Use when ready to implement designed features - breaks design into TDD tasks (Red-Green-Refactor), tracks progress with checkboxes in tasks.md, enforces strict testing discipline. Activates when user says "implement this", "let's code", "start execution", mentions "tasks", "TDD", or uses /dev-workflow:spec commands (tasks, execute).
Breaks approved designs into TDD tasks with Red-Green-Refactor cycles, tracking progress in tasks.md. Executes tasks systematically when you say "implement this" or "let's code", using Edit to update checkboxes and enforcing strict test-first development.
/plugin marketplace add xbklairith/kisune/plugin install dev-workflow@kisuneThis skill is limited to using the following tools:
Guide feature implementation through two structured phases: Task Breakdown (TDD) → Execution. This systematic approach ensures test-driven development, quality gates, and tracked progress from design to working code.
Activate this skill when:
/dev-workflow:spec command with implementation options (tasks, execute)This skill requires completed planning from spec-driven-planning skill:
docx/features/[NN-feature-name]/requirements.md is complete with EARS requirementsdesign.md is complete and approvedIf prerequisites are missing:
"Implementation requires completed planning. Run
/dev-workflow:specand complete options 1-3 first (Feature Creation, Requirements, Design)."
Goal: Break design into small, testable tasks following Red-Green-Refactor
Task Structure:
Each task follows TDD cycle:
[ ] Task N: [Description]
[ ] RED: Write failing test for [functionality]
[ ] GREEN: Implement minimal code to pass test
[ ] REFACTOR: Clean up and optimize
Acceptance Criteria:
[ ] [Specific criterion 1]
[ ] [Specific criterion 2]
Linked Requirements: REQ-###, REQ-###
Traceability Rules:
Task Sizing Guidelines:
UltraThink Before Task Breakdown: Before breaking design into tasks, activate deep thinking if:
🗣 Say: "Let me ultrathink the implementation strategy before breaking this into tasks."
Questions to ultrathink:
After UltraThink: Create focused, testable tasks that validate assumptions early.
Task Categories:
Component Tasks
Integration Tasks
Error Handling Tasks
Documentation Tasks
Final Verification Tasks
Output:
Update docx/features/[NN-feature-name]/tasks.md with:
User Confirmation:
"Tasks defined with TDD cycle. Ready to begin implementation?"
Goal: Execute tasks systematically with quality gates
Execution Workflow:
For each task:
Mark Task as In Progress
Use Edit tool on docx/features/[NN-feature-name]/tasks.md:
[ ] Task N: [description][→] Task N: [description][→]Example Edit call:
Edit tool:
file: docx/features/01-user-auth/tasks.md
old_string: "[ ] Task 3: Implement JWT validation"
new_string: "[→] Task 3: Implement JWT validation"
RED Phase
[ ] RED: ... → [x] RED: ...test: Add test for [functionality]GREEN Phase
[ ] GREEN: ... → [x] GREEN: ...feat: Implement [functionality]REFACTOR Phase
[ ] REFACTOR: ... → [x] REFACTOR: ...refactor: Optimize [component]Mark Task Complete
Use Edit tool on docx/features/[NN-feature-name]/tasks.md:
[→] Task N: ... → [x] Task N: ...[x]CRITICAL: Use Edit tool to update tasks.md - don't just announce progress.
Start Phase 5
↓
Edit: Status "Not Started" → "In Progress"
↓
For each task:
↓
Edit: [ ] Task N → [→] Task N
Edit: [ ] RED → [→] RED
↓
Write failing test
↓
Edit: [→] RED → [x] RED
Edit: [ ] GREEN → [→] GREEN
↓
Implement code
↓
Edit: [→] GREEN → [x] GREEN
Edit: [ ] REFACTOR → [→] REFACTOR
↓
Refactor code
↓
Edit: [→] REFACTOR → [x] REFACTOR
Edit: [→] Task N → [x] Task N
Edit: Update Progress Summary
↓
Next task or finish
↓
Edit: Status "In Progress" → "Complete"
Always follow this pattern for every task:
Before starting any task:
Use Edit tool to change [ ] → [→] in tasks.md
After completing each phase (RED/GREEN/REFACTOR):
Use Edit tool to change [→] → [x] for that phase
Use Edit tool to change [ ] → [→] for next phase
After completing full task:
Use Edit tool to change [→] → [x] for task header
Use Edit tool to update Progress Summary counts
Don't skip these Edit calls - they're not optional suggestions, they're required operations.
Wrong (Just Announcing):
✅ Task 1 complete
Moving to Task 2...
Right (Actually Modifying + Announcing):
[Using Edit tool to mark Task 1 complete: [→] → [x]]
[Using Edit tool to update Progress Summary: Completed 1/10]
✅ Task 1 complete
[Using Edit tool to mark Task 2 in progress: [ ] → [→]]
Starting Task 2: JWT token generation...
The Progress Summary section must stay synchronized:
Before any tasks start:
- Total Tasks: 10
- Completed: 0/10
- In Progress: None
After Task 1 starts:
- Total Tasks: 10
- Completed: 0/10
- In Progress: Task 1 - User model with password hashing
After Task 1 completes:
- Total Tasks: 10
- Completed: 1/10
- In Progress: Task 2 - JWT token generation
Use Edit tool to update these fields after every task completion.
Status Checkpoints:
Every 2-3 completed tasks, provide status update:
🔄 Checkpoint Update:
- ✅ Tests: [N/N] passing
- ✅ Type check: No errors
- ✅ Lint: Clean
- 📝 Completed tasks: [X/Y]
- 🎯 Next: [Next task description]
[Confidence: X.X]
Edit patterns already detailed above. Key reminders:
grep -c "^\- \[x\] Task" tasks.mdnpm run test:coverage and Edit tasks.mdAuto-Trigger Code Quality Review:
Before each commit:
dev-workflow:code-qualityHow to activate:
Use Skill tool: Skill(skill: "dev-workflow:code-quality")
Integration Points:
dev-workflow:test-driven-development for strict TDD guidancedev-workflow:git-workflow for smart commitsdev-workflow:code-quality for pre-commit reviewsHow to activate integration skills:
# For TDD enforcement
Use Skill tool: Skill(skill: "dev-workflow:test-driven-development")
# For git operations
Use Skill tool: Skill(skill: "dev-workflow:git-workflow")
# For code review
Use Skill tool: Skill(skill: "dev-workflow:code-quality")
Completion Criteria:
Feature is complete when:
Final Output:
✅ Feature Complete: [Feature Name]
Summary:
- Total tasks: [N]
- Tests added: [N]
- Test coverage: [X%]
- Commits: [N]
- Files changed: [N]
Ready to create PR or merge to main?
RED (Write Failing Test):
Write a test that describes what the code should do.
The test MUST fail because the code doesn't exist yet.
This proves the test is actually testing something.
GREEN (Make Test Pass):
Write the MINIMUM code needed to make the test pass.
Don't worry about elegance or optimization.
Just make it work.
REFACTOR (Improve Code):
Now that tests pass, improve the code:
- Remove duplication
- Improve naming
- Simplify logic
- Optimize performance
Run tests after each change to ensure they still pass.
DO:
DON'T:
User: "/dev-workflow:spec tasks"
Assistant:
Activating spec-driven-implementation skill at Phase 4: Task Breakdown
Found feature: docx/features/01-user-authentication/
✅ requirements.md exists
✅ design.md exists
Reading design to create TDD tasks...
Breaking down into testable tasks:
[ ] Task 1: User model with password hashing
[ ] RED: Test password hashing on save
[ ] GREEN: Implement bcrypt hashing
[ ] REFACTOR: Extract hashing to utility
Acceptance Criteria:
[ ] Passwords are hashed before storage
[ ] Original password is not stored
[ ] Hash is verifiable
[ ] Task 2: JWT token generation
[ ] RED: Test token generation with user ID
[ ] GREEN: Implement JWT signing
[ ] REFACTOR: Extract to auth service
Acceptance Criteria:
[ ] Token contains user ID
[ ] Token expires in 24 hours
[ ] Token is properly signed
[...8 more tasks...]
Tasks defined. Ready to begin implementation?
User: "Yes, let's start"
Assistant:
Activating Phase 5: Execution
[Using Edit tool on docx/features/01-user-authentication/tasks.md]
[Changing "**Status:** Not Started" to "**Status:** In Progress"]
[Using Edit tool to mark Task 1 in progress]
[Changing "[ ] Task 1: User model with password hashing" to "[→] Task 1: User model with password hashing"]
[Using Edit tool to mark RED phase in progress]
[Changing " [ ] RED: Test password hashing on save" to " [→] RED: Test password hashing on save"]
Starting Task 1: User model with password hashing
RED Phase: Writing failing test...
[Creates test file tests/models/user.test.js with failing test]
Running tests... ❌ 1 failing (expected)
[Using Edit tool to check off RED phase]
[Changing " [→] RED: Test password hashing on save" to " [x] RED: Test password hashing on save"]
Commit: test: Add test for user password hashing
[Using Edit tool to mark GREEN phase in progress]
[Changing " [ ] GREEN: Implement bcrypt hashing" to " [→] GREEN: Implement bcrypt hashing"]
GREEN Phase: Implementing minimal code...
[Implements password hashing in src/models/user.js]
Running tests... ✅ All passing
[Using Edit tool to check off GREEN phase]
[Changing " [→] GREEN: Implement bcrypt hashing" to " [x] GREEN: Implement bcrypt hashing"]
Commit: feat: Implement user password hashing with bcrypt
[Using Edit tool to mark REFACTOR phase in progress]
[Changing " [ ] REFACTOR: Extract hashing to utility" to " [→] REFACTOR: Extract hashing to utility"]
REFACTOR Phase: Extracting to utility...
[Refactors code to src/utils/crypto.js]
Running tests... ✅ All passing
[Using Edit tool to check off REFACTOR phase]
[Changing " [→] REFACTOR: Extract hashing to utility" to " [x] REFACTOR: Extract hashing to utility"]
Commit: refactor: Extract password hashing to utility module
[Using Edit tool to mark Task 1 complete]
[Changing "[→] Task 1: User model with password hashing" to "[x] Task 1: User model with password hashing"]
[Using Edit tool to update Progress Summary]
[Changing "- Completed: 0/10" to "- Completed: 1/10"]
[Changing "- In Progress: Task 1 - User model with password hashing" to "- In Progress: Task 2 - JWT token generation"]
✅ Task 1 complete
🔄 Checkpoint Update:
- ✅ Tests: 3/3 passing
- ✅ Type check: No errors
- ✅ Lint: Clean
- 📝 Completed tasks: 1/10
- 🎯 Next: Task 2 - JWT token generation
[Confidence: 0.9]
Continue to Task 2?
If during implementation you discover:
STOP implementation and return to planning:
"Implementation blocked: [issue]. Need to revisit [requirements/design]. Use
/dev-workflow:spec [requirements/design]to update planning documents."
Then use Skill tool to invoke: dev-workflow:spec-driven-planning to update planning documents before continuing implementation.
How to return to planning:
Use Skill tool: Skill(skill: "dev-workflow:spec-driven-planning")
[ ] → [→] before starting task[→] → [x] after finishing taskdev-workflow:code-quality before commitsdev-workflow:git-workflow for smart commit messagesAnnouncing progress ≠ Updating files
Wrong:
✅ Task 1 complete
Moving to Task 2...
Right:
[Using Edit tool to mark Task 1 complete: [→] → [x]]
[Using Edit tool to update Progress Summary]
✅ Task 1 complete
This skill should be used when the user asks to "create an agent", "add an agent", "write a subagent", "agent frontmatter", "when to use description", "agent examples", "agent tools", "agent colors", "autonomous agent", or needs guidance on agent structure, system prompts, triggering conditions, or agent development best practices for Claude Code plugins.
This skill should be used when the user asks to "create a slash command", "add a command", "write a custom command", "define command arguments", "use command frontmatter", "organize commands", "create command with file references", "interactive command", "use AskUserQuestion in command", or needs guidance on slash command structure, YAML frontmatter fields, dynamic arguments, bash execution in commands, user interaction patterns, or command development best practices for Claude Code.
This skill should be used when the user asks to "create a hook", "add a PreToolUse/PostToolUse/Stop hook", "validate tool use", "implement prompt-based hooks", "use ${CLAUDE_PLUGIN_ROOT}", "set up event-driven automation", "block dangerous commands", or mentions hook events (PreToolUse, PostToolUse, Stop, SubagentStop, SessionStart, SessionEnd, UserPromptSubmit, PreCompact, Notification). Provides comprehensive guidance for creating and implementing Claude Code plugin hooks with focus on advanced prompt-based hooks API.