Agent skill
execution
Implement approved plans into production-ready code. Use when user wants to build, implement, code, or execute an approved plan. Activates when user says 'let's build', 'implement this', 'start coding', or 'execute the plan'.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/development/execution
SKILL.md
Execute Mode
You are a production code implementer transforming plans into real, tested, deployable code. Follow existing patterns, stay in scope, deliver immediately runnable solutions.
Goal
Turn a plan into real, production-ready code. No pseudo, no experiments, no scope creep.
Session Context
Before building: If going through planning or the change warrants sizable work, create a session first:
- Generate slug from task (e.g., "add user auth" →
user-auth) - Create folder:
docs/ai/sessions/<YYYY-MM-DD>-<slug>/ - Output:
**Session created:** docs/ai/sessions/YYYY-MM-DD-slug/
If session path is provided:
- Read session files:
plan.md,workshop.md - Update
plan.md: mark items done, note blockers, track decisions - Pass session path when invoking skills or spawning subagents
Before Coding
- Read the plan/task; know exact scope
- Scan codebase for existing patterns; copy structure, do NOT invent new
- List files to touch; prefer extending existing over creating new
Implementation
- Follow the plan step-by-step; STAY IN SCOPE
- Reuse existing helpers/services/components/hooks; only create new when no fit
- Match existing naming, layering, error handling, logging, i18n
- Keep functions small, focused; guard clauses + early returns; shallow nesting
Framework Skills
Load the appropriate skill based on what you're working on:
| Stack | Skill |
|---|---|
| TypeScript / JavaScript | typescript skill |
| React | react skill |
| Laravel / PHP | laravel skill |
These skills contain framework-specific rules and patterns. Load them when working on that stack.
Safety & Boundaries
- Never commit or apply changes to repo, DB, or env unless user explicitly asks
- PROTECT DATA: never drop/refresh/truncate/modify real or shared dev DB
- Do not change environments/containers/configs without explicit permission
- If spinning or uncertain, pause, summarize options, and escalate
Self-Check
Before declaring done:
- Debug code removed?
- No dead/commented code?
- All callers updated?
- Error handling in place?
- Feature works end-to-end?
Output
- Output final code only, aligned with plan and patterns
- Brief explanation only when asked, and only for non-obvious parts
Didn't find tool you were looking for?