
    // Else: Load from .workflow/active//{session-id}/workflow-session.json
@@ -199,8 +195,7 @@ Task(

 **Session ID**: WFS-{session-id}
 **Workflow Type**: TDD
-**Execution Mode**: {agent-mode | cli-execute-mode}
-**Task JSON Template Path**: {template_path}
+**Note**: CLI tool usage is determined semantically from user's task description

 ## Phase 1: Discovery Results (Provided Context)

@@ -265,16 +260,15 @@ Refer to: @.claude/agents/action-planning-agent.md for:

 ##### 1. TDD 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 with TDD-specific metadata
  - `meta.tdd_workflow`: true (REQUIRED)
  - `meta.max_iterations`: 3 (Green phase test-fix cycle limit)
-  - `meta.use_codex`: false (manual fixes by default)
  - `context.tdd_cycles`: Array with quantified test cases and coverage
  - `flow_control.implementation_approach`: Exactly 3 steps with `tdd_phase` field
    1. Red Phase (`tdd_phase: "red"`): Write failing tests
    2. Green Phase (`tdd_phase: "green"`): Implement to pass tests
    3. Refactor Phase (`tdd_phase: "refactor"`): Improve code quality
+  - CLI tool usage determined semantically (add `command` field when user requests CLI execution)
 - **Details**: See action-planning-agent.md § TDD Task JSON Generation

 ##### 2. IMPL_PLAN.md (TDD Variant)
@@ -475,16 +469,14 @@ This section provides quick reference for TDD task JSON structure. For complete

 **Basic Usage**:
 ```bash
-# Agent mode (default, autonomous execution)
+# Standard execution
 /workflow:tools:task-generate-tdd --session WFS-auth

-# CLI tool mode (use Gemini/Qwen for generation)
-/workflow:tools:task-generate-tdd --session WFS-auth --cli-execute
+# With semantic CLI request (include in task description)
+# e.g., "Generate TDD tasks for auth module, use Codex for implementation"
 ```

-**Execution Modes**:
- **Agent mode** (default): Uses `action-planning-agent` with agent-mode task template
- **CLI mode** (`--cli-execute`): Uses Gemini/Qwen with cli-mode task template
+**CLI Tool Selection**: Determined semantically from user's task description. Include "use Codex/Gemini/Qwen" in your request for CLI execution.

 **Output**:
 - TDD task JSON files in `.task/` directory (IMPL-N.json format)
@@ -513,7 +505,7 @@ IMPL (Green phase) tasks include automatic test-fix cycle:
 3. **Success Path**: Tests pass → Complete task
 4. **Failure Path**: Tests fail → Enter iterative fix cycle:
   - **Gemini Diagnosis**: Analyze failures with bug-fix template
-   - **Fix Application**: Manual (default) or Codex (if meta.use_codex=true)
+   - **Fix Application**: Agent (default) or CLI (if `command` field present)
   - **Retest**: Verify fix resolves failures
   - **Repeat**: Up to max_iterations (default: 3)
 5. **Safety Net**: Auto-revert all changes if max iterations reached
@@ -522,5 +514,5 @@ IMPL (Green phase) tasks include automatic test-fix cycle:

 ## Configuration Options
 - **meta.max_iterations**: Number of fix attempts (default: 3 for TDD, 5 for test-gen)
- **meta.use_codex**: Enable Codex automated fixes (default: false, manual)
+- **CLI tool usage**: Determined semantically from user's task description via `command` field in implementation_approach

--- a/.claude/commands/workflow/tools/test-task-generate.md
+++ b/.claude/commands/workflow/tools/test-task-generate.md
@@ -1,11 +1,9 @@
 ---
 name: test-task-generate
 description: Generate test planning documents (IMPL_PLAN.md, test task JSONs, TODO_LIST.md) using action-planning-agent - produces test planning artifacts, does NOT execute tests
-argument-hint: "[--use-codex] [--cli-execute] --session WFS-test-session-id"
+argument-hint: "--session WFS-test-session-id"
 examples:
  - /workflow:tools:test-task-generate --session WFS-test-auth
-  - /workflow:tools:test-task-generate --use-codex --session WFS-test-auth
-  - /workflow:tools:test-task-generate --cli-execute --session WFS-test-auth
 ---

 # Generate Test Planning Documents Command
@@ -26,17 +24,17 @@ Generate test planning documents (IMPL_PLAN.md, test task JSONs, TODO_LIST.md) u

 ### Test Generation (IMPL-001)
 - **Agent Mode** (default): @code-developer generates tests within agent context
- **CLI Execute Mode** (`--cli-execute`): Use Codex CLI for autonomous test generation
+- **CLI Mode**: Use CLI tools when `command` field present in implementation_approach (determined semantically)

 ### Test Execution & Fix (IMPL-002+)
- **Manual Mode** (default): Gemini diagnosis → user applies fixes
- **Codex Mode** (`--use-codex`): Gemini diagnosis → Codex applies fixes with resume mechanism
+- **Agent Mode** (default): Gemini diagnosis → agent applies fixes
+- **CLI Mode**: Gemini diagnosis → CLI applies fixes (when `command` field present in implementation_approach)

 ## Execution Process

 ```
 Input Parsing:
-   ├─ Parse flags: --session, --use-codex, --cli-execute
+   ├─ Parse flags: --session
   └─ Validation: session_id REQUIRED

 Phase 1: Context Preparation (Command)
@@ -44,7 +42,7 @@ Phase 1: Context Preparation (Command)
   │  ├─ session_metadata_path
   │  ├─ test_analysis_results_path (REQUIRED)
   │  └─ test_context_package_path
-   └─ Provide metadata (session_id, execution_mode, use_codex, source_session_id)
+   └─ Provide metadata (session_id, source_session_id)

 Phase 2: Test Document Generation (Agent)
   ├─ Load TEST_ANALYSIS_RESULTS.md as primary requirements source
@@ -83,11 +81,11 @@ Phase 2: Test Document Generation (Agent)

 2. **Provide Metadata** (simple values):
   - `session_id`
-   - `execution_mode` (agent-mode | cli-execute-mode)
-   - `use_codex` flag (true | false)
   - `source_session_id` (if exists)
   - `mcp_capabilities` (available MCP tools)

+**Note**: CLI tool usage is now determined semantically from user's task description, not by flags.
+
 ### Phase 2: Test Document Generation (Agent Responsibility)

 **Purpose**: Generate test-specific IMPL_PLAN.md, task JSONs, and TODO_LIST.md - planning documents only, NOT test execution.
@@ -134,11 +132,14 @@ Output:
 ## CONTEXT METADATA
 Session ID: {test-session-id}
 Workflow Type: test_session
-Planning Mode: {agent-mode | cli-execute-mode}
-Use Codex: {true | false}
 Source Session: {source-session-id} (if exists)
 MCP Capabilities: {exa_code, exa_web, code_index}

+## CLI TOOL SELECTION
+Determine CLI tool usage per-step based on user's task description:
+- If user specifies "use Codex/Gemini/Qwen for X" → Add command field to relevant steps
+- Default: Agent execution (no command field) unless user explicitly requests CLI
+
 ## TEST-SPECIFIC REQUIREMENTS SUMMARY
 (Detailed specifications in your agent definition)

@@ -149,25 +150,26 @@ MCP Capabilities: {exa_code, exa_web, code_index}
 Task Configuration:
  IMPL-001 (Test Generation):
    - meta.type: "test-gen"
-    - meta.agent: "@code-developer" (agent-mode) OR CLI execution (cli-execute-mode)
+    - meta.agent: "@code-developer"
    - meta.test_framework: Specify existing framework (e.g., "jest", "vitest", "pytest")
    - flow_control: Test generation strategy from TEST_ANALYSIS_RESULTS.md
+    - CLI execution: Add `command` field when user requests (determined semantically)

  IMPL-002+ (Test Execution & Fix):
    - meta.type: "test-fix"
    - meta.agent: "@test-fix-agent"
-    - meta.use_codex: true/false (based on flag)
    - flow_control: Test-fix cycle with iteration limits and diagnosis configuration
+    - CLI execution: Add `command` field when user requests (determined semantically)

 ### Test-Fix Cycle Specification (IMPL-002+)
 Required flow_control fields:
  - max_iterations: 5
  - diagnosis_tool: "gemini"
  - diagnosis_template: "~/.claude/workflows/cli-templates/prompts/analysis/01-diagnose-bug-root-cause.txt"
-  - fix_mode: "manual" OR "codex" (based on use_codex flag)
  - cycle_pattern: "test → gemini_diagnose → fix → retest"
  - exit_conditions: ["all_tests_pass", "max_iterations_reached"]
  - auto_revert_on_failure: true
+  - CLI fix: Add `command` field when user specifies CLI tool usage

 ### Automation Framework Configuration
 Select automation tools based on test requirements from TEST_ANALYSIS_RESULTS.md:
@@ -191,8 +193,9 @@ PRIMARY requirements source - extract and map to task JSONs:
 ## EXPECTED DELIVERABLES
 1. Test Task JSON Files (.task/IMPL-*.json)
   - 6-field schema with quantified requirements from TEST_ANALYSIS_RESULTS.md
-   - Test-specific metadata: type, agent, use_codex, test_framework, coverage_target
+   - Test-specific metadata: type, agent, test_framework, coverage_target
   - flow_control includes: reusable_test_tools, test_commands (from project config)
+   - CLI execution via `command` field when user requests (determined semantically)
   - Artifact references from test-context-package.json
   - Absolute paths in context.files_to_test

@@ -213,9 +216,9 @@ Hard Constraints:
  - All requirements quantified from TEST_ANALYSIS_RESULTS.md
  - Test framework matches existing project framework
  - flow_control includes reusable_test_tools and test_commands from project
-  - use_codex flag correctly set in IMPL-002+ tasks
  - Absolute paths for all focus_paths
  - Acceptance criteria include verification commands
+  - CLI `command` field added only when user explicitly requests CLI tool usage

 ## SUCCESS CRITERIA
 - All test planning documents generated successfully
@@ -233,21 +236,18 @@ Hard Constraints:

 ### Usage Examples
 ```bash
-# Agent mode (default)
+# Standard execution
 /workflow:tools:test-task-generate --session WFS-test-auth

-# With automated Codex fixes
-/workflow:tools:test-task-generate --use-codex --session WFS-test-auth
-
-# CLI execution mode for test generation
-/workflow:tools:test-task-generate --cli-execute --session WFS-test-auth
+# With semantic CLI request (include in task description)
+# e.g., "Generate tests, use Codex for implementation and fixes"
 ```

-### Flag Behavior
- **No flags**: `meta.use_codex=false` (manual fixes), agent-mode test generation
- **--use-codex**: `meta.use_codex=true` (Codex automated fixes in IMPL-002+)
- **--cli-execute**: CLI tool execution mode for IMPL-001 test generation
- **Both flags**: CLI generation + automated Codex fixes
+### CLI Tool Selection
+CLI tool usage is determined semantically from user's task description:
+- Include "use Codex" for automated fixes
+- Include "use Gemini" for analysis
+- Default: Agent execution (no `command` field)

 ### Output
 - Test task JSON files in `.task/` directory (minimum 2)
--- a/.claude/skills/command-guide/index/all-commands.json
+++ b/.claude/skills/command-guide/index/all-commands.json
@@ -409,8 +409,8 @@
  {
    "name": "plan",
    "command": "/workflow:plan",
-    "description": "5-phase planning workflow with action-planning-agent task generation, outputs IMPL_PLAN.md and task JSONs with optional CLI auto-execution",
-    "arguments": "[--cli-execute] \\\"text description\\\"|file.md",
+    "description": "5-phase planning workflow with action-planning-agent task generation, outputs IMPL_PLAN.md and task JSONs",
+    "arguments": "\\\"text description\\\"|file.md",
    "category": "workflow",
    "subcategory": null,
    "usage_scenario": "planning",
@@ -531,7 +531,7 @@
    "name": "tdd-plan",
    "command": "/workflow:tdd-plan",
    "description": "TDD workflow planning with Red-Green-Refactor task chain generation, test-first development structure, and cycle tracking",
-    "arguments": "[--cli-execute] \\\"feature description\\\"|file.md",
+    "arguments": "\\\"feature description\\\"|file.md",
    "category": "workflow",
    "subcategory": null,
    "usage_scenario": "planning",
@@ -564,7 +564,7 @@
    "name": "test-fix-gen",
    "command": "/workflow:test-fix-gen",
    "description": "Create test-fix workflow session from session ID, description, or file path with test strategy generation and task planning",
-    "arguments": "[--use-codex] [--cli-execute] (source-session-id | \\\"feature description\\\" | /path/to/file.md)",
+    "arguments": "(source-session-id | \\\"feature description\\\" | /path/to/file.md)",
    "category": "workflow",
    "subcategory": null,
    "usage_scenario": "testing",
@@ -575,7 +575,7 @@
    "name": "test-gen",
    "command": "/workflow:test-gen",
    "description": "Create independent test-fix workflow session from completed implementation session, analyzes code to generate test tasks",
-    "arguments": "[--use-codex] [--cli-execute] source-session-id",
+    "arguments": "source-session-id",
    "category": "workflow",
    "subcategory": null,
    "usage_scenario": "testing",
@@ -608,7 +608,7 @@
    "name": "task-generate-agent",
    "command": "/workflow:tools:task-generate-agent",
    "description": "Generate implementation plan documents (IMPL_PLAN.md, task JSONs, TODO_LIST.md) using action-planning-agent - produces planning artifacts, does NOT execute code implementation",
-    "arguments": "--session WFS-session-id [--cli-execute]",
+    "arguments": "--session WFS-session-id",
    "category": "workflow",
    "subcategory": "tools",
    "usage_scenario": "implementation",
@@ -619,7 +619,7 @@
    "name": "task-generate-tdd",
    "command": "/workflow:tools:task-generate-tdd",
    "description": "Autonomous TDD task generation using action-planning-agent with Red-Green-Refactor cycles, test-first structure, and cycle validation",
-    "arguments": "--session WFS-session-id [--cli-execute]",
+    "arguments": "--session WFS-session-id",
    "category": "workflow",
    "subcategory": "tools",
    "usage_scenario": "implementation",
@@ -663,7 +663,7 @@
    "name": "test-task-generate",
    "command": "/workflow:tools:test-task-generate",
    "description": "Generate test planning documents (IMPL_PLAN.md, test task JSONs, TODO_LIST.md) using action-planning-agent - produces test planning artifacts, does NOT execute tests",
-    "arguments": "[--use-codex] [--cli-execute] --session WFS-test-session-id",
+    "arguments": "--session WFS-test-session-id",
    "category": "workflow",
    "subcategory": "tools",
    "usage_scenario": "implementation",
--- a/.claude/skills/command-guide/index/by-category.json
+++ b/.claude/skills/command-guide/index/by-category.json
@@ -295,8 +295,8 @@
      {
        "name": "plan",
        "command": "/workflow:plan",
-        "description": "5-phase planning workflow with action-planning-agent task generation, outputs IMPL_PLAN.md and task JSONs with optional CLI auto-execution",
-        "arguments": "[--cli-execute] \\\"text description\\\"|file.md",
+        "description": "5-phase planning workflow with action-planning-agent task generation, outputs IMPL_PLAN.md and task JSONs",
+        "arguments": "\\\"text description\\\"|file.md",
        "category": "workflow",
        "subcategory": null,
        "usage_scenario": "planning",
@@ -373,7 +373,7 @@
        "name": "tdd-plan",
        "command": "/workflow:tdd-plan",
        "description": "TDD workflow planning with Red-Green-Refactor task chain generation, test-first development structure, and cycle tracking",
-        "arguments": "[--cli-execute] \\\"feature description\\\"|file.md",
+        "arguments": "\\\"feature description\\\"|file.md",
        "category": "workflow",
        "subcategory": null,
        "usage_scenario": "planning",
@@ -406,7 +406,7 @@
        "name": "test-fix-gen",
        "command": "/workflow:test-fix-gen",
        "description": "Create test-fix workflow session from session ID, description, or file path with test strategy generation and task planning",
-        "arguments": "[--use-codex] [--cli-execute] (source-session-id | \\\"feature description\\\" | /path/to/file.md)",
+        "arguments": "(source-session-id | \\\"feature description\\\" | /path/to/file.md)",
        "category": "workflow",
        "subcategory": null,
        "usage_scenario": "testing",
@@ -417,7 +417,7 @@
        "name": "test-gen",
        "command": "/workflow:test-gen",
        "description": "Create independent test-fix workflow session from completed implementation session, analyzes code to generate test tasks",
-        "arguments": "[--use-codex] [--cli-execute] source-session-id",
+        "arguments": "source-session-id",
        "category": "workflow",
        "subcategory": null,
        "usage_scenario": "testing",
@@ -632,7 +632,7 @@
        "name": "task-generate-agent",
        "command": "/workflow:tools:task-generate-agent",
        "description": "Generate implementation plan documents (IMPL_PLAN.md, task JSONs, TODO_LIST.md) using action-planning-agent - produces planning artifacts, does NOT execute code implementation",
-        "arguments": "--session WFS-session-id [--cli-execute]",
+        "arguments": "--session WFS-session-id",
        "category": "workflow",
        "subcategory": "tools",
        "usage_scenario": "implementation",
@@ -643,7 +643,7 @@
        "name": "task-generate-tdd",
        "command": "/workflow:tools:task-generate-tdd",
        "description": "Autonomous TDD task generation using action-planning-agent with Red-Green-Refactor cycles, test-first structure, and cycle validation",
-        "arguments": "--session WFS-session-id [--cli-execute]",
+        "arguments": "--session WFS-session-id",
        "category": "workflow",
        "subcategory": "tools",
        "usage_scenario": "implementation",
@@ -687,7 +687,7 @@
        "name": "test-task-generate",
        "command": "/workflow:tools:test-task-generate",
        "description": "Generate test planning documents (IMPL_PLAN.md, test task JSONs, TODO_LIST.md) using action-planning-agent - produces test planning artifacts, does NOT execute tests",
-        "arguments": "[--use-codex] [--cli-execute] --session WFS-test-session-id",
+        "arguments": "--session WFS-test-session-id",
        "category": "workflow",
        "subcategory": "tools",
        "usage_scenario": "implementation",
--- a/.claude/skills/command-guide/index/by-use-case.json
+++ b/.claude/skills/command-guide/index/by-use-case.json
@@ -469,8 +469,8 @@
    {
      "name": "plan",
      "command": "/workflow:plan",
-      "description": "5-phase planning workflow with action-planning-agent task generation, outputs IMPL_PLAN.md and task JSONs with optional CLI auto-execution",
-      "arguments": "[--cli-execute] \\\"text description\\\"|file.md",
+      "description": "5-phase planning workflow with action-planning-agent task generation, outputs IMPL_PLAN.md and task JSONs",
+      "arguments": "\\\"text description\\\"|file.md",
      "category": "workflow",
      "subcategory": null,
      "usage_scenario": "planning",
@@ -492,7 +492,7 @@
      "name": "tdd-plan",
      "command": "/workflow:tdd-plan",
      "description": "TDD workflow planning with Red-Green-Refactor task chain generation, test-first development structure, and cycle tracking",
-      "arguments": "[--cli-execute] \\\"feature description\\\"|file.md",
+      "arguments": "\\\"feature description\\\"|file.md",
      "category": "workflow",
      "subcategory": null,
      "usage_scenario": "planning",
@@ -604,7 +604,7 @@
      "name": "task-generate-agent",
      "command": "/workflow:tools:task-generate-agent",
      "description": "Generate implementation plan documents (IMPL_PLAN.md, task JSONs, TODO_LIST.md) using action-planning-agent - produces planning artifacts, does NOT execute code implementation",
-      "arguments": "--session WFS-session-id [--cli-execute]",
+      "arguments": "--session WFS-session-id",
      "category": "workflow",
      "subcategory": "tools",
      "usage_scenario": "implementation",
@@ -615,7 +615,7 @@
      "name": "task-generate-tdd",
      "command": "/workflow:tools:task-generate-tdd",
      "description": "Autonomous TDD task generation using action-planning-agent with Red-Green-Refactor cycles, test-first structure, and cycle validation",
-      "arguments": "--session WFS-session-id [--cli-execute]",
+      "arguments": "--session WFS-session-id",
      "category": "workflow",
      "subcategory": "tools",
      "usage_scenario": "implementation",
@@ -626,7 +626,7 @@
      "name": "test-task-generate",
      "command": "/workflow:tools:test-task-generate",
      "description": "Generate test planning documents (IMPL_PLAN.md, test task JSONs, TODO_LIST.md) using action-planning-agent - produces test planning artifacts, does NOT execute tests",
-      "arguments": "[--use-codex] [--cli-execute] --session WFS-test-session-id",
+      "arguments": "--session WFS-test-session-id",
      "category": "workflow",
      "subcategory": "tools",
      "usage_scenario": "implementation",
@@ -742,7 +742,7 @@
      "name": "test-fix-gen",
      "command": "/workflow:test-fix-gen",
      "description": "Create test-fix workflow session from session ID, description, or file path with test strategy generation and task planning",
-      "arguments": "[--use-codex] [--cli-execute] (source-session-id | \\\"feature description\\\" | /path/to/file.md)",
+      "arguments": "(source-session-id | \\\"feature description\\\" | /path/to/file.md)",
      "category": "workflow",
      "subcategory": null,
      "usage_scenario": "testing",
@@ -753,7 +753,7 @@
      "name": "test-gen",
      "command": "/workflow:test-gen",
      "description": "Create independent test-fix workflow session from completed implementation session, analyzes code to generate test tasks",
-      "arguments": "[--use-codex] [--cli-execute] source-session-id",
+      "arguments": "source-session-id",
      "category": "workflow",
      "subcategory": null,
      "usage_scenario": "testing",
--- a/.claude/skills/command-guide/index/essential-commands.json
+++ b/.claude/skills/command-guide/index/essential-commands.json
@@ -24,8 +24,8 @@
  {
    "name": "plan",
    "command": "/workflow:plan",
-    "description": "5-phase planning workflow with action-planning-agent task generation, outputs IMPL_PLAN.md and task JSONs with optional CLI auto-execution",
-    "arguments": "[--cli-execute] \\\"text description\\\"|file.md",
+    "description": "5-phase planning workflow with action-planning-agent task generation, outputs IMPL_PLAN.md and task JSONs",
+    "arguments": "\\\"text description\\\"|file.md",
    "category": "workflow",
    "subcategory": null,
    "usage_scenario": "planning",
--- a/.claude/skills/command-guide/reference/agents/action-planning-agent.md
+++ b/.claude/skills/command-guide/reference/agents/action-planning-agent.md
@@ -43,7 +43,6 @@ You are a pure execution agent specialized in creating actionable implementation
  - `context_package_path`: Context package with brainstorming artifacts catalog
 - **Metadata**: Simple values
  - `session_id`: Workflow session identifier (WFS-[topic])
-  - `execution_mode`: agent-mode | cli-execute-mode
  - `mcp_capabilities`: Available MCP tools (exa_code, exa_web, code_index)

 **Legacy Support** (backward compatibility):
@@ -244,8 +243,7 @@ Generate individual `.task/IMPL-*.json` files with the following structure:
    "type": "test-gen|test-fix",
    "agent": "@code-developer|@test-fix-agent",
    "test_framework": "jest|vitest|pytest|junit|mocha",
-    "coverage_target": "80%",
-    "use_codex": true|false
+    "coverage_target": "80%"
  }
 }
 ```
@@ -253,7 +251,8 @@ Generate individual `.task/IMPL-*.json` files with the following structure:
 **Test-Specific Fields**:
 - `test_framework`: Existing test framework from project (required for test tasks)
 - `coverage_target`: Target code coverage percentage (optional)
- `use_codex`: Whether to use Codex for automated fixes in test-fix tasks (optional, default: false)
+
+**Note**: CLI tool usage for test-fix tasks is now controlled via `flow_control.implementation_approach` steps with `command` fields, not via `meta.use_codex`.

 #### Context Object

@@ -485,15 +484,31 @@ The `implementation_approach` supports **two execution modes** based on the pres
     - `bash(codex --full-auto exec '[task]' resume --last --skip-git-repo-check -s danger-full-access)` (multi-step)
     - `bash(cd [path] && gemini -p '[prompt]' --approval-mode yolo)` (write mode)

-**Mode Selection Strategy**:
- **Default to agent execution** for most tasks
- **Use CLI mode** when:
-  - User explicitly requests CLI tool (codex/gemini/qwen)
-  - Task requires multi-step autonomous reasoning beyond agent capability
-  - Complex refactoring needs specialized tool analysis
-  - Building on previous CLI execution context (use `resume --last`)
+**Semantic CLI Tool Selection**:

-**Key Principle**: The `command` field is **optional**. Agent must decide based on task complexity and user preference.
+Agent determines CLI tool usage per-step based on user semantics and task nature.
+
+**Source**: Scan `metadata.task_description` from context-package.json for CLI tool preferences.
+
+**User Semantic Triggers** (patterns to detect in task_description):
+- "use Codex/codex" → Add `command` field with Codex CLI
+- "use Gemini/gemini" → Add `command` field with Gemini CLI
+- "use Qwen/qwen" → Add `command` field with Qwen CLI
+- "CLI execution" / "automated" → Infer appropriate CLI tool
+
+**Task-Based Selection** (when no explicit user preference):
+- **Implementation/coding**: Codex preferred for autonomous development
+- **Analysis/exploration**: Gemini preferred for large context analysis
+- **Documentation**: Gemini/Qwen with write mode (`--approval-mode yolo`)
+- **Testing**: Depends on complexity - simple=agent, complex=Codex
+
+**Default Behavior**: Agent always executes the workflow. CLI commands are embedded in `implementation_approach` steps:
+- Agent orchestrates task execution
+- When step has `command` field, agent executes it via Bash
+- When step has no `command` field, agent implements directly
+- This maintains agent control while leveraging CLI tool power
+
+**Key Principle**: The `command` field is **optional**. Agent decides based on user semantics and task complexity.

 **Examples**:

--- a/.claude/skills/command-guide/reference/agents/cli-planning-agent.md
+++ b/.claude/skills/command-guide/reference/agents/cli-planning-agent.md
@@ -66,8 +66,7 @@ You are a specialized execution agent that bridges CLI analysis tools with task
  "task_config": {
    "agent": "@test-fix-agent",
    "type": "test-fix-iteration",
-    "max_iterations": 5,
-    "use_codex": false
+    "max_iterations": 5
  }
 }
 ```
@@ -263,7 +262,6 @@ function extractModificationPoints() {
    "analysis_report": ".process/iteration-{iteration}-analysis.md",
    "cli_output": ".process/iteration-{iteration}-cli-output.txt",
    "max_iterations": "{task_config.max_iterations}",
-    "use_codex": "{task_config.use_codex}",
    "parent_task": "{parent_task_id}",
    "created_by": "@cli-planning-agent",
    "created_at": "{timestamp}"
--- a/.claude/skills/command-guide/reference/agents/code-developer.md
+++ b/.claude/skills/command-guide/reference/agents/code-developer.md
@@ -24,8 +24,6 @@ You are a code execution specialist focused on implementing high-quality, produc
 - **Context-driven** - Use provided context and existing code patterns
 - **Quality over speed** - Write boring, reliable code that works

-
-
 ## Execution Process

 ### 1. Context Assessment
--- a/.claude/skills/command-guide/reference/agents/context-search-agent.md
+++ b/.claude/skills/command-guide/reference/agents/context-search-agent.md
@@ -100,10 +100,88 @@ if (!memory.has("README.md")) Read(README.md)

 ### Phase 2: Multi-Source Context Discovery

-Execute all 3 tracks in parallel for comprehensive coverage.
+Execute all tracks in parallel for comprehensive coverage.

 **Note**: Historical archive analysis (querying `.workflow/archives/manifest.json`) is optional and should be performed if the manifest exists. Inject findings into `conflict_detection.historical_conflicts[]`.

+#### Track 0: Exploration Synthesis (Optional)
+
+**Trigger**: When `explorations-manifest.json` exists in session `.process/` folder
+
+**Purpose**: Transform raw exploration data into prioritized, deduplicated insights. This is NOT simple aggregation - it synthesizes `critical_files` (priority-ranked), deduplicates patterns/integration_points, and generates `conflict_indicators`.
+
+```javascript
+// Check for exploration results from context-gather parallel explore phase
+const manifestPath = `.workflow/active/${session_id}/.process/explorations-manifest.json`;
+if (file_exists(manifestPath)) {
+  const manifest = JSON.parse(Read(manifestPath));
+
+  // Load full exploration data from each file
+  const explorationData = manifest.explorations.map(exp => ({
+    ...exp,
+    data: JSON.parse(Read(exp.path))
+  }));
+
+  // Build explorations array with summaries
+  const explorations = explorationData.map(exp => ({
+    angle: exp.angle,
+    file: exp.file,
+    path: exp.path,
+    index: exp.data._metadata?.exploration_index || exp.index,
+    summary: {
+      relevant_files_count: exp.data.relevant_files?.length || 0,
+      key_patterns: exp.data.patterns,
+      integration_points: exp.data.integration_points
+    }
+  }));
+
+  // SYNTHESIS (not aggregation): Transform raw data into prioritized insights
+  const aggregated_insights = {
+    // CRITICAL: Synthesize priority-ranked critical_files from multiple relevant_files lists
+    // - Deduplicate by path
+    // - Rank by: mention count across angles + individual relevance scores
+    // - Top 10-15 files only (focused, actionable)
+    critical_files: synthesizeCriticalFiles(explorationData.flatMap(e => e.data.relevant_files || [])),
+
+    // SYNTHESIS: Generate conflict indicators from pattern mismatches, constraint violations
+    conflict_indicators: synthesizeConflictIndicators(explorationData),
+
+    // Deduplicate clarification questions (merge similar questions)
+    clarification_needs: deduplicateQuestions(explorationData.flatMap(e => e.data.clarification_needs || [])),
+
+    // Preserve source attribution for traceability
+    constraints: explorationData.map(e => ({ constraint: e.data.constraints, source_angle: e.angle })).filter(c => c.constraint),
+
+    // Deduplicate patterns across angles (merge identical patterns)
+    all_patterns: deduplicatePatterns(explorationData.map(e => ({ patterns: e.data.patterns, source_angle: e.angle }))),
+
+    // Deduplicate integration points (merge by file:line location)
+    all_integration_points: deduplicateIntegrationPoints(explorationData.map(e => ({ points: e.data.integration_points, source_angle: e.angle })))
+  };
+
+  // Store for Phase 3 packaging
+  exploration_results = { manifest_path: manifestPath, exploration_count: manifest.exploration_count,
+                         complexity: manifest.complexity, angles: manifest.angles_explored,
+                         explorations, aggregated_insights };
+}
+
+// Synthesis helper functions (conceptual)
+function synthesizeCriticalFiles(allRelevantFiles) {
+  // 1. Group by path
+  // 2. Count mentions across angles
+  // 3. Average relevance scores
+  // 4. Rank by: (mention_count * 0.6) + (avg_relevance * 0.4)
+  // 5. Return top 10-15 with mentioned_by_angles attribution
+}
+
+function synthesizeConflictIndicators(explorationData) {
+  // 1. Detect pattern mismatches across angles
+  // 2. Identify constraint violations
+  // 3. Flag files mentioned with conflicting integration approaches
+  // 4. Assign severity: critical/high/medium/low
+}
+```
+
 #### Track 1: Reference Documentation

 Extract from Phase 0 loaded docs:
@@ -393,10 +471,39 @@ Calculate risk level based on:
    },
    "affected_modules": ["auth", "user-model", "middleware"],
    "mitigation_strategy": "Incremental refactoring with backward compatibility"
+  },
+  "exploration_results": {
+    "manifest_path": ".workflow/active/{session}/.process/explorations-manifest.json",
+    "exploration_count": 3,
+    "complexity": "Medium",
+    "angles": ["architecture", "dependencies", "testing"],
+    "explorations": [
+      {
+        "angle": "architecture",
+        "file": "exploration-architecture.json",
+        "path": ".workflow/active/{session}/.process/exploration-architecture.json",
+        "index": 1,
+        "summary": {
+          "relevant_files_count": 5,
+          "key_patterns": "Service layer with DI",
+          "integration_points": "Container.registerService:45-60"
+        }
+      }
+    ],
+    "aggregated_insights": {
+      "critical_files": [{"path": "src/auth/AuthService.ts", "relevance": 0.95, "mentioned_by_angles": ["architecture"]}],
+      "conflict_indicators": [{"type": "pattern_mismatch", "description": "...", "source_angle": "architecture", "severity": "medium"}],
+      "clarification_needs": [{"question": "...", "context": "...", "options": [], "source_angle": "architecture"}],
+      "constraints": [{"constraint": "Must follow existing DI pattern", "source_angle": "architecture"}],
+      "all_patterns": [{"patterns": "Service layer with DI", "source_angle": "architecture"}],
+      "all_integration_points": [{"points": "Container.registerService:45-60", "source_angle": "architecture"}]
+    }
  }
 }
 ```

+**Note**: `exploration_results` is populated when exploration files exist (from context-gather parallel explore phase). If no explorations, this field is omitted or empty.
+


 ## Quality Validation
--- a/.claude/skills/command-guide/reference/agents/test-fix-agent.md
+++ b/.claude/skills/command-guide/reference/agents/test-fix-agent.md
@@ -142,9 +142,9 @@ run_test_layer "L1-unit" "$UNIT_CMD"

 ### 3. Failure Diagnosis & Fixing Loop

-**Execution Modes**:
+**Execution Modes** (determined by `flow_control.implementation_approach`):

-**A. Manual Mode (Default, meta.use_codex=false)**:
+**A. Agent Mode (Default, no `command` field in steps)**:
 ```
 WHILE tests are failing AND iterations < max_iterations:
    1. Use Gemini to diagnose failure (bug-fix template)
@@ -155,17 +155,17 @@ WHILE tests are failing AND iterations < max_iterations:
 END WHILE
 ```

-**B. Codex Mode (meta.use_codex=true)**:
+**B. CLI Mode (`command` field present in implementation_approach steps)**:
 ```
 WHILE tests are failing AND iterations < max_iterations:
    1. Use Gemini to diagnose failure (bug-fix template)
-    2. Use Codex to apply fixes automatically with resume mechanism
+    2. Execute `command` field (e.g., Codex) to apply fixes automatically
    3. Re-run test suite
    4. Verify fix doesn't break other tests
 END WHILE
 ```

-**Codex Resume in Test-Fix Cycle** (when `meta.use_codex=true`):
+**Codex Resume in Test-Fix Cycle** (when step has `command` with Codex):
 - First iteration: Start new Codex session with full context
 - Subsequent iterations: Use `resume --last` to maintain fix history and apply consistent strategies

--- a/.claude/skills/command-guide/reference/commands/workflow/action-plan-verify.md
+++ b/.claude/skills/command-guide/reference/commands/workflow/action-plan-verify.md
@@ -101,7 +101,7 @@ Load only minimal necessary context from each artifact:
 - Dependencies (depends_on, blocks)
 - Context (requirements, focus_paths, acceptance, artifacts)
 - Flow control (pre_analysis, implementation_approach)
- Meta (complexity, priority, use_codex)
+- Meta (complexity, priority)

 ### 3. Build Semantic Models

--- a/.claude/skills/command-guide/reference/commands/workflow/lite-fix.md
+++ b/.claude/skills/command-guide/reference/commands/workflow/lite-fix.md
--- a/.claude/skills/command-guide/reference/commands/workflow/plan.md
+++ b/.claude/skills/command-guide/reference/commands/workflow/plan.md
@@ -1,7 +1,7 @@
 ---
 name: plan
-description: 5-phase planning workflow with action-planning-agent task generation, outputs IMPL_PLAN.md and task JSONs with optional CLI auto-execution
-argument-hint: "[--cli-execute] \"text description\"|file.md"
+description: 5-phase planning workflow with action-planning-agent task generation, outputs IMPL_PLAN.md and task JSONs
+argument-hint: "\"text description\"|file.md"
 allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 ---

@@ -69,7 +69,7 @@ Phase 3: Conflict Resolution (conditional)
      └─ conflict_risk < medium → Skip to Phase 4

 Phase 4: Task Generation
-   └─ /workflow:tools:task-generate-agent --session sessionId [--cli-execute]
+   └─ /workflow:tools:task-generate-agent --session sessionId
      └─ Output: IMPL_PLAN.md, task JSONs, TODO_LIST.md

 Return:
@@ -273,15 +273,10 @@ SlashCommand(command="/workflow:tools:conflict-resolution --session [sessionId]
 **Step 4.1: Dispatch** - Generate implementation plan and task JSONs

 ```javascript
-// Default (agent mode)
 SlashCommand(command="/workflow:tools:task-generate-agent --session [sessionId]")
-
-// With CLI execution (if --cli-execute flag present)
-SlashCommand(command="/workflow:tools:task-generate-agent --session [sessionId] --cli-execute")
 ```

-**Flag**:
- `--cli-execute`: Generate tasks with Codex execution commands
+**CLI Execution Note**: CLI tool usage is now determined semantically by action-planning-agent based on user's task description. If user specifies "use Codex/Gemini/Qwen for X", the agent embeds `command` fields in relevant `implementation_approach` steps.

 **Input**: `sessionId` from Phase 1

@@ -423,7 +418,7 @@ Phase 3: conflict-resolution [AUTO-TRIGGERED if conflict_risk ≥ medium]
    ↓ Output: Modified brainstorm artifacts (NO report file)
    ↓ Skip if conflict_risk is none/low → proceed directly to Phase 4
    ↓
-Phase 4: task-generate-agent --session sessionId [--cli-execute]
+Phase 4: task-generate-agent --session sessionId
    ↓ Input: sessionId + resolved brainstorm artifacts + session memory
    ↓ Output: IMPL_PLAN.md, task JSONs, TODO_LIST.md
    ↓
@@ -504,9 +499,7 @@ Return summary to user
 - **If conflict_risk ≥ medium**: Launch Phase 3 conflict-resolution with sessionId and contextPath
 - Wait for Phase 3 to finish executing (if executed), verify CONFLICT_RESOLUTION.md created
 - **If conflict_risk is none/low**: Skip Phase 3, proceed directly to Phase 4
- **Build Phase 4 command**:
-  - Base command: `/workflow:tools:task-generate-agent --session [sessionId]`
-  - Add `--cli-execute` if flag present
+- **Build Phase 4 command**: `/workflow:tools:task-generate-agent --session [sessionId]`
 - Pass session ID to Phase 4 command
 - Verify all Phase 4 outputs
 - Update TodoWrite after each phase (dynamically adjust for Phase 3 presence)
--- a/.claude/skills/command-guide/reference/commands/workflow/tdd-plan.md
+++ b/.claude/skills/command-guide/reference/commands/workflow/tdd-plan.md
@@ -1,7 +1,7 @@
 ---
 name: tdd-plan
 description: TDD workflow planning with Red-Green-Refactor task chain generation, test-first development structure, and cycle tracking
-argument-hint: "[--cli-execute] \"feature description\"|file.md"
+argument-hint: "\"feature description\"|file.md"
 allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 ---

@@ -11,9 +11,7 @@ allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)

 **This command is a pure orchestrator**: Dispatches 6 slash commands in sequence, parse outputs, pass context, and ensure complete TDD workflow creation with Red-Green-Refactor task generation.

-**Execution Modes**:
- **Agent Mode** (default): Use `/workflow:tools:task-generate-tdd` (autonomous agent-driven)
- **CLI Mode** (`--cli-execute`): Use `/workflow:tools:task-generate-tdd --cli-execute` (Gemini/Qwen)
+**CLI Tool Selection**: CLI tool usage is determined semantically from user's task description. Include "use Codex/Gemini/Qwen" in your request for CLI execution.

 **Task Attachment Model**:
 - SlashCommand dispatch **expands workflow** by attaching sub-tasks to current TodoWrite
@@ -235,13 +233,11 @@ SlashCommand(command="/workflow:tools:conflict-resolution --session [sessionId]
 **Step 5.1: Dispatch** - TDD task generation via action-planning-agent

 ```javascript
-// Agent Mode (default)
 SlashCommand(command="/workflow:tools:task-generate-tdd --session [sessionId]")
-
-// CLI Mode (--cli-execute flag)
-SlashCommand(command="/workflow:tools:task-generate-tdd --session [sessionId] --cli-execute")
 ```

+**Note**: CLI tool usage is determined semantically from user's task description.
+
 **Parse**: Extract feature count, task count (not chain count - tasks now contain internal TDD cycles)

 **Validate**:
@@ -454,8 +450,7 @@ Convert user input to TDD-structured format:
 - `/workflow:tools:test-context-gather` - Phase 3: Analyze existing test patterns and coverage
 - `/workflow:tools:conflict-resolution` - Phase 4: Detect and resolve conflicts (auto-triggered if conflict_risk ≥ medium)
 - `/compact` - Phase 4: Memory optimization (if context approaching limits)
- `/workflow:tools:task-generate-tdd` - Phase 5: Generate TDD tasks with agent-driven approach (default, autonomous)
- `/workflow:tools:task-generate-tdd --cli-execute` - Phase 5: Generate TDD tasks with CLI tools (Gemini/Qwen, when `--cli-execute` flag used)
+- `/workflow:tools:task-generate-tdd` - Phase 5: Generate TDD tasks (CLI tool usage determined semantically)

 **Follow-up Commands**:
 - `/workflow:action-plan-verify` - Recommended: Verify TDD plan quality and structure before execution
--- a/.claude/skills/command-guide/reference/commands/workflow/test-fix-gen.md
+++ b/.claude/skills/command-guide/reference/commands/workflow/test-fix-gen.md
@@ -1,7 +1,7 @@
 ---
 name: test-fix-gen
 description: Create test-fix workflow session from session ID, description, or file path with test strategy generation and task planning
-argument-hint: "[--use-codex] [--cli-execute] (source-session-id | \"feature description\" | /path/to/file.md)"
+argument-hint: "(source-session-id | \"feature description\" | /path/to/file.md)"
 allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 ---

@@ -43,7 +43,7 @@ fi
 - **Session Isolation**: Creates independent `WFS-test-[slug]` session
 - **Context-First**: Gathers implementation context via appropriate method
 - **Format Reuse**: Creates standard `IMPL-*.json` tasks with `meta.type: "test-fix"`
- **Manual First**: Default to manual fixes, use `--use-codex` for automation
+- **Semantic CLI Selection**: CLI tool usage determined from user's task description
 - **Automatic Detection**: Input pattern determines execution mode

 ### Coordinator Role
@@ -79,16 +79,14 @@ This command is a **pure planning coordinator**:

 ```bash
 # Basic syntax
-/workflow:test-fix-gen [FLAGS] <INPUT>
-
-# Flags (optional)
--use-codex        # Enable Codex automated fixes in IMPL-002
--cli-execute      # Enable CLI execution in IMPL-001
+/workflow:test-fix-gen <INPUT>

 # Input
 <INPUT>            # Session ID, description, or file path
 ```

+**Note**: CLI tool usage is determined semantically from the task description. To request CLI execution, include it in your description (e.g., "use Codex for automated fixes").
+
 ### Usage Examples

 #### Session Mode
@@ -96,11 +94,8 @@ This command is a **pure planning coordinator**:
 # Test validation for completed implementation
 /workflow:test-fix-gen WFS-user-auth-v2

-# With automated fixes
-/workflow:test-fix-gen --use-codex WFS-api-endpoints
-
-# With CLI execution
-/workflow:test-fix-gen --cli-execute --use-codex WFS-payment-flow
+# With semantic CLI request
+/workflow:test-fix-gen WFS-api-endpoints  # Add "use Codex" in description for automated fixes
 ```

 #### Prompt Mode - Text Description
@@ -108,17 +103,14 @@ This command is a **pure planning coordinator**:
 # Generate tests from feature description
 /workflow:test-fix-gen "Test the user authentication API endpoints in src/auth/api.ts"

-# With automated fixes
-/workflow:test-fix-gen --use-codex "Test user registration and login flows"
+# With CLI execution (semantic)
+/workflow:test-fix-gen "Test user registration and login flows, use Codex for automated fixes"
 ```

 #### Prompt Mode - File Reference
 ```bash
 # Generate tests from requirements file
 /workflow:test-fix-gen ./docs/api-requirements.md
-
-# With flags
-/workflow:test-fix-gen --use-codex --cli-execute ./specs/feature.md
 ```

 ### Mode Comparison
@@ -143,7 +135,7 @@ This command is a **pure planning coordinator**:
 5. **Complete All Phases**: Do not return until Phase 5 completes
 6. **Track Progress**: Update TodoWrite dynamically with task attachment/collapse pattern
 7. **Automatic Detection**: Mode auto-detected from input pattern
-8. **Parse Flags**: Extract `--use-codex` and `--cli-execute` flags for Phase 4
+8. **Semantic CLI Detection**: CLI tool usage determined from user's task description for Phase 4
 9. **Task Attachment Model**: SlashCommand dispatch **attaches** sub-tasks to current workflow. Orchestrator **executes** these attached tasks itself, then **collapses** them after completion
 10. **⚠️ CRITICAL: DO NOT STOP**: Continuous multi-phase workflow. After executing all attached tasks, immediately collapse them and execute next phase

@@ -283,13 +275,13 @@ For each targeted file/function, Gemini MUST generate:
 **Step 4.1: Dispatch** - Generate test task JSONs

 ```javascript
-SlashCommand(command="/workflow:tools:test-task-generate [--use-codex] [--cli-execute] --session [testSessionId]")
+SlashCommand(command="/workflow:tools:test-task-generate --session [testSessionId]")
 ```

 **Input**:
 - `testSessionId` from Phase 1
- `--use-codex` flag (if present) - Controls IMPL-002 fix mode
- `--cli-execute` flag (if present) - Controls IMPL-001 generation mode
+
+**Note**: CLI tool usage is determined semantically from user's task description.

 **Expected Behavior**:
 - Parse TEST_ANALYSIS_RESULTS.md from Phase 3 (multi-layered test plan)
@@ -422,7 +414,7 @@ CRITICAL - Next Steps:
 - **Phase 2**: Mode-specific context gathering (session summaries vs codebase analysis)
 - **Phase 3**: Multi-layered test requirements analysis (L0: Static, L1: Unit, L2: Integration, L3: E2E)
 - **Phase 4**: Multi-task generation with quality gate (IMPL-001, IMPL-001.5-review, IMPL-002)
- **Fix Mode Configuration**: `--use-codex` flag controls IMPL-002 fix mode (manual vs automated)
+- **Fix Mode Configuration**: CLI tool usage determined semantically from user's task description


 ---
@@ -521,16 +513,15 @@ If quality gate fails:
 - Task ID: `IMPL-002`
 - `meta.type: "test-fix"`
 - `meta.agent: "@test-fix-agent"`
- `meta.use_codex: true|false` (based on `--use-codex` flag)
 - `context.depends_on: ["IMPL-001"]`
 - `context.requirements`: Execute and fix tests

 **Test-Fix Cycle Specification**:
 **Note**: This specification describes what test-cycle-execute orchestrator will do. The agent only executes single tasks.
- **Cycle Pattern** (orchestrator-managed): test → gemini_diagnose → manual_fix (or codex) → retest
+- **Cycle Pattern** (orchestrator-managed): test → gemini_diagnose → fix (agent or CLI) → retest
 - **Tools Configuration** (orchestrator-controlled):
  - Gemini for analysis with bug-fix template → surgical fix suggestions
-  - Manual fix application (default) OR Codex if `--use-codex` flag (resume mechanism)
+  - Agent fix application (default) OR CLI if `command` field present in implementation_approach
 - **Exit Conditions** (orchestrator-enforced):
  - Success: All tests pass
  - Failure: Max iterations reached (5)
@@ -674,8 +665,7 @@ Key Points:
 4. **Mode Selection**:
   - Use **Session Mode** for completed workflow validation
   - Use **Prompt Mode** for ad-hoc test generation
-   - Use `--use-codex` for autonomous fix application
-   - Use `--cli-execute` for enhanced generation capabilities
+   - Include "use Codex" in description for autonomous fix application

 ## Related Commands

@@ -688,9 +678,7 @@ Key Points:
 - `/workflow:tools:test-context-gather` - Phase 2 (Session Mode): Gather source session context
 - `/workflow:tools:context-gather` - Phase 2 (Prompt Mode): Analyze codebase directly
 - `/workflow:tools:test-concept-enhanced` - Phase 3: Generate test requirements using Gemini
- `/workflow:tools:test-task-generate` - Phase 4: Generate test task JSONs using action-planning-agent (autonomous, default)
- `/workflow:tools:test-task-generate --use-codex` - Phase 4: With automated Codex fixes for IMPL-002 (when `--use-codex` flag used)
- `/workflow:tools:test-task-generate --cli-execute` - Phase 4: With CLI execution mode for IMPL-001 test generation (when `--cli-execute` flag used)
+- `/workflow:tools:test-task-generate` - Phase 4: Generate test task JSONs (CLI tool usage determined semantically)

 **Follow-up Commands**:
 - `/workflow:status` - Review generated test tasks
--- a/.claude/skills/command-guide/reference/commands/workflow/test-gen.md
+++ b/.claude/skills/command-guide/reference/commands/workflow/test-gen.md
@@ -1,7 +1,7 @@
 ---
 name: test-gen
 description: Create independent test-fix workflow session from completed implementation session, analyzes code to generate test tasks
-argument-hint: "[--use-codex] [--cli-execute] source-session-id"
+argument-hint: "source-session-id"
 allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 ---

@@ -16,7 +16,7 @@ allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 - **Context-First**: Prioritizes gathering code changes and summaries from source session
 - **Format Reuse**: Creates standard `IMPL-*.json` task, using `meta.type: "test-fix"` for agent assignment
 - **Parameter Simplification**: Tools auto-detect test session type via metadata, no manual cross-session parameters needed
- **Manual First**: Default to manual fixes, use `--use-codex` flag for automated Codex fix application
+- **Semantic CLI Selection**: CLI tool usage is determined by user's task description (e.g., "use Codex for fixes")

 **Task Attachment Model**:
 - SlashCommand dispatch **expands workflow** by attaching sub-tasks to current TodoWrite
@@ -48,7 +48,7 @@ allowed-tools: SlashCommand(*), TodoWrite(*), Read(*), Bash(*)
 5. **Complete All Phases**: Do not return to user until Phase 5 completes (summary returned)
 6. **Track Progress**: Update TodoWrite dynamically with task attachment/collapse pattern
 7. **Automatic Detection**: context-gather auto-detects test session and gathers source session context
-8. **Parse --use-codex Flag**: Extract flag from arguments and pass to Phase 4 (test-task-generate)
+8. **Semantic CLI Selection**: CLI tool usage determined from user's task description, passed to Phase 4
 9. **Command Boundary**: This command ends at Phase 5 summary. Test execution is NOT part of this command.
 10. **Task Attachment Model**: SlashCommand dispatch **attaches** sub-tasks to current workflow. Orchestrator **executes** these attached tasks itself, then **collapses** them after completion
 11. **⚠️ CRITICAL: DO NOT STOP**: Continuous multi-phase workflow. After executing all attached tasks, immediately collapse them and execute next phase
@@ -224,13 +224,13 @@ SlashCommand(command="/workflow:tools:test-concept-enhanced --session [testSessi
 **Step 4.1: Dispatch** - Generate test task JSON files and planning documents

 ```javascript
-SlashCommand(command="/workflow:tools:test-task-generate [--use-codex] [--cli-execute] --session [testSessionId]")
+SlashCommand(command="/workflow:tools:test-task-generate --session [testSessionId]")
 ```

 **Input**:
 - `testSessionId` from Phase 1
- `--use-codex` flag (if present in original command) - Controls IMPL-002 fix mode
- `--cli-execute` flag (if present in original command) - Controls IMPL-001 generation mode
+
+**Note**: CLI tool usage for fixes is determined semantically from user's task description (e.g., "use Codex for automated fixes").

 **Expected Behavior**:
 - Parse TEST_ANALYSIS_RESULTS.md from Phase 3
@@ -260,16 +260,15 @@ SlashCommand(command="/workflow:tools:test-task-generate [--use-codex] [--cli-ex
 - Task ID: `IMPL-002`
 - `meta.type: "test-fix"`
 - `meta.agent: "@test-fix-agent"`
- `meta.use_codex: true|false` (based on --use-codex flag)
 - `context.depends_on: ["IMPL-001"]`
 - `context.requirements`: Execute and fix tests
 - `flow_control.implementation_approach.test_fix_cycle`: Complete cycle specification
-  - **Cycle pattern**: test → gemini_diagnose → manual_fix (or codex if --use-codex) → retest
-  - **Tools configuration**: Gemini for analysis with bug-fix template, manual or Codex for fixes
+  - **Cycle pattern**: test → gemini_diagnose → fix (agent or CLI based on `command` field) → retest
+  - **Tools configuration**: Gemini for analysis with bug-fix template, agent or CLI for fixes
  - **Exit conditions**: Success (all pass) or failure (max iterations)
 - `flow_control.implementation_approach.modification_points`: 3-phase execution flow
  - Phase 1: Initial test execution
-  - Phase 2: Iterative Gemini diagnosis + manual/Codex fixes (based on flag)
+  - Phase 2: Iterative Gemini diagnosis + fixes (agent or CLI based on step's `command` field)
  - Phase 3: Final validation and certification

 <!-- TodoWrite: When test-task-generate dispatched, INSERT 3 test-task-generate tasks -->
@@ -327,7 +326,7 @@ Artifacts Created:

 Test Framework: [detected framework]
 Test Files to Generate: [count]
-Fix Mode: [Manual|Codex Automated] (based on --use-codex flag)
+Fix Mode: [Agent|CLI] (based on `command` field in implementation_approach steps)

 Review Generated Artifacts:
 - Test plan: .workflow/[testSessionId]/IMPL_PLAN.md
@@ -373,7 +372,7 @@ Ready for execution. Use appropriate workflow commands to proceed.
 - **Phase 2**: Cross-session context gathering from source implementation session
 - **Phase 3**: Test requirements analysis with Gemini for generation strategy
 - **Phase 4**: Dual-task generation (IMPL-001 for test generation, IMPL-002 for test execution)
- **Fix Mode Configuration**: `--use-codex` flag controls IMPL-002 fix mode (manual vs automated)
+- **Fix Mode Configuration**: CLI tool usage determined semantically from user's task description



@@ -444,7 +443,7 @@ Generates two task definition files:
  - Agent: @test-fix-agent
  - Dependency: IMPL-001 must complete first
  - Max iterations: 5
-  - Fix mode: Manual or Codex (based on --use-codex flag)
+  - Fix mode: Agent or CLI (based on `command` field in implementation_approach)

 See `/workflow:tools:test-task-generate` for complete task JSON schemas.

@@ -481,11 +480,10 @@ Created in `.workflow/active/WFS-test-[session]/`:
 **IMPL-002.json Structure**:
 - `meta.type: "test-fix"`
 - `meta.agent: "@test-fix-agent"`
- `meta.use_codex`: true/false (based on --use-codex flag)
 - `context.depends_on: ["IMPL-001"]`
 - `flow_control.implementation_approach.test_fix_cycle`: Complete cycle specification
  - Gemini diagnosis template
-  - Fix application mode (manual/codex)
+  - Fix application mode (agent or CLI based on `command` field)
  - Max iterations: 5
 - `flow_control.implementation_approach.modification_points`: 3-phase flow

@@ -503,13 +501,11 @@ See `/workflow:tools:test-task-generate` for complete JSON schemas.
 **Prerequisite Commands**:
 - `/workflow:plan` or `/workflow:execute` - Complete implementation session that needs test validation

-**Dispatched by This Command** (5 phases):
+**Dispatched by This Command** (4 phases):
 - `/workflow:session:start` - Phase 1: Create independent test workflow session
 - `/workflow:tools:test-context-gather` - Phase 2: Analyze test coverage and gather source session context
 - `/workflow:tools:test-concept-enhanced` - Phase 3: Generate test requirements and strategy using Gemini
- `/workflow:tools:test-task-generate` - Phase 4: Generate test task JSONs using action-planning-agent (autonomous, default)
- `/workflow:tools:test-task-generate --use-codex` - Phase 4: With automated Codex fixes for IMPL-002 (when `--use-codex` flag used)
- `/workflow:tools:test-task-generate --cli-execute` - Phase 4: With CLI execution mode for IMPL-001 test generation (when `--cli-execute` flag used)
+- `/workflow:tools:test-task-generate` - Phase 4: Generate test task JSONs (CLI tool usage determined semantically)

 **Follow-up Commands**:
 - `/workflow:status` - Review generated test tasks
--- a/.claude/skills/command-guide/reference/commands/workflow/tools/conflict-resolution.md
+++ b/.claude/skills/command-guide/reference/commands/workflow/tools/conflict-resolution.md
@@ -114,35 +114,44 @@ Task(subagent_type="cli-execution-agent", prompt=`
  - Risk: {conflict_risk}
  - Files: {existing_files_list}

+  ## Exploration Context (from context-package.exploration_results)
+  - Exploration Count: ${contextPackage.exploration_results?.exploration_count || 0}
+  - Angles Analyzed: ${JSON.stringify(contextPackage.exploration_results?.angles || [])}
+  - Pre-identified Conflict Indicators: ${JSON.stringify(contextPackage.exploration_results?.aggregated_insights?.conflict_indicators || [])}
+  - Critical Files: ${JSON.stringify(contextPackage.exploration_results?.aggregated_insights?.critical_files?.map(f => f.path) || [])}
+  - All Patterns: ${JSON.stringify(contextPackage.exploration_results?.aggregated_insights?.all_patterns || [])}
+  - All Integration Points: ${JSON.stringify(contextPackage.exploration_results?.aggregated_insights?.all_integration_points || [])}
+
  ## Analysis Steps

  ### 1. Load Context
  - Read existing files from conflict_detection.existing_files
  - Load plan from .workflow/active/{session_id}/.process/context-package.json
+  - **NEW**: Load exploration_results and use aggregated_insights for enhanced analysis
  - Extract role analyses and requirements

-  ### 2. Execute CLI Analysis (Enhanced with Scenario Uniqueness Detection)
+  ### 2. Execute CLI Analysis (Enhanced with Exploration + Scenario Uniqueness)

  Primary (Gemini):
  cd {project_root} && gemini -p "
-  PURPOSE: Detect conflicts between plan and codebase, including module scenario overlaps
+  PURPOSE: Detect conflicts between plan and codebase, using exploration insights
  TASK:
-  • Compare architectures
+  • **Review pre-identified conflict_indicators from exploration results**
+  • Compare architectures (use exploration key_patterns)
  • Identify breaking API changes
  • Detect data model incompatibilities
  • Assess dependency conflicts
-  • **NEW: Analyze module scenario uniqueness**
-    - Extract new module functionality from plan
-    - Search all existing modules with similar functionality
-    - Compare scenario coverage and identify overlaps
+  • **Analyze module scenario uniqueness**
+    - Use exploration integration_points for precise locations
+    - Cross-validate with exploration critical_files
    - Generate clarification questions for boundary definition
  MODE: analysis
  CONTEXT: @**/*.ts @**/*.js @**/*.tsx @**/*.jsx @.workflow/active/{session_id}/**/*
-  EXPECTED: Conflict list with severity ratings, including ModuleOverlap conflicts with:
-    - Existing module list with scenarios
-    - Overlap analysis matrix
+  EXPECTED: Conflict list with severity ratings, including:
+    - Validation of exploration conflict_indicators
+    - ModuleOverlap conflicts with overlap_analysis
    - Targeted clarification questions
-  RULES: $(cat ~/.claude/workflows/cli-templates/prompts/analysis/02-analyze-code-patterns.txt) | Focus on breaking changes, migration needs, and functional overlaps | analysis=READ-ONLY
+  RULES: $(cat ~/.claude/workflows/cli-templates/prompts/analysis/02-analyze-code-patterns.txt) | Focus on breaking changes, migration needs, and functional overlaps | Prioritize exploration-identified conflicts | analysis=READ-ONLY
  "

  Fallback: Qwen (same prompt) → Claude (manual analysis)
--- a/.claude/skills/command-guide/reference/commands/workflow/tools/context-gather.md
+++ b/.claude/skills/command-guide/reference/commands/workflow/tools/context-gather.md
@@ -36,24 +36,23 @@ Step 1: Context-Package Detection
      ├─ Valid package exists → Return existing (skip execution)
      └─ No valid package → Continue to Step 2

-Step 2: Invoke Context-Search Agent
-   ├─ Phase 1: Initialization & Pre-Analysis
-   │  ├─ Load project.json as primary context
-   │  ├─ Initialize code-index
-   │  └─ Classify complexity
-   ├─ Phase 2: Multi-Source Discovery
-   │  ├─ Track 1: Historical archive analysis
-   │  ├─ Track 2: Reference documentation
-   │  ├─ Track 3: Web examples (Exa MCP)
-   │  └─ Track 4: Codebase analysis (5-layer)
-   └─ Phase 3: Synthesis & Packaging
-      ├─ Apply relevance scoring
-      ├─ Integrate brainstorm artifacts
-      ├─ Perform conflict detection
-      └─ Generate context-package.json
+Step 2: Complexity Assessment & Parallel Explore (NEW)
+   ├─ Analyze task_description → classify Low/Medium/High
+   ├─ Select exploration angles (1-4 based on complexity)
+   ├─ Launch N cli-explore-agents in parallel
+   │  └─ Each outputs: exploration-{angle}.json
+   └─ Generate explorations-manifest.json

-Step 3: Output Verification
-   └─ Verify context-package.json created
+Step 3: Invoke Context-Search Agent (with exploration input)
+   ├─ Phase 1: Initialization & Pre-Analysis
+   ├─ Phase 2: Multi-Source Discovery
+   │  ├─ Track 0: Exploration Synthesis (prioritize & deduplicate)
+   │  ├─ Track 1-4: Existing tracks
+   └─ Phase 3: Synthesis & Packaging
+      └─ Generate context-package.json with exploration_results
+
+Step 4: Output Verification
+   └─ Verify context-package.json contains exploration_results
 ```

 ## Execution Flow
@@ -80,10 +79,139 @@ if (file_exists(contextPackagePath)) {
 }
 ```

-### Step 2: Invoke Context-Search Agent
+### Step 2: Complexity Assessment & Parallel Explore

 **Only execute if Step 1 finds no valid package**

+```javascript
+// 2.1 Complexity Assessment
+function analyzeTaskComplexity(taskDescription) {
+  const text = taskDescription.toLowerCase();
+  if (/architect|refactor|restructure|modular|cross-module/.test(text)) return 'High';
+  if (/multiple|several|integrate|migrate|extend/.test(text)) return 'Medium';
+  return 'Low';
+}
+
+const ANGLE_PRESETS = {
+  architecture: ['architecture', 'dependencies', 'modularity', 'integration-points'],
+  security: ['security', 'auth-patterns', 'dataflow', 'validation'],
+  performance: ['performance', 'bottlenecks', 'caching', 'data-access'],
+  bugfix: ['error-handling', 'dataflow', 'state-management', 'edge-cases'],
+  feature: ['patterns', 'integration-points', 'testing', 'dependencies'],
+  refactor: ['architecture', 'patterns', 'dependencies', 'testing']
+};
+
+function selectAngles(taskDescription, complexity) {
+  const text = taskDescription.toLowerCase();
+  let preset = 'feature';
+  if (/refactor|architect|restructure/.test(text)) preset = 'architecture';
+  else if (/security|auth|permission/.test(text)) preset = 'security';
+  else if (/performance|slow|optimi/.test(text)) preset = 'performance';
+  else if (/fix|bug|error|issue/.test(text)) preset = 'bugfix';
+
+  const count = complexity === 'High' ? 4 : (complexity === 'Medium' ? 3 : 1);
+  return ANGLE_PRESETS[preset].slice(0, count);
+}
+
+const complexity = analyzeTaskComplexity(task_description);
+const selectedAngles = selectAngles(task_description, complexity);
+const sessionFolder = `.workflow/active/${session_id}/.process`;
+
+// 2.2 Launch Parallel Explore Agents
+const explorationTasks = selectedAngles.map((angle, index) =>
+  Task(
+    subagent_type="cli-explore-agent",
+    description=`Explore: ${angle}`,
+    prompt=`
+## Task Objective
+Execute **${angle}** exploration for task planning context. Analyze codebase from this specific angle to discover relevant structure, patterns, and constraints.
+
+## Assigned Context
+- **Exploration Angle**: ${angle}
+- **Task Description**: ${task_description}
+- **Session ID**: ${session_id}
+- **Exploration Index**: ${index + 1} of ${selectedAngles.length}
+- **Output File**: ${sessionFolder}/exploration-${angle}.json
+
+## MANDATORY FIRST STEPS (Execute by Agent)
+**You (cli-explore-agent) MUST execute these steps in order:**
+1. Run: ~/.claude/scripts/get_modules_by_depth.sh (project structure)
+2. Run: rg -l "{keyword_from_task}" --type ts (locate relevant files)
+3. Execute: cat ~/.claude/workflows/cli-templates/schemas/explore-json-schema.json (get output schema reference)
+
+## Exploration Strategy (${angle} focus)
+
+**Step 1: Structural Scan** (Bash)
+- get_modules_by_depth.sh → identify modules related to ${angle}
+- find/rg → locate files relevant to ${angle} aspect
+- Analyze imports/dependencies from ${angle} perspective
+
+**Step 2: Semantic Analysis** (Gemini CLI)
+- How does existing code handle ${angle} concerns?
+- What patterns are used for ${angle}?
+- Where would new code integrate from ${angle} viewpoint?
+
+**Step 3: Write Output**
+- Consolidate ${angle} findings into JSON
+- Identify ${angle}-specific clarification needs
+
+## Expected Output
+
+**File**: ${sessionFolder}/exploration-${angle}.json
+
+**Schema Reference**: Schema obtained in MANDATORY FIRST STEPS step 3, follow schema exactly
+
+**Required Fields** (all ${angle} focused):
+- project_structure: Modules/architecture relevant to ${angle}
+- relevant_files: Files affected from ${angle} perspective
+  **IMPORTANT**: Use object format with relevance scores for synthesis:
+  \`[{path: "src/file.ts", relevance: 0.85, rationale: "Core ${angle} logic"}]\`
+  Scores: 0.7+ high priority, 0.5-0.7 medium, <0.5 low
+- patterns: ${angle}-related patterns to follow
+- dependencies: Dependencies relevant to ${angle}
+- integration_points: Where to integrate from ${angle} viewpoint (include file:line locations)
+- constraints: ${angle}-specific limitations/conventions
+- clarification_needs: ${angle}-related ambiguities (with options array)
+- _metadata.exploration_angle: "${angle}"
+
+## Success Criteria
+- [ ] Schema obtained via cat explore-json-schema.json
+- [ ] get_modules_by_depth.sh executed
+- [ ] At least 3 relevant files identified with ${angle} rationale
+- [ ] Patterns are actionable (code examples, not generic advice)
+- [ ] Integration points include file:line locations
+- [ ] Constraints are project-specific to ${angle}
+- [ ] JSON output follows schema exactly
+- [ ] clarification_needs includes options array
+
+## Output
+Write: ${sessionFolder}/exploration-${angle}.json
+Return: 2-3 sentence summary of ${angle} findings
+`
+  )
+);
+
+// 2.3 Generate Manifest after all complete
+const explorationFiles = bash(`find ${sessionFolder} -name "exploration-*.json" -type f`).split('\n').filter(f => f.trim());
+const explorationManifest = {
+  session_id,
+  task_description,
+  timestamp: new Date().toISOString(),
+  complexity,
+  exploration_count: selectedAngles.length,
+  angles_explored: selectedAngles,
+  explorations: explorationFiles.map(file => {
+    const data = JSON.parse(Read(file));
+    return { angle: data._metadata.exploration_angle, file: file.split('/').pop(), path: file, index: data._metadata.exploration_index };
+  })
+};
+Write(`${sessionFolder}/explorations-manifest.json`, JSON.stringify(explorationManifest, null, 2));
+```
+
+### Step 3: Invoke Context-Search Agent
+
+**Only execute after Step 2 completes**
+
 ```javascript
 Task(
  subagent_type="context-search-agent",
@@ -97,6 +225,12 @@ Task(
 - **Task Description**: ${task_description}
 - **Output Path**: .workflow/${session_id}/.process/context-package.json

+## Exploration Input (from Step 2)
+- **Manifest**: ${sessionFolder}/explorations-manifest.json
+- **Exploration Count**: ${explorationManifest.exploration_count}
+- **Angles**: ${explorationManifest.angles_explored.join(', ')}
+- **Complexity**: ${complexity}
+
 ## Mission
 Execute complete context-search-agent workflow for implementation planning:

@@ -107,7 +241,8 @@ Execute complete context-search-agent workflow for implementation planning:
 4. **Analysis**: Extract keywords, determine scope, classify complexity based on task description and project state

 ### Phase 2: Multi-Source Context Discovery
-Execute all 4 discovery tracks:
+Execute all discovery tracks:
+- **Track 0**: Exploration Synthesis (load ${sessionFolder}/explorations-manifest.json, prioritize critical_files, deduplicate patterns/integration_points)
 - **Track 1**: Historical archive analysis (query manifest.json for lessons learned)
 - **Track 2**: Reference documentation (CLAUDE.md, architecture docs)
 - **Track 3**: Web examples (use Exa MCP for unfamiliar tech/APIs)
@@ -130,6 +265,7 @@ Complete context-package.json with:
 - **dependencies**: {internal[], external[]} with dependency graph
 - **brainstorm_artifacts**: {guidance_specification, role_analyses[], synthesis_output} with content
 - **conflict_detection**: {risk_level, risk_factors, affected_modules[], mitigation_strategy, historical_conflicts[]}
+- **exploration_results**: {manifest_path, exploration_count, angles, explorations[], aggregated_insights} (from Track 0)

 ## Quality Validation
 Before completion verify:
@@ -146,7 +282,7 @@ Report completion with statistics.
 )
 ```

-### Step 3: Output Verification
+### Step 4: Output Verification

 After agent completes, verify output:

@@ -156,6 +292,12 @@ const outputPath = `.workflow/${session_id}/.process/context-package.json`;
 if (!file_exists(outputPath)) {
  throw new Error("❌ Agent failed to generate context-package.json");
 }
+
+// Verify exploration_results included
+const pkg = JSON.parse(Read(outputPath));
+if (pkg.exploration_results?.exploration_count > 0) {
+  console.log(`✅ Exploration results aggregated: ${pkg.exploration_results.exploration_count} angles`);
+}
 ```

 ## Parameter Reference
@@ -176,6 +318,7 @@ Refer to `context-search-agent.md` Phase 3.7 for complete `context-package.json`
 - **dependencies**: Internal and external dependency graphs
 - **brainstorm_artifacts**: Brainstorm documents with full content (if exists)
 - **conflict_detection**: Risk assessment with mitigation strategies and historical conflicts
+- **exploration_results**: Aggregated exploration insights (from parallel explore phase)

 ## Historical Archive Analysis

--- a/.claude/skills/command-guide/reference/commands/workflow/tools/task-generate-agent.md
+++ b/.claude/skills/command-guide/reference/commands/workflow/tools/task-generate-agent.md
@@ -1,10 +1,9 @@
 ---
 name: task-generate-agent
 description: Generate implementation plan documents (IMPL_PLAN.md, task JSONs, TODO_LIST.md) using action-planning-agent - produces planning artifacts, does NOT execute code implementation
-argument-hint: "--session WFS-session-id [--cli-execute]"
+argument-hint: "--session WFS-session-id"
 examples:
  - /workflow:tools:task-generate-agent --session WFS-auth
-  - /workflow:tools:task-generate-agent --session WFS-auth --cli-execute
 ---

 # Generate Implementation Plan Command
@@ -26,7 +25,7 @@ Generate implementation planning documents (IMPL_PLAN.md, task JSONs, TODO_LIST.

 ```
 Input Parsing:
-   ├─ Parse flags: --session, --cli-execute
+   ├─ Parse flags: --session
   └─ Validation: session_id REQUIRED

 Phase 1: Context Preparation (Command)
@@ -65,9 +64,10 @@ Phase 2: Planning Document Generation (Agent)

 2. **Provide Metadata** (simple values):
   - `session_id`
-   - `execution_mode` (agent-mode | cli-execute-mode)
   - `mcp_capabilities` (available MCP tools)

+**Note**: CLI tool usage is now determined semantically by action-planning-agent based on user's task description, not by flags.
+
 ### Phase 2: Planning Document Generation (Agent Responsibility)

 **Purpose**: Generate IMPL_PLAN.md, task JSONs, and TODO_LIST.md - planning documents only, NOT code implementation.
@@ -97,15 +97,28 @@ Output:

 ## CONTEXT METADATA
 Session ID: {session-id}
-Planning Mode: {agent-mode | cli-execute-mode}
 MCP Capabilities: {exa_code, exa_web, code_index}

+## CLI TOOL SELECTION
+Determine CLI tool usage per-step based on user's task description:
+- If user specifies "use Codex/Gemini/Qwen for X" → Add command field to relevant steps
+- Default: Agent execution (no command field) unless user explicitly requests CLI
+
+## EXPLORATION CONTEXT (from context-package.exploration_results)
+- Load exploration_results from context-package.json
+- Use aggregated_insights.critical_files for focus_paths generation
+- Apply aggregated_insights.constraints to acceptance criteria
+- Reference aggregated_insights.all_patterns for implementation approach
+- Use aggregated_insights.all_integration_points for precise modification locations
+- Use conflict_indicators for risk-aware task sequencing
+
 ## 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
+   - **focus_paths enhanced with exploration critical_files**
+   - Flow control with pre_analysis steps (include exploration integration_points analysis)

 2. Implementation Plan (IMPL_PLAN.md)
   - Context analysis and artifact references
--- a/.claude/skills/command-guide/reference/commands/workflow/tools/task-generate-tdd.md
+++ b/.claude/skills/command-guide/reference/commands/workflow/tools/task-generate-tdd.md
@@ -1,24 +1,23 @@
 ---
 name: task-generate-tdd
 description: Autonomous TDD task generation using action-planning-agent with Red-Green-Refactor cycles, test-first structure, and cycle validation
-argument-hint: "--session WFS-session-id [--cli-execute]"
+argument-hint: "--session WFS-session-id"
 examples:
  - /workflow:tools:task-generate-tdd --session WFS-auth
-  - /workflow:tools:task-generate-tdd --session WFS-auth --cli-execute
 ---

 # Autonomous TDD Task Generation Command

 ## Overview
-Autonomous TDD task JSON and IMPL_PLAN.md generation using action-planning-agent with two-phase execution: discovery and document generation. Supports both agent-driven execution (default) and CLI tool execution modes. Generates complete Red-Green-Refactor cycles contained within each task.
+Autonomous TDD task JSON and IMPL_PLAN.md generation using action-planning-agent with two-phase execution: discovery and document generation. Generates complete Red-Green-Refactor cycles contained within each task.

 ## Core Philosophy
 - **Agent-Driven**: Delegate execution to action-planning-agent for autonomous operation
 - **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 TDD template based on `--cli-execute` flag **before** invoking agent
- **Agent Simplicity**: Agent receives pre-selected template and focuses only on content generation
+- **Semantic CLI Selection**: CLI tool usage determined from user's task description, not flags
+- **Agent Simplicity**: Agent generates content with semantic CLI detection
 - **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`)
 - **TDD-First**: Every feature starts with a failing test (Red phase)
 - **Feature-Complete Tasks**: Each task contains complete Red-Green-Refactor cycle
@@ -57,7 +56,7 @@ Autonomous TDD task JSON and IMPL_PLAN.md generation using action-planning-agent

 ```
 Input Parsing:
-   ├─ Parse flags: --session, --cli-execute
+   ├─ Parse flags: --session
   └─ Validation: session_id REQUIRED

 Phase 1: Discovery & Context Loading (Memory-First)
@@ -69,7 +68,7 @@ Phase 1: Discovery & Context Loading (Memory-First)
   └─ Optional: MCP external research

 Phase 2: Agent Execution (Document Generation)
-   ├─ Pre-agent template selection (agent-mode OR cli-execute-mode)
+   ├─ Pre-agent template selection (semantic CLI detection)
   ├─ Invoke action-planning-agent
   ├─ Generate TDD Task JSON Files (.task/IMPL-*.json)
   │  └─ Each task: complete Red-Green-Refactor cycle internally
@@ -86,11 +85,8 @@ Phase 2: Agent Execution (Document Generation)
 ```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
  "workflow_type": "tdd",
+  // Note: CLI tool usage is determined semantically by action-planning-agent based on user's task description
  "session_metadata": {
    // If in memory: use cached content
    // Else: Load from .workflow/active//{session-id}/workflow-session.json
@@ -199,8 +195,7 @@ Task(

 **Session ID**: WFS-{session-id}
 **Workflow Type**: TDD
-**Execution Mode**: {agent-mode | cli-execute-mode}
-**Task JSON Template Path**: {template_path}
+**Note**: CLI tool usage is determined semantically from user's task description

 ## Phase 1: Discovery Results (Provided Context)

@@ -265,16 +260,15 @@ Refer to: @.claude/agents/action-planning-agent.md for:

 ##### 1. TDD 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 with TDD-specific metadata
  - `meta.tdd_workflow`: true (REQUIRED)
  - `meta.max_iterations`: 3 (Green phase test-fix cycle limit)
-  - `meta.use_codex`: false (manual fixes by default)
  - `context.tdd_cycles`: Array with quantified test cases and coverage
  - `flow_control.implementation_approach`: Exactly 3 steps with `tdd_phase` field
    1. Red Phase (`tdd_phase: "red"`): Write failing tests
    2. Green Phase (`tdd_phase: "green"`): Implement to pass tests
    3. Refactor Phase (`tdd_phase: "refactor"`): Improve code quality
+  - CLI tool usage determined semantically (add `command` field when user requests CLI execution)
 - **Details**: See action-planning-agent.md § TDD Task JSON Generation

 ##### 2. IMPL_PLAN.md (TDD Variant)
@@ -475,16 +469,14 @@ This section provides quick reference for TDD task JSON structure. For complete

 **Basic Usage**:
 ```bash
-# Agent mode (default, autonomous execution)
+# Standard execution
 /workflow:tools:task-generate-tdd --session WFS-auth

-# CLI tool mode (use Gemini/Qwen for generation)
-/workflow:tools:task-generate-tdd --session WFS-auth --cli-execute
+# With semantic CLI request (include in task description)
+# e.g., "Generate TDD tasks for auth module, use Codex for implementation"
 ```

-**Execution Modes**:
- **Agent mode** (default): Uses `action-planning-agent` with agent-mode task template
- **CLI mode** (`--cli-execute`): Uses Gemini/Qwen with cli-mode task template
+**CLI Tool Selection**: Determined semantically from user's task description. Include "use Codex/Gemini/Qwen" in your request for CLI execution.

 **Output**:
 - TDD task JSON files in `.task/` directory (IMPL-N.json format)
@@ -513,7 +505,7 @@ IMPL (Green phase) tasks include automatic test-fix cycle:
 3. **Success Path**: Tests pass → Complete task
 4. **Failure Path**: Tests fail → Enter iterative fix cycle:
   - **Gemini Diagnosis**: Analyze failures with bug-fix template
-   - **Fix Application**: Manual (default) or Codex (if meta.use_codex=true)
+   - **Fix Application**: Agent (default) or CLI (if `command` field present)
   - **Retest**: Verify fix resolves failures
   - **Repeat**: Up to max_iterations (default: 3)
 5. **Safety Net**: Auto-revert all changes if max iterations reached
@@ -522,5 +514,5 @@ IMPL (Green phase) tasks include automatic test-fix cycle:

 ## Configuration Options
 - **meta.max_iterations**: Number of fix attempts (default: 3 for TDD, 5 for test-gen)
- **meta.use_codex**: Enable Codex automated fixes (default: false, manual)
+- **CLI tool usage**: Determined semantically from user's task description via `command` field in implementation_approach

--- a/.claude/skills/command-guide/reference/commands/workflow/tools/test-task-generate.md
+++ b/.claude/skills/command-guide/reference/commands/workflow/tools/test-task-generate.md
@@ -1,11 +1,9 @@
 ---
 name: test-task-generate
 description: Generate test planning documents (IMPL_PLAN.md, test task JSONs, TODO_LIST.md) using action-planning-agent - produces test planning artifacts, does NOT execute tests
-argument-hint: "[--use-codex] [--cli-execute] --session WFS-test-session-id"
+argument-hint: "--session WFS-test-session-id"
 examples:
  - /workflow:tools:test-task-generate --session WFS-test-auth
-  - /workflow:tools:test-task-generate --use-codex --session WFS-test-auth
-  - /workflow:tools:test-task-generate --cli-execute --session WFS-test-auth
 ---

 # Generate Test Planning Documents Command
@@ -26,17 +24,17 @@ Generate test planning documents (IMPL_PLAN.md, test task JSONs, TODO_LIST.md) u

 ### Test Generation (IMPL-001)
 - **Agent Mode** (default): @code-developer generates tests within agent context
- **CLI Execute Mode** (`--cli-execute`): Use Codex CLI for autonomous test generation
+- **CLI Mode**: Use CLI tools when `command` field present in implementation_approach (determined semantically)

 ### Test Execution & Fix (IMPL-002+)
- **Manual Mode** (default): Gemini diagnosis → user applies fixes
- **Codex Mode** (`--use-codex`): Gemini diagnosis → Codex applies fixes with resume mechanism
+- **Agent Mode** (default): Gemini diagnosis → agent applies fixes
+- **CLI Mode**: Gemini diagnosis → CLI applies fixes (when `command` field present in implementation_approach)

 ## Execution Process

 ```
 Input Parsing:
-   ├─ Parse flags: --session, --use-codex, --cli-execute
+   ├─ Parse flags: --session
   └─ Validation: session_id REQUIRED

 Phase 1: Context Preparation (Command)
@@ -44,7 +42,7 @@ Phase 1: Context Preparation (Command)
   │  ├─ session_metadata_path
   │  ├─ test_analysis_results_path (REQUIRED)
   │  └─ test_context_package_path
-   └─ Provide metadata (session_id, execution_mode, use_codex, source_session_id)
+   └─ Provide metadata (session_id, source_session_id)

 Phase 2: Test Document Generation (Agent)
   ├─ Load TEST_ANALYSIS_RESULTS.md as primary requirements source
@@ -83,11 +81,11 @@ Phase 2: Test Document Generation (Agent)

 2. **Provide Metadata** (simple values):
   - `session_id`
-   - `execution_mode` (agent-mode | cli-execute-mode)
-   - `use_codex` flag (true | false)
   - `source_session_id` (if exists)
   - `mcp_capabilities` (available MCP tools)

+**Note**: CLI tool usage is now determined semantically from user's task description, not by flags.
+
 ### Phase 2: Test Document Generation (Agent Responsibility)

 **Purpose**: Generate test-specific IMPL_PLAN.md, task JSONs, and TODO_LIST.md - planning documents only, NOT test execution.
@@ -134,11 +132,14 @@ Output:
 ## CONTEXT METADATA
 Session ID: {test-session-id}
 Workflow Type: test_session
-Planning Mode: {agent-mode | cli-execute-mode}
-Use Codex: {true | false}
 Source Session: {source-session-id} (if exists)
 MCP Capabilities: {exa_code, exa_web, code_index}

+## CLI TOOL SELECTION
+Determine CLI tool usage per-step based on user's task description:
+- If user specifies "use Codex/Gemini/Qwen for X" → Add command field to relevant steps
+- Default: Agent execution (no command field) unless user explicitly requests CLI
+
 ## TEST-SPECIFIC REQUIREMENTS SUMMARY
 (Detailed specifications in your agent definition)

@@ -149,25 +150,26 @@ MCP Capabilities: {exa_code, exa_web, code_index}
 Task Configuration:
  IMPL-001 (Test Generation):
    - meta.type: "test-gen"
-    - meta.agent: "@code-developer" (agent-mode) OR CLI execution (cli-execute-mode)
+    - meta.agent: "@code-developer"
    - meta.test_framework: Specify existing framework (e.g., "jest", "vitest", "pytest")
    - flow_control: Test generation strategy from TEST_ANALYSIS_RESULTS.md
+    - CLI execution: Add `command` field when user requests (determined semantically)

  IMPL-002+ (Test Execution & Fix):
    - meta.type: "test-fix"
    - meta.agent: "@test-fix-agent"
-    - meta.use_codex: true/false (based on flag)
    - flow_control: Test-fix cycle with iteration limits and diagnosis configuration
+    - CLI execution: Add `command` field when user requests (determined semantically)

 ### Test-Fix Cycle Specification (IMPL-002+)
 Required flow_control fields:
  - max_iterations: 5
  - diagnosis_tool: "gemini"
  - diagnosis_template: "~/.claude/workflows/cli-templates/prompts/analysis/01-diagnose-bug-root-cause.txt"
-  - fix_mode: "manual" OR "codex" (based on use_codex flag)
  - cycle_pattern: "test → gemini_diagnose → fix → retest"
  - exit_conditions: ["all_tests_pass", "max_iterations_reached"]
  - auto_revert_on_failure: true
+  - CLI fix: Add `command` field when user specifies CLI tool usage

 ### Automation Framework Configuration
 Select automation tools based on test requirements from TEST_ANALYSIS_RESULTS.md:
@@ -191,8 +193,9 @@ PRIMARY requirements source - extract and map to task JSONs:
 ## EXPECTED DELIVERABLES
 1. Test Task JSON Files (.task/IMPL-*.json)
   - 6-field schema with quantified requirements from TEST_ANALYSIS_RESULTS.md
-   - Test-specific metadata: type, agent, use_codex, test_framework, coverage_target
+   - Test-specific metadata: type, agent, test_framework, coverage_target
   - flow_control includes: reusable_test_tools, test_commands (from project config)
+   - CLI execution via `command` field when user requests (determined semantically)
   - Artifact references from test-context-package.json
   - Absolute paths in context.files_to_test

@@ -213,9 +216,9 @@ Hard Constraints:
  - All requirements quantified from TEST_ANALYSIS_RESULTS.md
  - Test framework matches existing project framework
  - flow_control includes reusable_test_tools and test_commands from project
-  - use_codex flag correctly set in IMPL-002+ tasks
  - Absolute paths for all focus_paths
  - Acceptance criteria include verification commands
+  - CLI `command` field added only when user explicitly requests CLI tool usage

 ## SUCCESS CRITERIA
 - All test planning documents generated successfully
@@ -233,21 +236,18 @@ Hard Constraints:

 ### Usage Examples
 ```bash
-# Agent mode (default)
+# Standard execution
 /workflow:tools:test-task-generate --session WFS-test-auth

-# With automated Codex fixes
-/workflow:tools:test-task-generate --use-codex --session WFS-test-auth
-
-# CLI execution mode for test generation
-/workflow:tools:test-task-generate --cli-execute --session WFS-test-auth
+# With semantic CLI request (include in task description)
+# e.g., "Generate tests, use Codex for implementation and fixes"
 ```

-### Flag Behavior
- **No flags**: `meta.use_codex=false` (manual fixes), agent-mode test generation
- **--use-codex**: `meta.use_codex=true` (Codex automated fixes in IMPL-002+)
- **--cli-execute**: CLI tool execution mode for IMPL-001 test generation
- **Both flags**: CLI generation + automated Codex fixes
+### CLI Tool Selection
+CLI tool usage is determined semantically from user's task description:
+- Include "use Codex" for automated fixes
+- Include "use Gemini" for analysis
+- Default: Agent execution (no `command` field)

 ### Output
 - Test task JSON files in `.task/` directory (minimum 2)