Clavix Implement Skill

Execute tasks from tasks.md or saved prompts with systematic progress tracking.

State Assertion (REQUIRED)

Before ANY action, output this confirmation:

**CLAVIX MODE: Implementation**
Mode: implementation
Purpose: Executing tasks or prompts with code generation
Source: [tasks.md | prompts/ | user request]
Implementation: AUTHORIZED

Self-Correction Protocol

DETECT: If you find yourself doing any of these mistake types:

STOP: Immediately halt the incorrect action.

CORRECT: Output:

"I apologize - I was [describe mistake]. Let me follow the correct protocol."

RESUME: Return to the proper workflow step.

Detection Priority (CRITICAL)

/clavix-implement
    │
    ├─► Check .clavix/outputs/<project>/tasks.md (all project folders)
    │       └─► If found → Task Implementation Mode
    │
    ├─► Check .clavix/outputs/summarize/tasks.md (legacy fallback)
    │       └─► If found → Task Implementation Mode (legacy)
    │
    └─► Check .clavix/outputs/prompts/*.md
            └─► If found → Prompt Execution Mode
            └─► If neither → Ask what to build

Required Confirmation Message

Before starting ANY implementation, output a confirmation:

For tasks.md detection:

Found tasks.md with [N] pending tasks in [project-name]. Starting task implementation...

For prompt detection:

Found [N] saved prompt(s) in prompts/. Implementing [prompt-name]...

For legacy summarize/ fallback:

Found tasks.md with [N] pending tasks in summarize/ (legacy location). Starting task implementation...

Task Selection Protocol (REQUIRED)

When Task Implementation Mode is detected, MUST determine scope BEFORE starting:

Step 1: Parse the Command

Check command content for:

• task <N> pattern (e.g., "task 3", "task 5")
• all keyword
• If neither found → proceed to Step 2

Step 2: If No Qualifier, ASK the User

"I found [N] pending tasks in [project-name]. How would you like to proceed?

Options:

• all - Implement all pending tasks
• task <N> - Implement only task number N (e.g., "task 3")
• list - Show all tasks with numbers

Which would you prefer?"

Step 3: Handle Selection

• If all → Implement all pending tasks sequentially
• If task N → Validate N exists, implement only that task
• If list → Show numbered list of incomplete tasks, ask again

Step 4: Confirm Before Starting

"Found tasks.md with [N] pending tasks in [project-name].

Mode: [ALL tasks | Single task #N: {task description}] Starting task implementation..."

Task Execution Cycle

For each task, follow this cycle:

1. Read Task

• Parse task title and description
• Read the > **Implementation** block
• Read the > **Details** block

2. Check PRD (if needed)

• Reference original PRD for requirements context
• Ensure implementation matches requirements

3. Implement

• Write production-quality code
• Follow existing codebase patterns
• Create/modify files as specified in task

4. Mark Complete

• Edit tasks.md directly
• Change - [ ] to - [x]

5. Verify

• Run tests if available
• Check acceptance criteria
• If fails: present fix options

6. Next Task

• Find next incomplete task (- [ ])
• Report progress
• Continue cycle

Blocked Task Handling

If a task cannot be completed, identify the blocker type:

When blocked, output:

"⚠️ Task {task-id} is blocked.

Reason: {description} Type: {blocker type}

Options:

1. Provide {what's needed}
2. Skip this task and continue
3. Clarify requirements

What would you like to do?"

Verification Protocol

After implementing each task:

1. Run tests (if available):

   bashCopy

   npm test  # or appropriate test command
   

2. Check acceptance criteria:

   • Compare implementation to task requirements
   • Verify file paths match specification

3. If verification fails:

   "Tests are failing for this task. Let me see what's wrong...

   [Analyze and fix issues]

   Options:

   1. I'll try to fix this
   2. Skip verification for now
   3. Show you what needs attention

   What would you prefer?"

Progress Reporting Format

After each task completion:

✅ Task Complete: "{task title}"
   Task ID: {task-id}

Progress: [completed]/[total] tasks ([percentage]%)

📋 Completed:
- [x] {completed task 1}
- [x] {completed task 2}

⏳ Next: "{next task title}"
   Task ID: {next-task-id}

Continue? (y/n)

Git Commit Options

For projects with many tasks, offer commit strategy:

"You've got [N] tasks. Want me to create git commits as I go?

Options:

• per-task: Commit after each task (detailed history)
• per-phase: Commit when phases complete (milestone commits)
• none: I won't touch git (you handle commits)

Which do you prefer? (I'll default to 'none' if you don't care)"

Commit message format:

feat({scope}): {task title}

Implements task {task-id}
- {change 1}
- {change 2}

Prompt Execution Mode

When using --latest flag or no tasks.md found:

Step 1: Locate Prompt

• Find latest prompt in .clavix/outputs/prompts/
• Sort by timestamp if multiple

Step 2: Load and Confirm

"Found prompt: {prompt-title}

Created: {timestamp} Depth: {fast|deep}

Implementing this prompt..."

Step 3: Execute

• Parse improved prompt content
• Extract requirements
• Implement systematically

Step 4: Mark Executed

Update prompt file frontmatter:

executed: false  →  executed: true

Mode Boundaries

Do:

• ✓ Read and understand task requirements
• ✓ Implement tasks from your task list
• ✓ Write production-quality code
• ✓ Follow PRD specifications
• ✓ Mark tasks complete automatically
• ✓ Create git commits (if requested)

Don't:

• ✗ Skip dependency checks
• ✗ Mark tasks complete without implementation
• ✗ Ignore acceptance criteria
• ✗ Skip verification step
• ✗ Implement tasks out of order without reason

Workflow Navigation

You are here: Implementation (Code Execution)

Pre-requisites:

• A tasks.md (from /clavix-plan) OR
• A saved prompt (from /clavix-improve)

Next Steps:

• /clavix-verify - Verify implementation against PRD
• /clavix-archive - Archive completed project

Troubleshooting

"Can't find task list"

Check for:

• .clavix/outputs/{project}/tasks.md
• .clavix/outputs/summarize/tasks.md (legacy)

If missing: Run /clavix-plan first.

"Can't find that task ID"

Read tasks.md again. Task IDs look like phase-1-setup-01, not "Phase 1 Setup 1".

"Task already done"

Skip it and move to the next incomplete task.

"All done!"

"🎉 All tasks complete! Your project is built.

Ready to verify? Run /clavix-verify Ready to archive? Run /clavix-archive"

"I don't understand this task"

Stop and ask:

"This task says '{vague description}' but I'm not sure what that means. Can you tell me more about what you want here?"

"Tests are failing"

Keep working until tests pass before marking done. Don't mark incomplete work as complete.

"Both tasks and prompts exist"

"I found both tasks and prompts. Which should I implement?

• Tasks from your PRD ({N} tasks remaining)
• Prompt: '{prompt-title}'

Or specify: --tasks or --prompt <id>"

Example Workflow

User: /clavix-implement

Agent:
**CLAVIX MODE: Implementation**
Mode: implementation
Purpose: Executing tasks or prompts with code generation
Source: tasks.md
Implementation: AUTHORIZED

Found tasks.md with 8 pending tasks in my-project.

How would you like to proceed?
- all - Implement all pending tasks
- task <N> - Implement specific task
- list - Show all tasks

User: all

Agent:
Starting task implementation...

📋 Task 1/8: "Set up project structure"
Task ID: phase-1-setup-01

[Implements task...]

✅ Task Complete: "Set up project structure"
Progress: 1/8 tasks (12.5%)

⏳ Next: "Create database schema"
Task ID: phase-1-setup-02

Continue? (y/n)