راهکارهای حرفه‌ای توسعه نرم‌افزار با هوش مصنوعی نسخه ۲

عنوان دوره : Master.dev – Professional AI Dev Setup, v2

سطح : تمامی سطوح

زمان کل دوره : 5 ساعت و 38 دقیقه

تعداد فايل ها : 1080p: 5

سال آموزش : 2026

فایل تمرینی : ندارد

مربی دوره : Steve Kinney

نرم افزار : –

زیرنویس انگلیسی : ندارد

Master.dev – Professional AI Dev Setup, v2

Move beyond ad hoc prompting and learn how to design AI coding workflows that are reusable, reviewable, and safe. Build project instructions, skills, hooks, subagents, and automation recipes you can carry across tools like Claude Code, Codex, and Cursor.

Skills, Hooks, Subagents, and Automation Recipes for Real AI Coding Workflows

A professional and productive AI coding environment mean encoding the things you keep re-explaining into a repeatable system. This hands-on, lab-driven workshop teaches the shared primitives that Claude Code, Codex, and Cursor have all converged on: project instructions, skills, hooks, subagents, MCP, and sandbox/approval policy.

Each section introduces one primitive and immediately turns it into a working recipe built in your own editor, culminating in a capstone that chains everything into one end-to-end feature workflow. You will leave with a transferable craft for professional AI-assisted development, not a vendor-specific bag of tricks.

What You Will Master
The Senior Engineer Toolkit

Separate the seven core primitives—instructions, scoped rules, skills, hooks, subagents, MCP, and sandbox/approval policy—and know what belongs where.
Design an instruction architecture across CLAUDE.md/AGENTS.md, project rules, user rules, and team config without creating an unreadable AI bible.
Write strong skills as reusable runbooks with clear context, steps, and expected output.
Use progressive disclosure and context budgeting so agents load detailed guidance only when it is relevant.
Install hooks as deterministic guardrails for formatting, protected files, validation, and session setup so safety does not depend on the model remembering.
Use subagents for context isolation and parallel specialized work that returns concise summaries instead of raw logs.
Connect MCP tools for external context while reasoning clearly about credentials, blast radius, sandbox boundaries, and approval policy.
Run a plan → implement → verify → review → capture loop with humans accountable for final judgment.
Standardize the whole system into something a teammate can read, audit, and adopt.

Prerequisites

Professional JavaScript or TypeScript experience.
Basic comfort with the terminal and Git, including branches, diffs, and commits.
Some hands-on time with at least one AI coding assistant such as Claude Code, Codex, Cursor, Copilot, or a similar tool.
A laptop you can install software on, with Node.js LTS, npm, and Git available on your PATH.
One AI coding tool installed and signed in, plus a terminal and editor you are comfortable using.
Optionally, an MCP server you would like to try during the workshop.

Who is this for?

This workshop is for developers who already use an AI coding tool and want to move past ad hoc prompting into deliberate, repeatable workflows. It is especially well suited for senior engineers and tech leads who want their team’s use of AI to be reviewable, auditable, and teachable—not tribal knowledge. Prior experience with skills, hooks, subagents, MCP, sandboxing, or approval systems is not required.