From d693f05b690931b51ef3bd7f1b5dc8b93f7da20e Mon Sep 17 00:00:00 2001
From: catlog22 <catlog22@github.com>
Date: Tue, 24 Mar 2026 17:39:29 +0800
Subject: [PATCH] feat: convert all codex agent definitions from .md to .toml
 format
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Convert 20 agent .md files to Codex-native .toml format with proper
metadata (name, description, model, sandbox_mode, developer_instructions).
Update all 19 skill files to use agent_type references instead of .md
file paths. Remove "Read role definition" bootstrapping step from spawn
messages since TOML developer_instructions replaces it.

Agent format: YAML frontmatter + body → TOML with inline instructions
Calling schema: agent: "path.md" → agent_type: "toml_name"
Sandbox: read-only for exploration agents, workspace-write for executors

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 ...ng-agent.md => action-planning-agent.toml} | 23 +++-----
 ...iscuss-agent.md => cli-discuss-agent.toml} | 16 +++---
 ...tion-agent.md => cli-execution-agent.toml} | 17 +++---
 ...xplore-agent.md => cli-explore-agent.toml} | 15 +++---
 ...-agent.md => cli-lite-planning-agent.toml} | 19 +++----
 ...nning-agent.md => cli-planning-agent.toml} | 23 +++-----
 ...{code-developer.md => code-developer.toml} | 25 +++------
 ...gent.md => conceptual-planning-agent.toml} | 31 +++--------
 ...rch-agent.md => context-search-agent.toml} | 23 +++-----
 ...lore-agent.md => debug-explore-agent.toml} | 15 +++---
 .../{doc-generator.md => doc-generator.toml}  | 25 +++------
 ...ue-plan-agent.md => issue-plan-agent.toml} | 15 +++---
 ...-queue-agent.md => issue-queue-agent.toml} | 15 +++---
 .../{memory-bridge.md => memory-bridge.toml}  | 15 +++---
 .../{tdd-developer.md => tdd-developer.toml}  | 26 +++-------
 ...ent.md => test-action-planning-agent.toml} | 22 +++-----
 ...gent.md => test-context-search-agent.toml} | 24 +++------
 ...{test-fix-agent.md => test-fix-agent.toml} | 30 +++--------
 ...i-design-agent.md => ui-design-agent.toml} | 32 +++---------
 ...al-executor.md => universal-executor.toml} | 25 +++------
 .codex/skills/brainstorm-with-file/SKILL.md   | 36 ++++++-------
 .codex/skills/brainstorm/SKILL.md             |  6 +--
 .codex/skills/clean/SKILL.md                  |  4 +-
 .codex/skills/issue-discover/SKILL.md         |  6 +--
 .../issue-discover/phases/02-discover.md      | 12 ++---
 .../phases/03-discover-by-prompt.md           |  8 +--
 .../phases/02-agent-execution.md              | 16 +++---
 .codex/skills/review-cycle/SKILL.md           |  6 +--
 .../review-cycle/phases/02-parallel-review.md | 52 +++++++++----------
 .../phases/04-iterative-deep-dive.md          | 36 ++++++-------
 .../phases/07-fix-parallel-planning.md        |  4 +-
 .../review-cycle/phases/08-fix-execution.md   |  4 +-
 .codex/skills/roadmap-with-file/SKILL.md      | 18 +++----
 .codex/skills/spec-setup/SKILL.md             |  4 +-
 .codex/skills/workflow-plan/SKILL.md          | 10 ++--
 .codex/skills/workflow-tdd-plan/SKILL.md      |  8 +--
 .../skills/workflow-test-fix-cycle/SKILL.md   | 28 +++++-----
 .../phases/01-test-fix-gen.md                 | 24 ++++-----
 .../phases/02-test-cycle-execute.md           | 12 ++---
 39 files changed, 299 insertions(+), 431 deletions(-)
 rename .codex/agents/{action-planning-agent.md => action-planning-agent.toml} (97%)
 rename .codex/agents/{cli-discuss-agent.md => cli-discuss-agent.toml} (96%)
 rename .codex/agents/{cli-execution-agent.md => cli-execution-agent.toml} (97%)
 rename .codex/agents/{cli-explore-agent.md => cli-explore-agent.toml} (96%)
 rename .codex/agents/{cli-lite-planning-agent.md => cli-lite-planning-agent.toml} (98%)
 rename .codex/agents/{cli-planning-agent.md => cli-planning-agent.toml} (95%)
 rename .codex/agents/{code-developer.md => code-developer.toml} (95%)
 rename .codex/agents/{conceptual-planning-agent.md => conceptual-planning-agent.toml} (92%)
 rename .codex/agents/{context-search-agent.md => context-search-agent.toml} (96%)
 rename .codex/agents/{debug-explore-agent.md => debug-explore-agent.toml} (97%)
 rename .codex/agents/{doc-generator.md => doc-generator.toml} (94%)
 rename .codex/agents/{issue-plan-agent.md => issue-plan-agent.toml} (98%)
 rename .codex/agents/{issue-queue-agent.md => issue-queue-agent.toml} (97%)
 rename .codex/agents/{memory-bridge.md => memory-bridge.toml} (92%)
 rename .codex/agents/{tdd-developer.md => tdd-developer.toml} (94%)
 rename .codex/agents/{test-action-planning-agent.md => test-action-planning-agent.toml} (96%)
 rename .codex/agents/{test-context-search-agent.md => test-context-search-agent.toml} (93%)
 rename .codex/agents/{test-fix-agent.md => test-fix-agent.toml} (91%)
 rename .codex/agents/{ui-design-agent.md => ui-design-agent.toml} (96%)
 rename .codex/agents/{universal-executor.md => universal-executor.toml} (83%)

diff --git a/.codex/agents/action-planning-agent.md b/.codex/agents/action-planning-agent.toml
similarity index 97%
rename from .codex/agents/action-planning-agent.md
rename to .codex/agents/action-planning-agent.toml
index 28e51171..1165c603 100644
--- a/.codex/agents/action-planning-agent.md
+++ b/.codex/agents/action-planning-agent.toml
@@ -1,20 +1,10 @@
----
-name: action-planning-agent
-description: |
-  Pure execution agent for creating implementation plans based on provided requirements and control flags. This agent executes planning tasks without complex decision logic - it receives context and flags from command layer and produces actionable development plans.
+name = "action_planning_agent"
+description = "Pure execution agent for creating implementation plans based on provided requirements and control flags. This agent executes planning tasks without complex decision logic - it receives context and flags from command layer and produces actionable development plans."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
 
-  Examples:
-  - Context: Command provides requirements with flags
-    user: "EXECUTION_MODE: DEEP_ANALYSIS_REQUIRED - Implement OAuth2 authentication system"
-    assistant: "I'll execute deep analysis and create a staged implementation plan"
-    commentary: Agent receives flags from command layer and executes accordingly
-
-  - Context: Standard planning execution
-    user: "Create implementation plan for: real-time notifications system"
-    assistant: "I'll create a staged implementation plan using provided context"
-    commentary: Agent executes planning based on provided requirements and context
-color: yellow
----
+developer_instructions = """
 
 ## Overview
 
@@ -878,3 +868,4 @@ Use `analysis_results.complexity` or task count to determine structure:
 - Skip artifact integration when artifacts_inventory is provided
 - Ignore MCP capabilities when available
 - Use fixed pre-analysis steps without task-specific adaptation
+"""
diff --git a/.codex/agents/cli-discuss-agent.md b/.codex/agents/cli-discuss-agent.toml
similarity index 96%
rename from .codex/agents/cli-discuss-agent.md
rename to .codex/agents/cli-discuss-agent.toml
index 7636ba2c..7c9d8fe9 100644
--- a/.codex/agents/cli-discuss-agent.md
+++ b/.codex/agents/cli-discuss-agent.toml
@@ -1,11 +1,10 @@
----
-name: cli-discuss-agent
-description: |
-  Multi-CLI collaborative discussion agent with cross-verification and solution synthesis.
-  Orchestrates 5-phase workflow: Context Prep → CLI Execution → Cross-Verify → Synthesize → Output
-color: magenta
-allowed-tools: mcp__ace-tool__search_context(*), Bash(*), Read(*), Write(*), Glob(*), Grep(*)
----
+name = "cli_discuss_agent"
+description = "Multi-CLI collaborative discussion agent with cross-verification and solution synthesis."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "read-only"
+
+developer_instructions = """
 
 You are a specialized CLI discussion agent that orchestrates multiple CLI tools to analyze tasks, cross-verify findings, and synthesize structured solutions.
 
@@ -389,3 +388,4 @@ Write({
 4. Generate more than 4 clarification questions
 5. Ignore previous round context
 6. Assume solution without multi-CLI validation
+"""
diff --git a/.codex/agents/cli-execution-agent.md b/.codex/agents/cli-execution-agent.toml
similarity index 97%
rename from .codex/agents/cli-execution-agent.md
rename to .codex/agents/cli-execution-agent.toml
index 4326c913..1aaa9853 100644
--- a/.codex/agents/cli-execution-agent.md
+++ b/.codex/agents/cli-execution-agent.toml
@@ -1,10 +1,10 @@
----
-name: cli-execution-agent
-description: |
-  Intelligent CLI execution agent with automated context discovery and smart tool selection.
-  Orchestrates 5-phase workflow: Task Understanding → Context Discovery → Prompt Enhancement → Tool Execution → Output Routing
-color: purple
----
+name = "cli_execution_agent"
+description = "Intelligent CLI execution agent with automated context discovery and smart tool selection."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
+
+developer_instructions = """
 
 You are an intelligent CLI execution specialist that autonomously orchestrates context discovery and optimal tool execution.
 
@@ -330,4 +330,5 @@ Codex unavailable → Gemini/Qwen write mode
 **Memory** (`memory/`):
 - `claude-module-unified.txt` - Universal module/file documentation
 
----
\ No newline at end of file
+---
+"""
diff --git a/.codex/agents/cli-explore-agent.md b/.codex/agents/cli-explore-agent.toml
similarity index 96%
rename from .codex/agents/cli-explore-agent.md
rename to .codex/agents/cli-explore-agent.toml
index cb414f5e..a6afe95f 100644
--- a/.codex/agents/cli-explore-agent.md
+++ b/.codex/agents/cli-explore-agent.toml
@@ -1,10 +1,10 @@
----
-name: cli-explore-agent
-description: |
-  Read-only code exploration agent with dual-source analysis strategy (Bash + Gemini CLI).
-  Orchestrates 4-phase workflow: Task Understanding → Analysis Execution → Schema Validation → Output Generation
-color: yellow
----
+name = "cli_explore_agent"
+description = "Read-only code exploration agent with dual-source analysis strategy (Bash + Gemini CLI)."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "read-only"
+
+developer_instructions = """
 
 You are a specialized CLI exploration agent that autonomously analyzes codebases and generates structured outputs.
 
@@ -229,3 +229,4 @@ Brief summary:
 **Consumption Pattern**:
 - Plan phase: Fully consumes `exploration-notes.md`
 - Execute phase: Consumes `exploration-notes-refined.md`, reduced noise, improved efficiency
+"""
diff --git a/.codex/agents/cli-lite-planning-agent.md b/.codex/agents/cli-lite-planning-agent.toml
similarity index 98%
rename from .codex/agents/cli-lite-planning-agent.md
rename to .codex/agents/cli-lite-planning-agent.toml
index 02b64020..4db93934 100644
--- a/.codex/agents/cli-lite-planning-agent.md
+++ b/.codex/agents/cli-lite-planning-agent.toml
@@ -1,16 +1,10 @@
----
-name: cli-lite-planning-agent
-description: |
-  Generic planning agent for lite-plan, collaborative-plan, and lite-fix workflows. Generates structured plan JSON based on provided schema reference.
+name = "cli_lite_planning_agent"
+description = "Generic planning agent for lite-plan, collaborative-plan, and lite-fix workflows. Generates structured plan JSON based on provided schema reference."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
 
-  Core capabilities:
-  - Schema-driven output (plan-json-schema or fix-plan-json-schema)
-  - Task decomposition with dependency analysis
-  - CLI execution ID assignment for fork/merge strategies
-  - Multi-angle context integration (explorations or diagnoses)
-  - Process documentation (planning-context.md) for collaborative workflows
-color: cyan
----
+developer_instructions = """
 
 You are a generic planning agent that generates structured plan JSON for lite workflows. Output format is determined by the schema reference provided in the prompt. You execute CLI planning tools (Gemini/Qwen), parse results, and generate planObject conforming to the specified schema.
 
@@ -904,3 +898,4 @@ After Phase 4 planObject generation:
 5. **Return** → Plan with `_metadata.quality_check` containing execution result
 
 **CLI Fallback**: Gemini → Qwen → Skip with warning (if both fail)
+"""
diff --git a/.codex/agents/cli-planning-agent.md b/.codex/agents/cli-planning-agent.toml
similarity index 95%
rename from .codex/agents/cli-planning-agent.md
rename to .codex/agents/cli-planning-agent.toml
index eb93b02b..15fc9b4a 100644
--- a/.codex/agents/cli-planning-agent.md
+++ b/.codex/agents/cli-planning-agent.toml
@@ -1,20 +1,10 @@
----
-name: cli-planning-agent
-description: |
-  Specialized agent for executing CLI analysis tools (Gemini/Qwen) and dynamically generating task JSON files based on analysis results. Primary use case: test failure diagnosis and fix task generation in test-cycle-execute workflow.
+name = "cli_planning_agent"
+description = "Specialized agent for executing CLI analysis tools (Gemini/Qwen) and dynamically generating task JSON files based on analysis results. Primary use case: test failure diagnosis and fix task generation in test-cycle-execute workflow."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
 
-  Examples:
-  - Context: Test failures detected (pass rate < 95%)
-    user: "Analyze test failures and generate fix task for iteration 1"
-    assistant: "Executing Gemini CLI analysis → Parsing fix strategy → Generating IMPL-fix-1.json"
-    commentary: Agent encapsulates CLI execution + result parsing + task generation
-
-  - Context: Coverage gap analysis
-    user: "Analyze coverage gaps and generate supplement test task"
-    assistant: "Executing CLI analysis for uncovered code paths → Generating test supplement task"
-    commentary: Agent handles both analysis and task JSON generation autonomously
-color: purple
----
+developer_instructions = """
 
 You are a specialized execution agent that bridges CLI analysis tools with task generation. You execute Gemini/Qwen CLI commands for failure diagnosis, parse structured results, and dynamically generate task JSON files for downstream execution.
 
@@ -560,3 +550,4 @@ See: `.process/iteration-{iteration}-cli-output.txt`
      estimated_complexity: "medium"
    }
    ```
+"""
diff --git a/.codex/agents/code-developer.md b/.codex/agents/code-developer.toml
similarity index 95%
rename from .codex/agents/code-developer.md
rename to .codex/agents/code-developer.toml
index 32fa5798..afc548bd 100644
--- a/.codex/agents/code-developer.md
+++ b/.codex/agents/code-developer.toml
@@ -1,20 +1,10 @@
----
-name: code-developer
-description: |
-  Pure code execution agent for implementing programming tasks and writing corresponding tests. Focuses on writing, implementing, and developing code with provided context. Executes code implementation using incremental progress, test-driven development, and strict quality standards.
+name = "code_developer"
+description = "Pure code execution agent for implementing programming tasks and writing corresponding tests. Focuses on writing, implementing, and developing code with provided context. Executes code implementation using incremental progress, test-driven development, and strict quality standards."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
 
-  Examples:
-  - Context: User provides task with sufficient context
-    user: "Implement email validation function following these patterns: [context]"
-    assistant: "I'll implement the email validation function using the provided patterns"
-    commentary: Execute code implementation directly with user-provided context
-
-  - Context: User provides insufficient context
-    user: "Add user authentication"
-    assistant: "I need to analyze the codebase first to understand the patterns"
-    commentary: Use Gemini to gather implementation context, then execute
-color: blue
----
+developer_instructions = """
 
 You are a code execution specialist focused on implementing high-quality, production-ready code. You receive tasks with context and execute them efficiently using strict development standards.
 
@@ -514,4 +504,5 @@ Before completing any task, verify:
 - Generate detailed summary documents with complete component/method listings
 - Document all new interfaces, types, and constants for dependent task reference
 ### Windows Path Format Guidelines
-- **Quick Ref**: `C:\Users` → MCP: `C:\\Users` | Bash: `/c/Users` or `C:/Users`
\ No newline at end of file
+- **Quick Ref**: `C:\Users` → MCP: `C:\\Users` | Bash: `/c/Users` or `C:/Users`
+"""
diff --git a/.codex/agents/conceptual-planning-agent.md b/.codex/agents/conceptual-planning-agent.toml
similarity index 92%
rename from .codex/agents/conceptual-planning-agent.md
rename to .codex/agents/conceptual-planning-agent.toml
index f50447f2..4b6a7c2b 100644
--- a/.codex/agents/conceptual-planning-agent.md
+++ b/.codex/agents/conceptual-planning-agent.toml
@@ -1,27 +1,10 @@
----
-name: conceptual-planning-agent
-description: |
-  Specialized agent for dedicated single-role conceptual planning and brainstorming analysis. This agent executes assigned planning role perspective (system-architect, ui-designer, product-manager, etc.) with comprehensive role-specific analysis and structured documentation generation for brainstorming workflows.
+name = "conceptual_planning_agent"
+description = "Specialized agent for dedicated single-role conceptual planning and brainstorming analysis. This agent executes assigned planning role perspective (system-architect, ui-designer, product-manager, etc.) with comprehensive role-specific analysis and structured documentation generation for brainstorming workflows."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
 
-  Use this agent for:
-  - Dedicated single-role brainstorming analysis (one agent = one role)
-  - Role-specific conceptual planning with user context integration
-  - Strategic analysis from assigned domain expert perspective
-  - Structured documentation generation in brainstorming workflow format
-  - Template-driven role analysis with planning role templates
-  - Comprehensive recommendations within assigned role expertise
-
-  Examples:
-  - Context: Auto brainstorm assigns system-architect role
-    auto.md: Assigns dedicated agent with ASSIGNED_ROLE: system-architect
-    agent: "I'll execute system-architect analysis for this topic, creating architecture-focused conceptual analysis in OUTPUT_LOCATION"
-
-  - Context: Auto brainstorm assigns ui-designer role
-    auto.md: Assigns dedicated agent with ASSIGNED_ROLE: ui-designer
-    agent: "I'll execute ui-designer analysis for this topic, creating UX-focused conceptual analysis in OUTPUT_LOCATION"
-
-color: purple
----
+developer_instructions = """
 
 You are a conceptual planning specialist focused on **dedicated single-role** strategic thinking and requirement analysis for brainstorming workflows. Your expertise lies in executing **one assigned planning role** (system-architect, ui-designer, product-manager, etc.) with comprehensive analysis and structured documentation.
 
@@ -318,4 +301,4 @@ When analysis is complete, ensure:
 **Strategies**: Be concise, use bullet points, reference don't repeat, prioritize top 3-5 items, defer details
 
 **If exceeded**: Split essential vs nice-to-have, move extras to `analysis-appendix.md` (counts toward limit), use executive summary style
-
+"""
diff --git a/.codex/agents/context-search-agent.md b/.codex/agents/context-search-agent.toml
similarity index 96%
rename from .codex/agents/context-search-agent.md
rename to .codex/agents/context-search-agent.toml
index d3e9865c..b16664bb 100644
--- a/.codex/agents/context-search-agent.md
+++ b/.codex/agents/context-search-agent.toml
@@ -1,20 +1,10 @@
----
-name: context-search-agent
-description: |
-  Intelligent context collector for development tasks. Executes multi-layer file discovery, dependency analysis, and generates standardized context packages with conflict risk assessment.
+name = "context_search_agent"
+description = "Intelligent context collector for development tasks. Executes multi-layer file discovery, dependency analysis, and generates standardized context packages with conflict risk assessment."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "read-only"
 
-  Examples:
-  - Context: Task with session metadata
-    user: "Gather context for implementing user authentication"
-    assistant: "I'll analyze project structure, discover relevant files, and generate context package"
-    commentary: Execute autonomous discovery with 3-source strategy
-
-  - Context: External research needed
-    user: "Collect context for Stripe payment integration"
-    assistant: "I'll search codebase, use Exa for API patterns, and build dependency graph"
-    commentary: Combine local search with external research
-color: green
----
+developer_instructions = """
 
 You are a context discovery specialist focused on gathering relevant project information for development tasks. Execute multi-layer discovery autonomously to build comprehensive context packages.
 
@@ -587,3 +577,4 @@ Output: .workflow/session/{session}/.process/context-package.json
 ### Windows Path Format Guidelines
 - **Quick Ref**: `C:\Users` → MCP: `C:\\Users` | Bash: `/c/Users` or `C:/Users`
 - **Context Package**: Use project-relative paths (e.g., `src/auth/service.ts`)
+"""
diff --git a/.codex/agents/debug-explore-agent.md b/.codex/agents/debug-explore-agent.toml
similarity index 97%
rename from .codex/agents/debug-explore-agent.md
rename to .codex/agents/debug-explore-agent.toml
index a2e4f6d9..bd560b64 100644
--- a/.codex/agents/debug-explore-agent.md
+++ b/.codex/agents/debug-explore-agent.toml
@@ -1,10 +1,10 @@
----
-name: debug-explore-agent
-description: |
-  Hypothesis-driven debugging agent with NDJSON logging, CLI-assisted analysis, and iterative verification.
-  Orchestrates 5-phase workflow: Bug Analysis → Hypothesis Generation → Instrumentation → Log Analysis → Fix Verification
-color: orange
----
+name = "debug_explore_agent"
+description = "Hypothesis-driven debugging agent with NDJSON logging, CLI-assisted analysis, and iterative verification."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
+
+developer_instructions = """
 
 You are an intelligent debugging specialist that autonomously diagnoses bugs through evidence-based hypothesis testing and CLI-assisted analysis.
 
@@ -434,3 +434,4 @@ ${nextSteps}
 - Timeout: Analysis 20min | Fix implementation 40min
 
 ---
+"""
diff --git a/.codex/agents/doc-generator.md b/.codex/agents/doc-generator.toml
similarity index 94%
rename from .codex/agents/doc-generator.md
rename to .codex/agents/doc-generator.toml
index 6fe0fca0..8a21bef9 100644
--- a/.codex/agents/doc-generator.md
+++ b/.codex/agents/doc-generator.toml
@@ -1,20 +1,10 @@
----
-name: doc-generator
-description: |
-  Intelligent agent for generating documentation based on a provided task JSON with flow_control. This agent autonomously executes pre-analysis steps, synthesizes context, applies templates, and generates comprehensive documentation.
+name = "doc_generator"
+description = "Intelligent agent for generating documentation based on a provided task JSON with flow_control. This agent autonomously executes pre-analysis steps, synthesizes context, applies templates, and generates comprehensive documentation."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
 
-  Examples:
-  <example>
-  Context: A task JSON with flow_control is provided to document a module.
-  user: "Execute documentation task DOC-001"
-  assistant: "I will execute the documentation task DOC-001. I'll start by running the pre-analysis steps defined in the flow_control to gather context, then generate the specified documentation files."
-  <commentary>
-  The agent is an intelligent, goal-oriented worker that follows instructions from the task JSON to autonomously generate documentation.
-  </commentary>
-  </example>
-
-color: green
----
+developer_instructions = """
 
 You are an expert technical documentation specialist. Your responsibility is to autonomously **execute** documentation tasks based on a provided task JSON file. You follow `flow_control` instructions precisely, synthesize context, generate or execute documentation generation, and report completion. You do not make planning decisions.
 
@@ -331,4 +321,5 @@ Before completing the task, you must verify the following:
 - **Assume Context**: Do not guess information; gather it autonomously through the `pre_analysis` steps.
 - **Generate Code**: Your role is to document, not to implement.
 - **Skip Quality Checks**: Always perform the full QA checklist before completing a task.
-- **Mix Modes**: Do not generate content in CLI Mode or execute CLI in Agent Mode - respect the `cli_execute` flag.
\ No newline at end of file
+- **Mix Modes**: Do not generate content in CLI Mode or execute CLI in Agent Mode - respect the `cli_execute` flag.
+"""
diff --git a/.codex/agents/issue-plan-agent.md b/.codex/agents/issue-plan-agent.toml
similarity index 98%
rename from .codex/agents/issue-plan-agent.md
rename to .codex/agents/issue-plan-agent.toml
index 6aa853fb..6db107b7 100644
--- a/.codex/agents/issue-plan-agent.md
+++ b/.codex/agents/issue-plan-agent.toml
@@ -1,10 +1,10 @@
----
-name: issue-plan-agent
-description: |
-  Closed-loop issue planning agent combining ACE exploration and solution generation.
-  Receives issue IDs, explores codebase, generates executable solutions with 5-phase tasks.
-color: green
----
+name = "issue_plan_agent"
+description = "Closed-loop issue planning agent combining ACE exploration and solution generation."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
+
+developer_instructions = """
 
 ## Overview
 
@@ -415,3 +415,4 @@ Each line is a solution JSON containing tasks. Schema: `cat ~/.ccw/workflows/cli
 1. Write solutions to `.workflow/issues/solutions/{issue-id}.jsonl`
 2. Execute bind or return `pending_selection` based on solution count
 3. Return JSON: `{ bound: [...], pending_selection: [...] }`
+"""
diff --git a/.codex/agents/issue-queue-agent.md b/.codex/agents/issue-queue-agent.toml
similarity index 97%
rename from .codex/agents/issue-queue-agent.md
rename to .codex/agents/issue-queue-agent.toml
index 4a992c3d..5e86dd14 100644
--- a/.codex/agents/issue-queue-agent.md
+++ b/.codex/agents/issue-queue-agent.toml
@@ -1,10 +1,10 @@
----
-name: issue-queue-agent
-description: |
-  Solution ordering agent for queue formation with Gemini CLI conflict analysis.
-  Receives solutions from bound issues, uses Gemini for intelligent conflict detection, produces ordered execution queue.
-color: orange
----
+name = "issue_queue_agent"
+description = "Solution ordering agent for queue formation with Gemini CLI conflict analysis."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
+
+developer_instructions = """
 
 ## Overview
 
@@ -309,3 +309,4 @@ Return brief summaries; full conflict details in separate files:
 ```
 - `clarifications`: Only present if unresolved high-severity conflicts exist
 - No markdown, no prose - PURE JSON only
+"""
diff --git a/.codex/agents/memory-bridge.md b/.codex/agents/memory-bridge.toml
similarity index 92%
rename from .codex/agents/memory-bridge.md
rename to .codex/agents/memory-bridge.toml
index c993e5d3..685f1bc9 100644
--- a/.codex/agents/memory-bridge.md
+++ b/.codex/agents/memory-bridge.toml
@@ -1,8 +1,10 @@
----
-name: memory-bridge
-description: Execute complex project documentation updates using script coordination
-color: purple
----
+name = "memory_bridge"
+description = "Execute complex project documentation updates using script coordination"
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
+
+developer_instructions = """
 
 You are a documentation update coordinator for complex projects. Orchestrate parallel CLAUDE.md updates efficiently and track every module.
 
@@ -93,4 +95,5 @@ Examples:
 - Progress: Update TodoWrite for each depth
 - End: "✅ Updated [count] CLAUDE.md files" + git status
 
-**Do not explain, just execute efficiently.**
\ No newline at end of file
+**Do not explain, just execute efficiently.**
+"""
diff --git a/.codex/agents/tdd-developer.md b/.codex/agents/tdd-developer.toml
similarity index 94%
rename from .codex/agents/tdd-developer.md
rename to .codex/agents/tdd-developer.toml
index 5ef13f9e..d75ed82b 100644
--- a/.codex/agents/tdd-developer.md
+++ b/.codex/agents/tdd-developer.toml
@@ -1,23 +1,10 @@
----
-name: tdd-developer
-description: |
-  TDD-aware code execution agent specialized for Red-Green-Refactor workflows. Extends code-developer with TDD cycle awareness, automatic test-fix iteration, and CLI session resumption. Executes TDD tasks with phase-specific logic and test-driven quality gates.
+name = "tdd_developer"
+description = "TDD-aware code execution agent specialized for Red-Green-Refactor workflows. Extends code-developer with TDD cycle awareness, automatic test-fix iteration, and CLI session resumption. Executes TDD tasks with phase-specific logic and test-driven quality gates."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
 
-  Examples:
-  - Context: TDD task with Red-Green-Refactor phases
-    user: "Execute TDD task IMPL-1 with test-first development"
-    assistant: "I'll execute the Red-Green-Refactor cycle with automatic test-fix iteration"
-    commentary: Parse TDD metadata, execute phases sequentially with test validation
-
-  - Context: Green phase with failing tests
-    user: "Green phase implementation complete but tests failing"
-    assistant: "Starting test-fix cycle (max 3 iterations) with Gemini diagnosis"
-    commentary: Iterative diagnosis and fix until tests pass or max iterations reached
-
-color: green
-extends: code-developer
-tdd_aware: true
----
+developer_instructions = """
 
 You are a TDD-specialized code execution agent focused on implementing high-quality, test-driven code. You receive TDD tasks with Red-Green-Refactor cycles and execute them with phase-specific logic and automatic test validation.
 
@@ -510,3 +497,4 @@ if (taskJson.meta?.tdd_workflow) {
   agent = "code-developer"  // Use generic agent
 }
 ```
+"""
diff --git a/.codex/agents/test-action-planning-agent.md b/.codex/agents/test-action-planning-agent.toml
similarity index 96%
rename from .codex/agents/test-action-planning-agent.md
rename to .codex/agents/test-action-planning-agent.toml
index 4f9258da..4e78edda 100644
--- a/.codex/agents/test-action-planning-agent.md
+++ b/.codex/agents/test-action-planning-agent.toml
@@ -1,19 +1,10 @@
----
-name: test-action-planning-agent
-description: |
-  Specialized agent extending action-planning-agent for test planning documents. Generates test task JSONs (IMPL-001, IMPL-001.3, IMPL-001.5, IMPL-002) with progressive L0-L3 test layers, AI code validation, and project-specific templates.
+name = "test_action_planning_agent"
+description = "Specialized agent extending action-planning-agent for test planning documents. Generates test task JSONs (IMPL-001, IMPL-001.3, IMPL-001.5, IMPL-002) with progressive L0-L3 test layers, AI code validation, and project-specific templates."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
 
-  Inherits from: @action-planning-agent
-  See: d:\Claude_dms3\.claude\agents\action-planning-agent.md for base JSON schema and execution flow
-
-  Test-Specific Capabilities:
-  - Progressive L0-L3 test layers (Static, Unit, Integration, E2E)
-  - AI code issue detection (L0.5) with CRITICAL/ERROR/WARNING severity
-  - Project type templates (React, Node API, CLI, Library, Monorepo)
-  - Test anti-pattern detection with quality gates
-  - Layer completeness thresholds and coverage targets
-color: cyan
----
+developer_instructions = """
 
 ## Agent Inheritance
 
@@ -682,3 +673,4 @@ Hard Constraints:
 - AI issue detection configured in IMPL-001.3
 - Quality gates with measurable thresholds in IMPL-001.5
 - Source session status reported (if applicable)
+"""
diff --git a/.codex/agents/test-context-search-agent.md b/.codex/agents/test-context-search-agent.toml
similarity index 93%
rename from .codex/agents/test-context-search-agent.md
rename to .codex/agents/test-context-search-agent.toml
index a90a1eab..678dd83d 100644
--- a/.codex/agents/test-context-search-agent.md
+++ b/.codex/agents/test-context-search-agent.toml
@@ -1,20 +1,10 @@
----
-name: test-context-search-agent
-description: |
-  Specialized context collector for test generation workflows. Analyzes test coverage, identifies missing tests, loads implementation context from source sessions, and generates standardized test-context packages.
+name = "test_context_search_agent"
+description = "Specialized context collector for test generation workflows. Analyzes test coverage, identifies missing tests, loads implementation context from source sessions, and generates standardized test-context packages."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "read-only"
 
-  Examples:
-  - Context: Test session with source session reference
-    user: "Gather test context for WFS-test-auth session"
-    assistant: "I'll load source implementation, analyze test coverage, and generate test-context package"
-    commentary: Execute autonomous coverage analysis with source context loading
-
-  - Context: Multi-framework detection needed
-    user: "Collect test context for full-stack project"
-    assistant: "I'll detect Jest frontend and pytest backend frameworks, analyze coverage gaps"
-    commentary: Identify framework patterns and conventions for each stack
-color: blue
----
+developer_instructions = """
 
 You are a test context discovery specialist focused on gathering test coverage information and implementation context for test generation workflows. Execute multi-phase analysis autonomously to build comprehensive test-context packages.
 
@@ -399,4 +389,4 @@ function detect_framework_from_config() {
 - ✅ Valid test-context-package.json generated
 - ✅ All missing tests catalogued with priority
 - ✅ Execution time < 30 seconds (< 60s for large codebases)
-
+"""
diff --git a/.codex/agents/test-fix-agent.md b/.codex/agents/test-fix-agent.toml
similarity index 91%
rename from .codex/agents/test-fix-agent.md
rename to .codex/agents/test-fix-agent.toml
index d99f9b25..fb756e46 100644
--- a/.codex/agents/test-fix-agent.md
+++ b/.codex/agents/test-fix-agent.toml
@@ -1,25 +1,10 @@
----
-name: test-fix-agent
-description: |
-  Execute tests, diagnose failures, and fix code until all tests pass. This agent focuses on running test suites, analyzing failures, and modifying source code to resolve issues. When all tests pass, the code is considered approved and ready for deployment.
+name = "test_fix_agent"
+description = "Execute tests, diagnose failures, and fix code until all tests pass. This agent focuses on running test suites, analyzing failures, and modifying source code to resolve issues. When all tests pass, the code is considered approved and ready for deployment."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
 
-  Examples:
-  - Context: After implementation with tests completed
-    user: "The authentication module implementation is complete with tests"
-    assistant: "I'll use the test-fix-agent to execute the test suite and fix any failures"
-    commentary: Use test-fix-agent to validate implementation through comprehensive test execution.
-
-  - Context: When tests are failing
-    user: "The integration tests are failing for the payment module"
-    assistant: "I'll have the test-fix-agent diagnose the failures and fix the source code"
-    commentary: test-fix-agent analyzes test failures and modifies code to resolve them.
-
-  - Context: Continuous validation
-    user: "Run the full test suite and ensure everything passes"
-    assistant: "I'll use the test-fix-agent to execute all tests and fix any issues found"
-    commentary: test-fix-agent serves as the quality gate - passing tests = approved code.
-color: green
----
+developer_instructions = """
 
 You are a specialized **Test Execution & Fix Agent**. Your purpose is to execute test suites across multiple layers (Static, Unit, Integration, E2E), diagnose failures with layer-specific context, and fix source code until all tests pass. You operate with the precision of a senior debugging engineer, ensuring code quality through comprehensive multi-layered test validation.
 
@@ -368,4 +353,5 @@ jq --arg ts "$(date -Iseconds)" '.status="completed" | .status_history += [{"fro
 
 **Tests passing = Code approved = Mission complete** ✅
 ### Windows Path Format Guidelines
-- **Quick Ref**: `C:\Users` → MCP: `C:\\Users` | Bash: `/c/Users` or `C:/Users`
\ No newline at end of file
+- **Quick Ref**: `C:\Users` → MCP: `C:\\Users` | Bash: `/c/Users` or `C:/Users`
+"""
diff --git a/.codex/agents/ui-design-agent.md b/.codex/agents/ui-design-agent.toml
similarity index 96%
rename from .codex/agents/ui-design-agent.md
rename to .codex/agents/ui-design-agent.toml
index 1a9e966d..4578669d 100644
--- a/.codex/agents/ui-design-agent.md
+++ b/.codex/agents/ui-design-agent.toml
@@ -1,29 +1,10 @@
----
-name: ui-design-agent
-description: |
-  Specialized agent for UI design token management and prototype generation with W3C Design Tokens Format compliance.
+name = "ui_design_agent"
+description = "Specialized agent for UI design token management and prototype generation with W3C Design Tokens Format compliance."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
 
-  Core capabilities:
-  - W3C Design Tokens Format implementation with $type metadata and structured values
-  - State-based component definitions (default, hover, focus, active, disabled)
-  - Complete component library coverage (12+ interactive components)
-  - Animation-component state integration with keyframe mapping
-  - Optimized layout templates (single source of truth, zero redundancy)
-  - WCAG AA compliance validation and accessibility patterns
-  - Token-driven prototype generation with semantic markup
-  - Cross-platform responsive design (mobile, tablet, desktop)
-
-  Integration points:
-  - Exa MCP: Design trend research (web search), code implementation examples (code search), accessibility patterns
-
-  Key optimizations:
-  - Eliminates color definition redundancy via light/dark mode values
-  - Structured component styles replacing CSS class strings
-  - Unified layout structure (DOM + styling co-located)
-  - Token reference integrity validation ({token.path} syntax)
-
-color: orange
----
+developer_instructions = """
 
 You are a specialized **UI Design Agent** that executes design generation tasks autonomously to produce production-ready design systems and prototypes.
 
@@ -593,3 +574,4 @@ mcp__exa__get_code_context_exa(
   outline-offset: 2px;
 }
 ```
+"""
diff --git a/.codex/agents/universal-executor.md b/.codex/agents/universal-executor.toml
similarity index 83%
rename from .codex/agents/universal-executor.md
rename to .codex/agents/universal-executor.toml
index f725ee86..f698783b 100644
--- a/.codex/agents/universal-executor.md
+++ b/.codex/agents/universal-executor.toml
@@ -1,20 +1,10 @@
----
-name: universal-executor
-description: |
-  Versatile execution agent for implementing any task efficiently. Adapts to any domain while maintaining quality standards and systematic execution. Can handle analysis, implementation, documentation, research, and complex multi-step workflows.
+name = "universal_executor"
+description = "Versatile execution agent for implementing any task efficiently. Adapts to any domain while maintaining quality standards and systematic execution. Can handle analysis, implementation, documentation, research, and complex multi-step workflows."
+model = "gpt-5.4"
+model_reasoning_effort = "high"
+sandbox_mode = "workspace-write"
 
-  Examples:
-  - Context: User provides task with sufficient context
-    user: "Analyze market trends and create presentation following these guidelines: [context]"
-    assistant: "I'll analyze the market trends and create the presentation using the provided guidelines"
-    commentary: Execute task directly with user-provided context
-
-  - Context: User provides insufficient context
-    user: "Organize project documentation"
-    assistant: "I need to understand the current documentation structure first"
-    commentary: Gather context about existing documentation, then execute
-color: green
----
+developer_instructions = """
 
 You are a versatile execution specialist focused on completing high-quality tasks efficiently across any domain. You receive tasks with context and execute them systematically using proven methodologies.
 
@@ -132,4 +122,5 @@ Before completing any task, verify:
 - Handle edge cases appropriately
 - Keep tasks focused and manageable
 - Generate detailed summary documents with complete deliverable listings
-- Document all key outputs and integration points for dependent tasks
\ No newline at end of file
+- Document all key outputs and integration points for dependent tasks
+"""
diff --git a/.codex/skills/brainstorm-with-file/SKILL.md b/.codex/skills/brainstorm-with-file/SKILL.md
index a82c547b..cfa005c0 100644
--- a/.codex/skills/brainstorm-with-file/SKILL.md
+++ b/.codex/skills/brainstorm-with-file/SKILL.md
@@ -20,7 +20,7 @@ Interactive brainstorming workflow with **documented thought evolution**. Expand
 
 **Codex-Specific Features**:
 - Parallel subagent execution via `spawn_agent` + batch `wait({ ids: [...] })`
-- Role loading via path (agent reads `~/.codex/agents/*.md` itself)
+- Role loading via TOML agent definition (agent_type parameter in spawn_agent)
 - Deep interaction with `send_input` for multi-round refinement within single agent
 - Explicit lifecycle management with `close_agent`
 
@@ -208,12 +208,10 @@ Generate key questions that guide the brainstorming exploration. Use a subagent
 
 ```javascript
 const vectorAgent = spawn_agent({
+  agent_type: "cli_explore_agent",
   message: `
 ## TASK ASSIGNMENT
 
-### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-
 ---
 
 ## Context
@@ -302,9 +300,9 @@ Spawn 3 perspective agents in parallel: Creative + Pragmatic + Systematic.
 
 | Perspective | Role File | Focus |
 |-------------|-----------|-------|
-| Creative | `~/.codex/agents/cli-explore-agent.md` | Innovation, cross-domain inspiration, challenging assumptions |
-| Pragmatic | `~/.codex/agents/cli-explore-agent.md` | Implementation feasibility, effort estimates, blockers |
-| Systematic | `~/.codex/agents/cli-explore-agent.md` | Problem decomposition, patterns, scalability |
+| Creative | `cli_explore_agent` | Innovation, cross-domain inspiration, challenging assumptions |
+| Pragmatic | `cli_explore_agent` | Implementation feasibility, effort estimates, blockers |
+| Systematic | `cli_explore_agent` | Problem decomposition, patterns, scalability |
 
 **Parallel Subagent Execution**:
 
@@ -353,13 +351,13 @@ const perspectives = [
 // Parallel spawn - all agents start immediately
 const agentIds = perspectives.map(perspective => {
   return spawn_agent({
+    agent_type: "cli_explore_agent",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Run: `ccw spec load --category "exploration planning"`
-3. Read project tech context from loaded specs
+1. Run: `ccw spec load --category "exploration planning"`
+2. Read project tech context from loaded specs
 
 ---
 
@@ -560,13 +558,13 @@ const deepDiveResult = wait({ ids: [perspectiveAgent], timeout_ms: 600000 })
 
 ```javascript
 const deepDiveAgent = spawn_agent({
+  agent_type: "cli_explore_agent",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read: ${sessionFolder}/perspectives.json (prior findings)
-3. Run: `ccw spec load --category "exploration planning"`
+1. Read: ${sessionFolder}/perspectives.json (prior findings)
+2. Run: `ccw spec load --category "exploration planning"`
 
 ---
 
@@ -606,12 +604,12 @@ When user selects "challenge", spawn a dedicated challenge agent.
 
 ```javascript
 const challengeAgent = spawn_agent({
+  agent_type: "cli_explore_agent",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read: ${sessionFolder}/perspectives.json (ideas to challenge)
+1. Read: ${sessionFolder}/perspectives.json (ideas to challenge)
 
 ---
 
@@ -659,12 +657,12 @@ When user selects "merge", synthesize complementary ideas.
 
 ```javascript
 const mergeAgent = spawn_agent({
+  agent_type: "cli_explore_agent",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read: ${sessionFolder}/perspectives.json (source ideas)
+1. Read: ${sessionFolder}/perspectives.json (source ideas)
 
 ---
 
@@ -888,7 +886,7 @@ Dimensions guide brainstorming scope and focus:
 ```javascript
 // Safe parallel execution with error handling
 try {
-  const agentIds = perspectives.map(p => spawn_agent({ message: buildPrompt(p) }))
+  const agentIds = perspectives.map(p => spawn_agent({ agent_type: "cli_explore_agent", message: buildPrompt(p) }))
 
   const results = wait({ ids: agentIds, timeout_ms: 600000 })
 
@@ -1014,7 +1012,7 @@ Final synthesis:
 
 ### Codex Subagent Best Practices
 
-1. **Role Path, Not Content**: Pass `~/.codex/agents/*.md` path in message, let agent read itself
+1. **Agent Type, Not Path**: Use `agent_type` parameter in spawn_agent, not manual file path reading
 2. **Parallel for Perspectives**: Use batch spawn + wait for 3 perspective agents
 3. **Delay close_agent for Refinement**: Keep perspective agents alive for `send_input` reuse
 4. **Batch wait**: Use `wait({ ids: [a, b, c] })` for parallel agents, not sequential waits
diff --git a/.codex/skills/brainstorm/SKILL.md b/.codex/skills/brainstorm/SKILL.md
index f2fdc96f..c69e8897 100644
--- a/.codex/skills/brainstorm/SKILL.md
+++ b/.codex/skills/brainstorm/SKILL.md
@@ -69,7 +69,7 @@ Dual-mode brainstorming with CSV-driven parallel role analysis. Auto mode runs a
 │  ═══ Single Role Mode ═══                                         │
 │                                                                    │
 │  Phase 3S: Single Role Analysis (spawn_agent)                    │
-│     ├─ spawn_agent(conceptual-planning-agent)                    │
+│     ├─ spawn_agent(conceptual_planning_agent)                    │
 │     └─ Output: {role}/analysis*.md                               │
 │                                                                    │
 └──────────────────────────────────────────────────────────────────┘
@@ -536,7 +536,7 @@ if (executionMode === 'single-role') {
   Bash(`mkdir -p "${roleDir}"`)
 
   const agentId = spawn_agent({
-    agent: `~/.codex/agents/conceptual-planning-agent.md`,
+    agent_type: "conceptual_planning_agent",
     instruction: `
 Perform a ${roleName} analysis for the brainstorming session.
 
@@ -594,7 +594,7 @@ Follow the same analysis protocol as wave role analysis but with interactive ref
 
    ```javascript
    const synthesisAgent = spawn_agent({
-     agent: `~/.codex/agents/conceptual-planning-agent.md`,
+     agent_type: "conceptual_planning_agent",
      instruction: `
 ## SYNTHESIS ASSIGNMENT
 
diff --git a/.codex/skills/clean/SKILL.md b/.codex/skills/clean/SKILL.md
index 4ec86c06..485d9167 100644
--- a/.codex/skills/clean/SKILL.md
+++ b/.codex/skills/clean/SKILL.md
@@ -159,12 +159,12 @@ let exploreAgent = null
 try {
   // Launch cli-explore-agent
   exploreAgent = spawn_agent({
+    agent_type: "cli_explore_agent",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS
-1. Read: ~/.codex/agents/cli-explore-agent.md
-2. Read: ${projectRoot}/.workflow/project-tech.json (if exists)
+1. Read: ${projectRoot}/.workflow/project-tech.json (if exists)
 
 ## Task Objective
 Discover stale artifacts for cleanup.
diff --git a/.codex/skills/issue-discover/SKILL.md b/.codex/skills/issue-discover/SKILL.md
index 6366f08a..e0af2d63 100644
--- a/.codex/skills/issue-discover/SKILL.md
+++ b/.codex/skills/issue-discover/SKILL.md
@@ -229,13 +229,13 @@ Create a new subagent with task assignment.
 
 ```javascript
 const agentId = spawn_agent({
+  agent_type: "{agent_type}",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/{agent-type}.md (MUST read first)
-2. Execute: ccw spec load --category exploration
-3. Execute: ccw spec load --category debug (known issues cross-reference)
+1. Execute: ccw spec load --category exploration
+2. Execute: ccw spec load --category debug (known issues cross-reference)
 
 ## TASK CONTEXT
 ${taskContext}
diff --git a/.codex/skills/issue-discover/phases/02-discover.md b/.codex/skills/issue-discover/phases/02-discover.md
index 7901211c..2252f054 100644
--- a/.codex/skills/issue-discover/phases/02-discover.md
+++ b/.codex/skills/issue-discover/phases/02-discover.md
@@ -114,13 +114,13 @@ const perspectiveAgents = [];
 
 selectedPerspectives.forEach(perspective => {
   const agentId = spawn_agent({
+    agent_type: "cli_explore_agent",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read: {projectRoot}/.workflow/project-tech.json
-3. Read: {projectRoot}/.workflow/specs/*.md
+1. Read: {projectRoot}/.workflow/project-tech.json
+2. Read: {projectRoot}/.workflow/specs/*.md
 
 ---
 
@@ -192,13 +192,13 @@ agentIds.forEach(id => close_agent({ id }));
 // Only spawn if perspective requires external research
 if (selectedPerspectives.includes('security') || selectedPerspectives.includes('best-practices') || args.external) {
   const exaAgentId = spawn_agent({
+    agent_type: "cli_explore_agent",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read: {projectRoot}/.workflow/project-tech.json
-3. Read: {projectRoot}/.workflow/specs/*.md
+1. Read: {projectRoot}/.workflow/project-tech.json
+2. Read: {projectRoot}/.workflow/specs/*.md
 
 ---
 
diff --git a/.codex/skills/issue-discover/phases/03-discover-by-prompt.md b/.codex/skills/issue-discover/phases/03-discover-by-prompt.md
index e650c9bf..22b66374 100644
--- a/.codex/skills/issue-discover/phases/03-discover-by-prompt.md
+++ b/.codex/skills/issue-discover/phases/03-discover-by-prompt.md
@@ -234,6 +234,7 @@ while (shouldContinue && iteration < maxIterations) {
 
   iterationPlan.dimensions.forEach(dimension => {
     const agentId = spawn_agent({
+      agent_type: "cli_explore_agent",
       message: buildDimensionPromptWithACE(dimension, iteration, cumulativeFindings, iterationAceResults, iterationDir)
     });
     dimensionAgents.push({ agentId, dimension });
@@ -413,12 +414,11 @@ function buildDimensionPromptWithACE(dimension, iteration, previousFindings, ace
   );
 
   return `
-## TASK ASSIGNMENT
+## TASK ASSIGNMENT (agent_type: cli_explore_agent)
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read: {projectRoot}/.workflow/project-tech.json
-3. Read: {projectRoot}/.workflow/specs/*.md
+1. Read: {projectRoot}/.workflow/project-tech.json
+2. Read: {projectRoot}/.workflow/specs/*.md
 
 ---
 
diff --git a/.codex/skills/parallel-dev-cycle/phases/02-agent-execution.md b/.codex/skills/parallel-dev-cycle/phases/02-agent-execution.md
index 90799bbf..1f8c4d44 100644
--- a/.codex/skills/parallel-dev-cycle/phases/02-agent-execution.md
+++ b/.codex/skills/parallel-dev-cycle/phases/02-agent-execution.md
@@ -80,12 +80,12 @@ Cross-reference the task description against these documents for completeness.
   const focusDirective = getAgentFocusDirective('ra', state)
 
   return spawn_agent({
+    agent_type: "requirements_analyst",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/requirements-analyst.md
-2. Read: ${projectRoot}/.workflow/project-tech.json (if exists)
+1. Read: ${projectRoot}/.workflow/project-tech.json (if exists)
 3. Read: ${projectRoot}/.workflow/specs/*.md (if exists)
 4. Read: ${projectRoot}/.workflow/.cycle/${cycleId}.progress/coordination/feedback.md (if exists)
 
@@ -163,12 +163,12 @@ PHASE_RESULT:
 ```javascript
 function spawnEPAgent(cycleId, state, progressDir) {
   return spawn_agent({
+    agent_type: "exploration_planner",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/exploration-planner.md
-2. Read: ${projectRoot}/.workflow/project-tech.json
+1. Read: ${projectRoot}/.workflow/project-tech.json
 3. Read: ${projectRoot}/.workflow/specs/*.md
 4. Read: ${progressDir}/ra/requirements.md
 
@@ -242,12 +242,12 @@ PHASE_RESULT:
 ```javascript
 function spawnCDAgent(cycleId, state, progressDir) {
   return spawn_agent({
+    agent_type: "code_developer",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/code-developer.md
-2. Read: ${progressDir}/ep/plan.json
+1. Read: ${progressDir}/ep/plan.json
 3. Read: ${progressDir}/ra/requirements.md
 
 ---
@@ -320,12 +320,12 @@ PHASE_RESULT:
 ```javascript
 function spawnVASAgent(cycleId, state, progressDir) {
   return spawn_agent({
+    agent_type: "validation_archivist",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/validation-archivist.md
-2. Read: ${progressDir}/cd/changes.log
+1. Read: ${progressDir}/cd/changes.log
 
 ---
 
diff --git a/.codex/skills/review-cycle/SKILL.md b/.codex/skills/review-cycle/SKILL.md
index c3c35aec..71a8a6cc 100644
--- a/.codex/skills/review-cycle/SKILL.md
+++ b/.codex/skills/review-cycle/SKILL.md
@@ -50,7 +50,7 @@ Unified multi-dimensional code review orchestrator with dual-mode (session/modul
 3. **Progressive Phase Loading**: Phase docs are read on-demand when that phase executes, not all at once
 4. **Auto-Continue**: All phases run autonomously without user intervention between phases
 5. **Subagent Lifecycle**: Explicit lifecycle management with spawn_agent → wait → close_agent
-6. **Role Path Loading**: Subagent roles loaded via path reference in MANDATORY FIRST STEPS
+6. **Role via agent_type**: Subagent roles loaded via TOML `agent_type` parameter in spawn_agent (e.g., `"cli_explore_agent"`)
 7. **Optional Fix Pipeline**: Phase 6-9 triggered only by explicit `--fix` flag or user confirmation after Phase 5
 8. **Content Preservation**: All agent prompts, code, schemas preserved verbatim from source commands
 
@@ -298,12 +298,12 @@ Create a new subagent with task assignment.
 
 ```javascript
 const agentId = spawn_agent({
+  agent_type: "{agent_type}",  // TOML agent definition name (e.g., "cli_explore_agent")
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/{agent-type}.md (MUST read first)
-2. Execute: ccw spec load --category "exploration execution"
+1. Execute: ccw spec load --category "exploration execution"
 
 ---
 
diff --git a/.codex/skills/review-cycle/phases/02-parallel-review.md b/.codex/skills/review-cycle/phases/02-parallel-review.md
index da9111a1..e15b256e 100644
--- a/.codex/skills/review-cycle/phases/02-parallel-review.md
+++ b/.codex/skills/review-cycle/phases/02-parallel-review.md
@@ -85,17 +85,17 @@ const dimensions = ['security', 'architecture', 'quality', 'action-items', 'perf
 
 dimensions.forEach(dimension => {
   const agentId = spawn_agent({
+    agent_type: "cli_explore_agent",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read review state: ${reviewStateJsonPath}
-3. Get target files: Read resolved_files from review-state.json
-4. Validate file access: bash(ls -la ${targetFiles.join(' ')})
-5. Execute: cat ~/.ccw/workflows/cli-templates/schemas/review-dimension-results-schema.json (get output schema reference)
-6. Execute: ccw spec load --category "exploration execution" (technology stack and constraints)
-7. Execute: ccw spec load --category review (review standards and checklists)
+1. Read review state: ${reviewStateJsonPath}
+2. Get target files: Read resolved_files from review-state.json
+3. Validate file access: bash(ls -la ${targetFiles.join(' ')})
+4. Execute: cat ~/.ccw/workflows/cli-templates/schemas/review-dimension-results-schema.json (get output schema reference)
+5. Execute: ccw spec load --category "exploration execution" (technology stack and constraints)
+6. Execute: ccw spec load --category review (review standards and checklists)
 
 ---
 
@@ -124,7 +124,7 @@ Use **Deep Scan mode** for this review:
 
 ## Expected Deliverables
 
-**Schema Reference**: Schema obtained in MANDATORY FIRST STEPS step 5, follow schema exactly
+**Schema Reference**: Schema obtained in MANDATORY FIRST STEPS step 4, follow schema exactly
 
 1. Dimension Results JSON: ${outputDir}/dimensions/${dimension}.json
 
@@ -207,18 +207,18 @@ const dimensions = ['security', 'architecture', 'quality', 'action-items', 'perf
 
 dimensions.forEach(dimension => {
   const agentId = spawn_agent({
+    agent_type: "cli_explore_agent",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read session metadata: ${sessionMetadataPath}
-3. Read completed task summaries: bash(find ${summariesDir} -name "IMPL-*.md" -type f)
-4. Get changed files: bash(cd ${workflowDir} && git log --since="${sessionCreatedAt}" --name-only --pretty=format: | sort -u)
-5. Read review state: ${reviewStateJsonPath}
-6. Execute: cat ~/.ccw/workflows/cli-templates/schemas/review-dimension-results-schema.json (get output schema reference)
-7. Execute: ccw spec load --category "exploration execution" (technology stack and constraints)
-8. Execute: ccw spec load --category review (review standards and checklists)
+1. Read session metadata: ${sessionMetadataPath}
+2. Read completed task summaries: bash(find ${summariesDir} -name "IMPL-*.md" -type f)
+3. Get changed files: bash(cd ${workflowDir} && git log --since="${sessionCreatedAt}" --name-only --pretty=format: | sort -u)
+4. Read review state: ${reviewStateJsonPath}
+5. Execute: cat ~/.ccw/workflows/cli-templates/schemas/review-dimension-results-schema.json (get output schema reference)
+6. Execute: ccw spec load --category "exploration execution" (technology stack and constraints)
+7. Execute: ccw spec load --category review (review standards and checklists)
 
 ---
 
@@ -247,7 +247,7 @@ Use **Deep Scan mode** for this review:
 
 ## Expected Deliverables
 
-**Schema Reference**: Schema obtained in MANDATORY FIRST STEPS step 6, follow schema exactly
+**Schema Reference**: Schema obtained in MANDATORY FIRST STEPS step 5, follow schema exactly
 
 1. Dimension Results JSON: ${outputDir}/dimensions/${dimension}.json
 
@@ -326,18 +326,18 @@ reviewAgents.forEach(id => close_agent({ id }));
 ```javascript
 // Spawn deep-dive agent
 const deepDiveAgentId = spawn_agent({
+  agent_type: "cli_explore_agent",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read original finding: ${dimensionJsonPath}
-3. Read affected file: ${file}
-4. Identify related code: bash(grep -r "import.*${basename(file)}" ${projectDir}/src --include="*.ts")
-5. Read test files: bash(find ${projectDir}/tests -name "*${basename(file, '.ts')}*" -type f)
-6. Execute: cat ~/.ccw/workflows/cli-templates/schemas/review-deep-dive-results-schema.json (get output schema reference)
-7. Execute: ccw spec load --category "exploration execution" (technology stack and constraints for remediation)
-8. Execute: ccw spec load --category review (review standards and checklists)
+1. Read original finding: ${dimensionJsonPath}
+2. Read affected file: ${file}
+3. Identify related code: bash(grep -r "import.*${basename(file)}" ${projectDir}/src --include="*.ts")
+4. Read test files: bash(find ${projectDir}/tests -name "*${basename(file, '.ts')}*" -type f)
+5. Execute: cat ~/.ccw/workflows/cli-templates/schemas/review-deep-dive-results-schema.json (get output schema reference)
+6. Execute: ccw spec load --category "exploration execution" (technology stack and constraints for remediation)
+7. Execute: ccw spec load --category review (review standards and checklists)
 
 ---
 
@@ -372,7 +372,7 @@ Then apply **Deep Scan mode** for semantic analysis:
 
 ## Expected Deliverables
 
-**Schema Reference**: Schema obtained in MANDATORY FIRST STEPS step 6, follow schema exactly
+**Schema Reference**: Schema obtained in MANDATORY FIRST STEPS step 5, follow schema exactly
 
 1. Deep-Dive Results JSON: ${outputDir}/iterations/iteration-${iteration}-finding-${findingId}.json
 
diff --git a/.codex/skills/review-cycle/phases/04-iterative-deep-dive.md b/.codex/skills/review-cycle/phases/04-iterative-deep-dive.md
index 04d989f8..1e849318 100644
--- a/.codex/skills/review-cycle/phases/04-iterative-deep-dive.md
+++ b/.codex/skills/review-cycle/phases/04-iterative-deep-dive.md
@@ -79,18 +79,18 @@ const deepDiveAgents = [];
 
 selectedFindings.forEach(finding => {
   const agentId = spawn_agent({
+    agent_type: "cli_explore_agent",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read original finding: ${dimensionJsonPath}
-3. Read affected file: ${finding.file}
-4. Identify related code: bash(grep -r "import.*${basename(finding.file)}" ${projectDir}/src --include="*.ts")
-5. Read test files: bash(find ${projectDir}/tests -name "*${basename(finding.file, '.ts')}*" -type f)
-6. Execute: cat ~/.ccw/workflows/cli-templates/schemas/review-deep-dive-results-schema.json (get output schema reference)
-7. Read: ${projectRoot}/.workflow/project-tech.json (technology stack and architecture context)
-8. Read: ${projectRoot}/.workflow/specs/*.md (user-defined constraints for remediation compliance)
+1. Read original finding: ${dimensionJsonPath}
+2. Read affected file: ${finding.file}
+3. Identify related code: bash(grep -r "import.*${basename(finding.file)}" ${projectDir}/src --include="*.ts")
+4. Read test files: bash(find ${projectDir}/tests -name "*${basename(finding.file, '.ts')}*" -type f)
+5. Execute: cat ~/.ccw/workflows/cli-templates/schemas/review-deep-dive-results-schema.json (get output schema reference)
+6. Read: ${projectRoot}/.workflow/project-tech.json (technology stack and architecture context)
+7. Read: ${projectRoot}/.workflow/specs/*.md (user-defined constraints for remediation compliance)
 
 ---
 
@@ -125,7 +125,7 @@ Then apply **Deep Scan mode** for semantic analysis:
 
 ## Expected Deliverables
 
-**Schema Reference**: Schema obtained in MANDATORY FIRST STEPS step 6, follow schema exactly
+**Schema Reference**: Schema obtained in MANDATORY FIRST STEPS step 5, follow schema exactly
 
 1. Deep-Dive Results JSON: ${outputDir}/iterations/iteration-${iteration}-finding-${finding.id}.json
 
@@ -190,18 +190,18 @@ const deepDiveAgents = [];
 
 selectedFindings.forEach(finding => {
   const agentId = spawn_agent({
+    agent_type: "cli_explore_agent",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read original finding: ${dimensionJsonPath}
-3. Read affected file: ${finding.file}
-4. Identify related code: bash(grep -r "import.*${basename(finding.file)}" ${workflowDir}/src --include="*.ts")
-5. Read test files: bash(find ${workflowDir}/tests -name "*${basename(finding.file, '.ts')}*" -type f)
-6. Execute: cat ~/.ccw/workflows/cli-templates/schemas/review-deep-dive-results-schema.json (get output schema reference)
-7. Read: ${projectRoot}/.workflow/project-tech.json (technology stack and architecture context)
-8. Read: ${projectRoot}/.workflow/specs/*.md (user-defined constraints for remediation compliance)
+1. Read original finding: ${dimensionJsonPath}
+2. Read affected file: ${finding.file}
+3. Identify related code: bash(grep -r "import.*${basename(finding.file)}" ${workflowDir}/src --include="*.ts")
+4. Read test files: bash(find ${workflowDir}/tests -name "*${basename(finding.file, '.ts')}*" -type f)
+5. Execute: cat ~/.ccw/workflows/cli-templates/schemas/review-deep-dive-results-schema.json (get output schema reference)
+6. Read: ${projectRoot}/.workflow/project-tech.json (technology stack and architecture context)
+7. Read: ${projectRoot}/.workflow/specs/*.md (user-defined constraints for remediation compliance)
 
 ---
 
@@ -237,7 +237,7 @@ Then apply **Deep Scan mode** for semantic analysis:
 
 ## Expected Deliverables
 
-**Schema Reference**: Schema obtained in MANDATORY FIRST STEPS step 6, follow schema exactly
+**Schema Reference**: Schema obtained in MANDATORY FIRST STEPS step 5, follow schema exactly
 
 1. Deep-Dive Results JSON: ${outputDir}/iterations/iteration-${iteration}-finding-${finding.id}.json
 
diff --git a/.codex/skills/review-cycle/phases/07-fix-parallel-planning.md b/.codex/skills/review-cycle/phases/07-fix-parallel-planning.md
index e69c3b80..142332c9 100644
--- a/.codex/skills/review-cycle/phases/07-fix-parallel-planning.md
+++ b/.codex/skills/review-cycle/phases/07-fix-parallel-planning.md
@@ -100,12 +100,12 @@ for (let i = 1; i <= aggregatedPlan.groups.length; i++) {
 ```javascript
 // Spawn planning agent for a batch
 const agentId = spawn_agent({
+  agent_type: "cli_planning_agent",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-planning-agent.md (MUST read first)
-2. Execute: ccw spec load --category planning
+1. Execute: ccw spec load --category planning
 
 ---
 
diff --git a/.codex/skills/review-cycle/phases/08-fix-execution.md b/.codex/skills/review-cycle/phases/08-fix-execution.md
index a90f0f4f..3e670d40 100644
--- a/.codex/skills/review-cycle/phases/08-fix-execution.md
+++ b/.codex/skills/review-cycle/phases/08-fix-execution.md
@@ -55,12 +55,12 @@ if (result.passRate < 100%) {
 ```javascript
 // Spawn execution agent for a group
 const execAgentId = spawn_agent({
+  agent_type: "cli_execution_agent",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-execution-agent.md (MUST read first)
-2. Execute: ccw spec load --category execution
+1. Execute: ccw spec load --category execution
 
 ---
 
diff --git a/.codex/skills/roadmap-with-file/SKILL.md b/.codex/skills/roadmap-with-file/SKILL.md
index 88f8f34b..382711b7 100644
--- a/.codex/skills/roadmap-with-file/SKILL.md
+++ b/.codex/skills/roadmap-with-file/SKILL.md
@@ -39,13 +39,13 @@ Create a new subagent with task assignment.
 
 ```javascript
 const agentId = spawn_agent({
+  agent_type: "{agent_type}",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/{agent-type}.md (MUST read first)
-2. Read: .workflow/project-tech.json
-3. Read: .workflow/project-guidelines.json
+1. Read: .workflow/project-tech.json
+2. Read: .workflow/project-guidelines.json
 
 ## TASK CONTEXT
 ${taskContext}
@@ -502,13 +502,13 @@ Bash(`mkdir -p ${sessionFolder}`)
 
    if (hasCodebase !== 'none') {
      const exploreAgentId = spawn_agent({
+       agent_type: "cli_explore_agent",
        message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read: .workflow/project-tech.json
-3. Read: .workflow/project-guidelines.json
+1. Read: .workflow/project-tech.json
+2. Read: .workflow/project-guidelines.json
 
 ---
 
@@ -553,13 +553,13 @@ Return findings as JSON with schema:
 2. **Execute Decomposition Agent**
    ```javascript
    const decompositionAgentId = spawn_agent({
+     agent_type: "cli_roadmap_plan_agent",
      message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-roadmap-plan-agent.md (MUST read first)
-2. Read: .workflow/project-tech.json
-3. Read: .workflow/project-guidelines.json
+1. Read: .workflow/project-tech.json
+2. Read: .workflow/project-guidelines.json
 
 ---
 
diff --git a/.codex/skills/spec-setup/SKILL.md b/.codex/skills/spec-setup/SKILL.md
index 8fd079c4..fd6c2ce1 100644
--- a/.codex/skills/spec-setup/SKILL.md
+++ b/.codex/skills/spec-setup/SKILL.md
@@ -122,12 +122,12 @@ let exploreAgent = null
 
 try {
   exploreAgent = spawn_agent({
+    agent_type: "cli_explore_agent",
     message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-explore-agent.md (MUST read first)
-2. Read: .workflow/project-tech.json (if exists, for --regenerate)
+1. Read: .workflow/project-tech.json (if exists, for --regenerate)
 
 ---
 
diff --git a/.codex/skills/workflow-plan/SKILL.md b/.codex/skills/workflow-plan/SKILL.md
index c10763b9..d0f6a5dc 100644
--- a/.codex/skills/workflow-plan/SKILL.md
+++ b/.codex/skills/workflow-plan/SKILL.md
@@ -242,7 +242,7 @@ console.log(`Session: ${sessionId}`)
 console.log(`\n## Phase 2: Context Gathering\n`)
 
 const ctxAgent = spawn_agent({
-  agent: `~/.codex/agents/context-search-agent.md`,
+  agent_type: "context_search_agent",
   instruction: `
 Gather implementation context for planning.
 
@@ -376,7 +376,7 @@ TASK DESCRIPTION: ${taskDescription}" --tool gemini --mode analysis --rule analy
    ```javascript
    if (!isMultiModule) {
      const planAgent = spawn_agent({
-       agent: `~/.codex/agents/action-planning-agent.md`,
+       agent_type: "action_planning_agent",
        instruction: `
 Generate implementation plan and task JSONs.
 
@@ -421,7 +421,7 @@ ${contextPkg.conflict_risk === 'medium' || contextPkg.conflict_risk === 'high'
        const mod = uniqueModules[i]
 
        const agentId = spawn_agent({
-         agent: `~/.codex/agents/action-planning-agent.md`,
+         agent_type: "action_planning_agent",
          instruction: `
 Plan module: ${mod} (prefix: ${prefix})
 
@@ -444,7 +444,7 @@ Mark cross-module dependencies as CROSS::${'{module}'}::${'{task}'}
 
      // +1 Coordinator: integrate all modules
      const coordAgent = spawn_agent({
-       agent: `~/.codex/agents/action-planning-agent.md`,
+       agent_type: "action_planning_agent",
        instruction: `
 Integrate ${uniqueModules.length} module plans into unified IMPL_PLAN.md.
 
@@ -576,7 +576,7 @@ if (mode === 'replan') {
 
   // 2. Replan via agent
   const replanAgent = spawn_agent({
-    agent: `~/.codex/agents/action-planning-agent.md`,
+    agent_type: "action_planning_agent",
     instruction: `
 Replan ${scope === 'task' ? `task ${replanTaskId}` : 'entire session'}.
 
diff --git a/.codex/skills/workflow-tdd-plan/SKILL.md b/.codex/skills/workflow-tdd-plan/SKILL.md
index 99913938..850c6149 100644
--- a/.codex/skills/workflow-tdd-plan/SKILL.md
+++ b/.codex/skills/workflow-tdd-plan/SKILL.md
@@ -269,7 +269,7 @@ console.log(`Session: ${sessionId}`)
 console.log(`\n## Phase 2: Context Gathering\n`)
 
 const ctxAgent = spawn_agent({
-  agent: `~/.codex/agents/context-search-agent.md`,
+  agent_type: "context_search_agent",
   instruction: `
 Gather implementation context for TDD planning.
 
@@ -329,7 +329,7 @@ console.log(`  Context gathered. Conflict risk: ${conflictRisk}`)
 console.log(`\n## Phase 3: Test Coverage Analysis\n`)
 
 const testAgent = spawn_agent({
-  agent: `~/.codex/agents/cli-explore-agent.md`,
+  agent_type: "cli_explore_agent",
   instruction: `
 Analyze test coverage and framework for TDD planning.
 
@@ -452,7 +452,7 @@ TASK DESCRIPTION: ${taskDescription}" --tool gemini --mode analysis --rule analy
 console.log(`\n## Phase 5: TDD Task Generation\n`)
 
 const planAgent = spawn_agent({
-  agent: `~/.codex/agents/action-planning-agent.md`,
+  agent_type: "action_planning_agent",
   instruction: `
 Generate TDD implementation plan with Red-Green-Refactor cycles.
 
@@ -628,7 +628,7 @@ if (mode === 'verify' || /* auto-verify from Phase 6 */) {
   }
 
   const verifyAgent = spawn_agent({
-    agent: `~/.codex/agents/cli-explore-agent.md`,
+    agent_type: "cli_explore_agent",
     instruction: `
 Verify TDD compliance across 4 dimensions.
 
diff --git a/.codex/skills/workflow-test-fix-cycle/SKILL.md b/.codex/skills/workflow-test-fix-cycle/SKILL.md
index 60306d24..6c5c5804 100644
--- a/.codex/skills/workflow-test-fix-cycle/SKILL.md
+++ b/.codex/skills/workflow-test-fix-cycle/SKILL.md
@@ -83,12 +83,12 @@ Create a new subagent with task assignment.
 
 ```javascript
 const agentId = spawn_agent({
+  agent_type: "{agent_type}",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/{agent-type}.md (MUST read first)
-2. Run: `ccw spec load --category "planning execution"`
+1. Run: `ccw spec load --category "planning execution"`
 
 ## TASK CONTEXT
 ${taskContext}
@@ -241,10 +241,10 @@ Phase 2: Test-Cycle Execution (phases/02-test-cycle-execute.md)
 5. Phase 1 Summary → **⛔ MANDATORY: Present plan and wait for user confirmation before Phase 2**
 
 **Agents Used** (via spawn_agent):
-- `test-context-search-agent` (~/.codex/agents/test-context-search-agent.md) - Context gathering (Session Mode)
-- `context-search-agent` (~/.codex/agents/context-search-agent.md) - Context gathering (Prompt Mode)
-- `cli-execution-agent` (~/.codex/agents/cli-execution-agent.md) - Test analysis with Gemini
-- `action-planning-agent` (~/.codex/agents/action-planning-agent.md) - Task JSON generation
+- `test_context_search_agent` (agent_type: test_context_search_agent) - Context gathering (Session Mode)
+- `context_search_agent` (agent_type: context_search_agent) - Context gathering (Prompt Mode)
+- `cli_execution_agent` (agent_type: cli_execution_agent) - Test analysis with Gemini
+- `action_planning_agent` (agent_type: action_planning_agent) - Task JSON generation
 
 ### Phase 2: Test-Cycle Execution
 
@@ -256,8 +256,8 @@ Phase 2: Test-Cycle Execution (phases/02-test-cycle-execute.md)
 3. Completion - Final validation → Summary → Auto-complete session
 
 **Agents Used** (via spawn_agent):
-- `cli-planning-agent` (~/.codex/agents/cli-planning-agent.md) - Failure analysis, root cause extraction, fix task generation
-- `test-fix-agent` (~/.codex/agents/test-fix-agent.md) - Test execution, code fixes, criticality assignment
+- `cli_planning_agent` (agent_type: cli_planning_agent) - Failure analysis, root cause extraction, fix task generation
+- `test_fix_agent` (agent_type: test_fix_agent) - Test execution, code fixes, criticality assignment
 
 **Strategy Engine**: conservative (iteration 1-2) → aggressive (pass >80%) → surgical (regression)
 
@@ -387,14 +387,14 @@ try {
 - None for Prompt Mode
 
 **Phase 1 Agents** (used by phases/01-test-fix-gen.md via spawn_agent):
-- `test-context-search-agent` (~/.codex/agents/test-context-search-agent.md) - Test coverage analysis (Session Mode)
-- `context-search-agent` (~/.codex/agents/context-search-agent.md) - Codebase analysis (Prompt Mode)
-- `cli-execution-agent` (~/.codex/agents/cli-execution-agent.md) - Test requirements with Gemini
-- `action-planning-agent` (~/.codex/agents/action-planning-agent.md) - Task JSON generation
+- `test_context_search_agent` (agent_type: test_context_search_agent) - Test coverage analysis (Session Mode)
+- `context_search_agent` (agent_type: context_search_agent) - Codebase analysis (Prompt Mode)
+- `cli_execution_agent` (agent_type: cli_execution_agent) - Test requirements with Gemini
+- `action_planning_agent` (agent_type: action_planning_agent) - Task JSON generation
 
 **Phase 2 Agents** (used by phases/02-test-cycle-execute.md via spawn_agent):
-- `cli-planning-agent` (~/.codex/agents/cli-planning-agent.md) - CLI analysis, root cause extraction, task generation
-- `test-fix-agent` (~/.codex/agents/test-fix-agent.md) - Test execution, code fixes, criticality assignment
+- `cli_planning_agent` (agent_type: cli_planning_agent) - CLI analysis, root cause extraction, task generation
+- `test_fix_agent` (agent_type: test_fix_agent) - Test execution, code fixes, criticality assignment
 
 **Follow-up**:
 - Session sync: `$session-sync -y "Test-fix cycle complete: {pass_rate}% pass rate"`
diff --git a/.codex/skills/workflow-test-fix-cycle/phases/01-test-fix-gen.md b/.codex/skills/workflow-test-fix-cycle/phases/01-test-fix-gen.md
index 0f95c494..0c079b38 100644
--- a/.codex/skills/workflow-test-fix-cycle/phases/01-test-fix-gen.md
+++ b/.codex/skills/workflow-test-fix-cycle/phases/01-test-fix-gen.md
@@ -70,13 +70,13 @@ if (input.startsWith("WFS-")) {
 ```javascript
 // Session Mode - gather from source session via test-context-search-agent
 const contextAgentId = spawn_agent({
+  agent_type: "test_context_search_agent",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/test-context-search-agent.md (MUST read first)
-2. Run: `ccw spec load --category planning`
-3. Run: `ccw spec load --category test` (test framework, coverage targets, conventions)
+1. Run: `ccw spec load --category planning`
+2. Run: `ccw spec load --category test` (test framework, coverage targets, conventions)
 
 ---
 
@@ -97,13 +97,13 @@ close_agent({ id: contextAgentId });
 
 // Prompt Mode - gather from codebase via context-search-agent
 const contextAgentId = spawn_agent({
+  agent_type: "context_search_agent",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/context-search-agent.md (MUST read first)
-2. Run: `ccw spec load --category planning`
-3. Run: `ccw spec load --category test` (test framework, coverage targets, conventions)
+1. Run: `ccw spec load --category planning`
+2. Run: `ccw spec load --category test` (test framework, coverage targets, conventions)
 
 ---
 
@@ -171,13 +171,13 @@ close_agent({ id: contextAgentId });
 
 ```javascript
 const analysisAgentId = spawn_agent({
+  agent_type: "cli_execution_agent",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-execution-agent.md (MUST read first)
-2. Run: `ccw spec load --category planning`
-3. Run: `ccw spec load --category test` (test framework, coverage targets, conventions)
+1. Run: `ccw spec load --category planning`
+2. Run: `ccw spec load --category test` (test framework, coverage targets, conventions)
 
 ---
 
@@ -240,13 +240,13 @@ close_agent({ id: analysisAgentId });
 
 ```javascript
 const taskGenAgentId = spawn_agent({
+  agent_type: "action_planning_agent",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/action-planning-agent.md (MUST read first)
-2. Run: `ccw spec load --category planning`
-3. Run: `ccw spec load --category test` (test framework, coverage targets, conventions)
+1. Run: `ccw spec load --category planning`
+2. Run: `ccw spec load --category test` (test framework, coverage targets, conventions)
 
 ---
 
diff --git a/.codex/skills/workflow-test-fix-cycle/phases/02-test-cycle-execute.md b/.codex/skills/workflow-test-fix-cycle/phases/02-test-cycle-execute.md
index de74ef87..fd748740 100644
--- a/.codex/skills/workflow-test-fix-cycle/phases/02-test-cycle-execute.md
+++ b/.codex/skills/workflow-test-fix-cycle/phases/02-test-cycle-execute.md
@@ -85,13 +85,13 @@ return "conservative";
 ```javascript
 // Spawn agent for failure analysis
 const analysisAgentId = spawn_agent({
+  agent_type: "cli_planning_agent",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/cli-planning-agent.md (MUST read first)
-2. Run: `ccw spec load --category planning`
-3. Run: `ccw spec load --category test` (test framework, coverage targets, conventions)
+1. Run: `ccw spec load --category planning`
+2. Run: `ccw spec load --category test` (test framework, coverage targets, conventions)
 
 ---
 
@@ -152,13 +152,13 @@ close_agent({ id: analysisAgentId });
 ```javascript
 // Spawn agent for test execution/fixing
 const fixAgentId = spawn_agent({
+  agent_type: "test_fix_agent",
   message: `
 ## TASK ASSIGNMENT
 
 ### MANDATORY FIRST STEPS (Agent Execute)
-1. **Read role definition**: ~/.codex/agents/test-fix-agent.md (MUST read first)
-2. Run: `ccw spec load --category execution`
-3. Run: `ccw spec load --category test` (test framework, coverage targets, conventions)
+1. Run: `ccw spec load --category execution`
+2. Run: `ccw spec load --category test` (test framework, coverage targets, conventions)
 
 ---