# Implementing

> Use when an approved implementation plan exists and you are about to start implementation

- Skill: `majiayu000/implementing` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add majiayu000/implementing`
- Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/implementing/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: majiayu000 (https://skillmd.com/u/majiayu000)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/majiayu000/implementing

---


# Implementing

## Overview

Execute an existing plan exactly as written, with verification and checkpoints.
If the plan is wrong or unclear, stop and ask to change it.

## When to Use

- A plan document exists and is approved
- The work needs disciplined, step-by-step execution

Skip if: there is no plan (write one first) or the task is trivial.

## Quick Reference

1. Read the full plan.
2. Call out blockers or questions.
3. Track progress per task.
4. Verify each task before moving on.
5. Check in after a small batch.

## Execution Workflow

1. **Load the plan**
   - Read end to end.
   - Identify unclear steps or missing info.

2. **Confirm before starting**
   - If anything is unclear, ask and wait.
   - Do not "fix" the plan silently.

3. **Execute tasks in order**
   - One task at a time.
   - Follow steps exactly.
   - Run the listed verification.
   - Mark the task complete.

4. **Checkpoint**
   - After 2-3 tasks (or any risky task), summarize:
     - What changed
     - Verification results
     - Open questions
   - Wait for approval before continuing.

5. **Finish**
   - Run any final verification from the plan.
   - Report completion and remaining risks.

## Rules

- Do not add scope or extra features
- Do not reorder tasks unless the plan says so
- Stop and ask if blocked or if verification fails

## Common Mistakes

- Skipping verification to "save time"
- Changing the plan without approval
- Executing multiple tasks in parallel without permission