From 914139abb1198dec43574d6096d2d02311a49698 Mon Sep 17 00:00:00 2001
From: Baiheng Xie <874256269@qq.com>
Date: Wed, 4 Feb 2026 15:48:30 +0800
Subject: [PATCH 1/2] feat: add guide for test-driven development workflow with
 agent orchestration

---
 .claude/commands/tdd.md | 82 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 82 insertions(+)
 create mode 100644 .claude/commands/tdd.md

diff --git a/.claude/commands/tdd.md b/.claude/commands/tdd.md
new file mode 100644
index 0000000..d0dcbd3
--- /dev/null
+++ b/.claude/commands/tdd.md
@@ -0,0 +1,82 @@
+Guide test-driven development through RED → GREEN → REFACTOR cycle with agent orchestration.
+
+Reference `pytest-patterns` skill for testing conventions.
+
+Task: $ARGUMENTS
+
+## Workflow
+
+### Phase 1: RED (Write Failing Test)
+
+1. **Understand requirement** from task description
+2. **Find test location**:
+   - Search existing tests in `tests/` directory
+   - Match structure: `src/foo/bar.py` → `tests/foo/test_bar.py`
+3. **Write test that fails**:
+   - Test behavior, not implementation
+   - Use descriptive test names
+   - Follow `pytest-patterns` conventions
+4. **Run test to verify failure**:
+   ```bash
+   uv run pytest path/to/test_file.py::test_name -v
+   ```
+
+### Phase 2: GREEN (Minimal Implementation)
+
+**For complex implementations** (multi-file changes, architectural decisions, unclear scope):
+- Use the `planner` agent to create implementation plan
+- Follow the plan to implement minimal code
+
+**For simple implementations** (single function, obvious change):
+- Implement minimal code directly
+
+**Verify test passes**:
+```bash
+uv run pytest path/to/test_file.py::test_name -v
+```
+
+### Phase 3: REFACTOR (Improve Code)
+
+1. **Use `refactor-cleaner` agent** to identify:
+   - Dead code to remove
+   - Duplicated logic to consolidate
+   - Cleanup opportunities
+
+2. **Apply refactorings** while keeping tests green:
+   ```bash
+   uv run pytest path/to/test_file.py -v
+   ```
+
+3. **Verify coverage**:
+   ```bash
+   uv run pytest path/to/test_file.py --cov=src/module/being/tested --cov-report=term-missing
+   ```
+
+### Phase 4: REVIEW (Quality Check)
+
+**Use `code-reviewer` agent** to verify:
+- Security concerns
+- Code quality
+- Test coverage adequacy
+- Adherence to project patterns
+
+## Agent Usage
+
+- **planner**: Complex implementations in GREEN phase
+- **refactor-cleaner**: Cleanup in REFACTOR phase
+- **code-reviewer**: Final quality check in REVIEW phase
+
+## Guidelines
+
+- One behavior per test
+- Self-contained test setup
+- Clear assertion messages with context
+- Tests < 1 second (mark slow tests with `@pytest.mark.integration`)
+- Create test/implementation files following project structure if they don't exist
+
+## Output
+
+- Show test run results
+- Report coverage percentage
+- Indicate active phase (RED/GREEN/REFACTOR/REVIEW)
+- Show agent outputs when used

From 1c897a17fddb5ab5775c8ab983d627b62e2cf7f8 Mon Sep 17 00:00:00 2001
From: Baiheng Xie <874256269@qq.com>
Date: Wed, 4 Feb 2026 15:53:48 +0800
Subject: [PATCH 2/2] refactor: split verbose command

---
 .claude/commands/tdd.md              | 81 +---------------------------
 .claude/skills/tdd-workflow/SKILL.md | 68 +++++++++++++++++++++++
 2 files changed, 70 insertions(+), 79 deletions(-)
 create mode 100644 .claude/skills/tdd-workflow/SKILL.md

diff --git a/.claude/commands/tdd.md b/.claude/commands/tdd.md
index d0dcbd3..1d96907 100644
--- a/.claude/commands/tdd.md
+++ b/.claude/commands/tdd.md
@@ -1,82 +1,5 @@
-Guide test-driven development through RED → GREEN → REFACTOR cycle with agent orchestration.
+Guide test-driven development through RED → GREEN → REFACTOR → REVIEW cycle.
 
-Reference `pytest-patterns` skill for testing conventions.
+Reference `pytest-patterns` and `tdd-workflow` skills. Use `planner` agent for complex implementations, `refactor-cleaner` agent for cleanup, `code-reviewer` agent for final review.
 
 Task: $ARGUMENTS
-
-## Workflow
-
-### Phase 1: RED (Write Failing Test)
-
-1. **Understand requirement** from task description
-2. **Find test location**:
-   - Search existing tests in `tests/` directory
-   - Match structure: `src/foo/bar.py` → `tests/foo/test_bar.py`
-3. **Write test that fails**:
-   - Test behavior, not implementation
-   - Use descriptive test names
-   - Follow `pytest-patterns` conventions
-4. **Run test to verify failure**:
-   ```bash
-   uv run pytest path/to/test_file.py::test_name -v
-   ```
-
-### Phase 2: GREEN (Minimal Implementation)
-
-**For complex implementations** (multi-file changes, architectural decisions, unclear scope):
-- Use the `planner` agent to create implementation plan
-- Follow the plan to implement minimal code
-
-**For simple implementations** (single function, obvious change):
-- Implement minimal code directly
-
-**Verify test passes**:
-```bash
-uv run pytest path/to/test_file.py::test_name -v
-```
-
-### Phase 3: REFACTOR (Improve Code)
-
-1. **Use `refactor-cleaner` agent** to identify:
-   - Dead code to remove
-   - Duplicated logic to consolidate
-   - Cleanup opportunities
-
-2. **Apply refactorings** while keeping tests green:
-   ```bash
-   uv run pytest path/to/test_file.py -v
-   ```
-
-3. **Verify coverage**:
-   ```bash
-   uv run pytest path/to/test_file.py --cov=src/module/being/tested --cov-report=term-missing
-   ```
-
-### Phase 4: REVIEW (Quality Check)
-
-**Use `code-reviewer` agent** to verify:
-- Security concerns
-- Code quality
-- Test coverage adequacy
-- Adherence to project patterns
-
-## Agent Usage
-
-- **planner**: Complex implementations in GREEN phase
-- **refactor-cleaner**: Cleanup in REFACTOR phase
-- **code-reviewer**: Final quality check in REVIEW phase
-
-## Guidelines
-
-- One behavior per test
-- Self-contained test setup
-- Clear assertion messages with context
-- Tests < 1 second (mark slow tests with `@pytest.mark.integration`)
-- Create test/implementation files following project structure if they don't exist
-
-## Output
-
-- Show test run results
-- Report coverage percentage
-- Indicate active phase (RED/GREEN/REFACTOR/REVIEW)
-- Show agent outputs when used
diff --git a/.claude/skills/tdd-workflow/SKILL.md b/.claude/skills/tdd-workflow/SKILL.md
new file mode 100644
index 0000000..995f365
--- /dev/null
+++ b/.claude/skills/tdd-workflow/SKILL.md
@@ -0,0 +1,68 @@
+---
+name: tdd-workflow
+description: TDD cycle phases (RED → GREEN → REFACTOR → REVIEW) and agent orchestration
+---
+
+# TDD Workflow
+
+## Phases
+
+### Phase 1: RED (Write Failing Test)
+
+1. **Understand requirement** from task description
+2. **Find test location**:
+   - Search existing tests in `tests/` directory
+   - Match structure: `src/server/auth.py` → `tests/server/test_auth.py`
+3. **Write test that fails**:
+   - Test behavior, not implementation
+   - Use descriptive test names
+   - Follow `pytest-patterns` conventions
+4. **Run test to verify failure**:
+   ```bash
+   uv run pytest path/to/test_file.py::test_name -v
+   ```
+
+### Phase 2: GREEN (Minimal Implementation)
+
+**For complex implementations** (multi-file changes, architectural decisions, unclear scope):
+- Use the `planner` agent to create implementation plan
+- Follow the plan to implement minimal code
+
+**For simple implementations** (single function, obvious change):
+- Implement minimal code directly
+
+**Verify test passes**:
+```bash
+uv run pytest path/to/test_file.py::test_name -v
+```
+
+### Phase 3: REFACTOR (Improve Code)
+
+1. **Use `refactor-cleaner` agent** to identify:
+   - Dead code to remove
+   - Duplicated logic to consolidate
+   - Cleanup opportunities
+
+2. **Apply refactorings** while keeping tests green:
+   ```bash
+   uv run pytest path/to/test_file.py -v
+   ```
+
+3. **Verify coverage**:
+   ```bash
+   uv run pytest path/to/test_file.py --cov=src/module/being/tested --cov-report=term-missing
+   ```
+
+### Phase 4: REVIEW (Quality Check)
+
+**Use `code-reviewer` agent** to verify:
+- Security concerns
+- Code quality
+- Test coverage adequacy
+- Adherence to project patterns
+
+## Agent Orchestration
+
+- **planner**: Complex implementations in GREEN phase
+- **refactor-cleaner**: Cleanup in REFACTOR phase
+- **code-reviewer**: Final quality check in REVIEW phase